Paper deep dive
Talking to Digital Twins: Selective Disclosure and Belief Measurement in Financial Social Media
Boone Bowles, Raymond Duch, Sorin Sorescu
Intelligence
Status: not_run | Model: - | Prompt: - | Confidence: 0%
Entities (0)
Relation Signals (0)
No relation signals yet.
Cypher Suggestions (0)
No Cypher suggestions yet.
Abstract
Abstract:Social media affect financial markets, but public posts by financial media personas are voluntary disclosures. What is not disclosed is therefore usually unobserved. We address this measurement problem by conducting repeated, real-time interviews of "digital twins" built from monitored finfluencers' X accounts under a fixed protocol. The interviews recover stock-level public-persona belief proxies even when no public recommendation is made. Because the interviews are generated and archived before the relevant return windows, the design avoids the look-ahead bias that arises when LLMs are queried ex post. The evidence shows that information obtained from these digital-twin interviews predicts the cross section of large-cap stock returns in the expected direction. Repeated real-time interviews therefore show how selective disclosure can be turned into measurable panels of market views.
Tags
Links
- Source: https://arxiv.org/abs/2608.01181v1
- Canonical: https://arxiv.org/abs/2608.01181v1
Trouble viewing inline? Open PDF directly â
Full Text
217,997 characters extracted from source content.
Expand or collapse full text
ABSTRACT Social media affect financial markets, but public posts by financial media personas are voluntary disclosures. What is not disclosed is therefore usually unobserved. We address this measurement problem by conducting repeated, real-time interviews of âdigital twinsâ built from monitored finfluencersâ X accounts under a fixed protocol. The interviews recover stock-level public-persona belief proxies even when no public recommendation is made. Because the interviews are generated and archived before the relevant return windows, the design avoids the look-ahead bias that arises when LLMs are queried ex post. The evidence shows that information obtained from these digital-twin interviews predicts the cross section of large-cap stock returns in the expected direction. Repeated real-time interviews therefore show how selective disclosure can be turned into measurable panels of market views. Keywords: Finfluencers, social media, digital twins, belief aggregation, return predictability JEL Classification Numbers: G12, G14, G41 Abstract Talking to Digital Twins: Selective Disclosure and Belief Measurement in Financial Social Media Social media affect financial markets, but public posts by financial media personas are voluntary disclosures. What is not disclosed is therefore usually unobserved. We address this measurement problem by conducting repeated, real-time interviews of âdigital twinsâ built from monitored finfluencersâ X accounts under a fixed protocol. The interviews recover stock-level public-persona belief proxies even when no public recommendation is made. Because the interviews are generated and archived before the relevant return windows, the design avoids the look-ahead bias that arises when LLMs are queried ex post. The evidence shows that information obtained from these digital-twin interviews predicts the cross section of large-cap stock returns in the expected direction. Repeated real-time interviews therefore show how selective disclosure can be turned into measurable panels of market views. Keywords: Finfluencers, social media, digital twins, belief aggregation, return predictability JEL Classification Numbers: G12, G14, G41 1. Introduction A central premise of financial economics is that markets aggregate dispersed information through prices. In Hayekâs (1945) formulation, prices act as a mechanism for communicating and coordinating knowledge that is distributed across agents. Grossman and Stiglitz (1980) point out that when information is costly, fully revealing prices would eliminate the incentive to acquire it, implying that some degree of inefficiency must persist in equilibrium. More recent research emphasizes that beyond cost, limited attention also causes imperfect adjustment of prices to new information, leading to predictable return dynamics (Hong and Stein, 1999; Sims, 2003; Reis, 2006). The rise of digital communication has further transformed the link between information and prices by enabling decentralized diffusion of price-relevant content. Investors encounter market narratives, stock recommendations, and trading rules of thumb through creators (âfinfluencersâ) on popular platforms such as TikTok and X.111Recent surveys and regulatory analyses document both the scale of finfluencer activity and associated concerns about advice quality, disclosures, and investor protection (Espeute and Preece, 2024; Financial Industry Regulatory Authority, 2025; Ontario Securities Commission, 2025). In modern digital environments, these activities interact with amplification via algorithms and influencer popularity, further enhancing transmission of opinion. An expanding body of work demonstrates that such content can also move prices, predict near-term returns, and affect market efficiency (Antweiler and Frank, 2004; Adams et al., 2023; NĂșñez-Mora and Mendoza-Urdiales, 2023; Kakhbod et al., 2023; Gordillo MartĂnez, 2024). Although the literature indicates that choices of social media creators are economically meaningful, these posts create a measurement problem. Specifically, silence is ambiguous. A creator may be silent because she has no view, because the stock is not salient, because the view is uncertain, because the post would not attract attention, or because disclosure would be strategically costly. Thus, a post-based measure mixes two objects: the underlying view and the decision to reveal it. The absence of a recommendation may be as important as the recommendation itself, yet silence is normally unobserved as a belief state. In this paper, we address the challenge of selective disclosure by introducing a novel measurement approach based on structured elicitation of beliefs. Rather than passively observing what finfluencers choose to reveal in public posts, we recover their views using repeated, standardized interviews applied to AI-based âdigital twinsâ constructed from their observable social-media personas. Advances in large language models enable us to create virtual representations of real personas that can be queried systematically over time.222LLMs can now condition responses on rich text supplied at the time of the query. Several developments make this possible. First, transformer models use attention mechanisms to relate different parts of a text prompt to one another, allowing the model to connect an account biography, recent posts, a ticker, and a question in the same response (Vaswani et al., 2017). Second, large-scale pretraining and in-context learning allow a pretrained model to perform new tasks from instructions and examples in the prompt, without training a separate model for each account (Brown et al., 2020). Third, instruction tuning and human-feedback training make the model better at following detailed prompts and returning structured answers, which matters because our protocol asks for scores, confidence, speculation, horizons, and catalysts (Ouyang et al., 2022). Fourth, retrieval and context augmentation show how model outputs can account for external information supplied at query time, rather than relying only on knowledge stored in model parameters (Lewis et al., 2020). Finally, recent social-science work shows that LLM agents can produce meaningful persona-conditioned responses when given rich background information, including interviews, memories, or assigned personas (Horton et al., 2023; Park et al., 2024). By implementing a fixed daily interview protocol, we generate a panel dataset of forward-looking beliefs that is time-stamped, comparable across influencers, and not restricted to voluntary disclosure. Specifically, we implement a method to ask finfluencers questions instead of just waiting for them to post. Each day, we run repeated, real-time interviews of âdigital twinsâ of finfluencers on X.333The real-time aspect of this study is similar to epidemiological research in medicine such as the Framingham Heart Study (Mahmood et al., 2014; Andersson et al., 2019), where a cohort of subjects is followed forward to ascertain outcomes, with no look-ahead bias. A digital twin in this setting is an artificial interviewee built from account information and monitored recent content and instructed to respond from the perspective of the associated public persona. The interview is daily and standardized; public posts are voluntary and non-uniform. The design, therefore, separates the recovery of views from the creatorâs decision to disclose them publicly. We use the term digital twin operationally. The twin does not reveal what the human finfluencer privately believes or actually owns. It is a standardized representation of the finfluencerâs public persona, built from account-specific public information and interviewed under the same protocol as the other twins. We ask each digital twin to share views about broad stock market conditions and to make specific buy, hold, or sell recommendations about each large-cap stock in our sample. The same monitored personas are queried repeatedly. The questions are fixed, and so is the cross-section of stocks. The answers are generated contemporaneously, timestamped, and aligned to market event time before the relevant stock return outcomes are observed. The timing of these interviews is central to our research design because pre-trained LLMs create a special challenge for financial predictability tests. If a researcher asks a model today to reconstruct what a market participant would have believed last year, the model may have been trained on, or have access to, information that was not available at the time. That is look-ahead bias in a new form (Sul et al., 2017; Ludwig et al., 2024; Sarkar and Vafa, 2024). Our interviews are not run ex post. They are generated live, timestamped, archived, and then compared with subsequent returns. The tests therefore evaluate interview records that existed before the return windows being studied. The interview protocol has two immediate advantages. The first is a coverage gain. Because we query monitored personas directly and repeatedly, rather than relying only on voluntary public posts, the interviews recover stock-level views for many names that never receive an actual public recommendation. They also create a daily path of views around names that eventually become public recommendations, including days before and after the disclosure date. The second advantage is a measurement gain. The stock-pick interviews not only ask whether a digital twin is buy-leaning or sell-leaning on a given stock. They also ask how confident the response is, whether the recommendation is speculative, and what catalyst supports the view. This allows us to separate three distinct states of the belief panel: direction, disagreement, and uncertainty. Direction tells us which stocks the panel favors. Disagreement tells us which stocks divide the panel. Uncertainty tells us where the panel lacks grounded conviction. These distinctions are difficult to recover from public posts alone, especially when no post is made. We first validate whether the digital twin produces responses consistent with its human finfluencer counterpart. We examine overlaps (same day, same ticker, same account) between recommendations made by human finfluencers and those made by their digital twin counterparts. Digital twinsâ recommendations match those of their human counterparts 91.5%91.5\% of the time. Moreover, before later public recommendations appear, digital twinsâ interviews already lean toward the direction that eventually becomes public. And, after removing common date-question effects from the macro interviews, the digital twinâs own identity still explains a substantial portion of the variation in the outcome of these macro interviews. These validations show that the information produced by digital-twin interviews is not merely generic market commentary. We then use the results from these digital-twin interviews to predict cross-sectional stock-level returns over the next ten trading days. For each stock-event, we summarize the interview output with Net Buy Share, a simple buy-minus-sell measure: the share of digital-twin recommendations that are buys minus the share that are sells. We find that stocks with more buy-leaning interview recommendations outperform their benchmark, while stocks with more sell-leaning recommendations underperform, suggesting that recommendations elicited from finfluencersâ digital twins contain relevant stock-selection information. Importantly, this predictability is strongest in the silent region, i.e., for stocks that are not explicitly mentioned by the human finfluencers. This silent region accounts for 84.8%84.8\% of the interview stock-event observations. Within the silent region, in controlled specifications, a ten-percentage-point increase in this buy-minus-sell interview measure predicts 2424 basis points higher future excess return at the five-trading-day horizon and 5050 basis points at the ten-trading-day horizon. The interviews, therefore, do not merely restate public recommendations. They recover stock-level views when the public channel is silent. This evidence is useful for the measurement argument because the covered universe consists of a set of large, liquid, and heavily followed firms for which short-horizon cross-sectional predictability is usually difficult to detect (Fama and French, 2008). We therefore do not interpret the result primarily as a trading-strategy claim. Rather, the fact that the interview signal has predictive content in this relatively efficient part of the market is evidence that the digital-twin instrument is recovering economically meaningful belief proxies. This finding shapes the core contribution of our study. The paper is not primarily a horse race between scraped posts and AI interviews, and it is not mainly a claim that finfluencers can pick stocks. The core contribution is that standardized digital-twin interviews make a previously unmeasured part of selective disclosure visible. Return predictability is important because it shows that the recovered silent-region beliefs have economic content. In this sense, the return tests are corroborating evidence for the measurement strategy. We use social mediaâthe X platform in particularâas a laboratory for our research design because investors increasingly encounter market narratives, stock recommendations, and trading rules of thumb through social media creators.444Recent surveys and regulatory analyses document both the scale of finfluencer activity and associated concerns about advice quality, disclosures, and investor protection (Espeute and Preece, 2024; U.S. Securities and Exchange Commission Investor Advisory Committee, 2024; Financial Industry Regulatory Authority, 2025; Ontario Securities Commission, 2025). A growing empirical literature shows that social-media sentiment, online discussion, and influencer content can move prices, predict near-term returns, and affect market efficiency.555See Antweiler and Frank (2004); Bollen et al. (2011); Sprenger et al. (2014); Ranco et al. (2015); Adams et al. (2023); NĂșñez-Mora and Mendoza-Urdiales (2023); Kakhbod et al. (2023); Gordillo MartĂnez (2024). Social networks also shape how information spreads, with influential users acting as key nodes in amplification and diffusion (Hong et al., 2004; Engelberg et al., 2018; Cookson and Niessner, 2019). The existing social-media finance literature has made substantial progress by extracting sentiment, attention, disagreement, and recommendations from observed text. However, the observed text is only the part of the creatorsâ equity research analysis that becomes public. The unpublished part of their analysisâthe silent regionâis not well understood. The distinction between public posts and the silent region is important because the aggregation of information into prices is neither frictionless nor complete. Our work contributes in several dimensions. First, we contribute to the literature on investor beliefs, sentiment, and disagreement. Models of noise trading and sentiment-driven demand show that optimism and pessimism can move prices away from fundamentals, especially when limits to arbitrage prevent rational traders from correcting mispricing (De Long et al., 1990; Barberis et al., 1998; Shleifer, 2000). Empirical evidence shows that investor sentiment predicts returns, volatility, and trading volume, particularly for hard-to-value securities (Baker and Wurgler, 2006). Disagreement is also central to asset pricing, with dispersion in beliefs linked to trading volume, volatility, and expected returns (Miller, 1977; Hong and Stein, 2007; Daniel et al., 2023). Our setting contributes by measuring public-facing belief proxies for a class of retail-facing intermediaries whose disclosures are episodic, strategic, and incomplete. Second, our paper contributes to the literature on attention and retail-facing information. Investors are more likely to buy stocks that attract their attention, generating predictable demand imbalances (Barber and Odean, 2008; Da et al., 2011). In digital environments, search intensity and online activity provide direct proxies for attention and have been shown to predict trading volume and returns (Da et al., 2011). We take a different approach. Rather than harvesting attention proxies from observed behavior, we repeatedly query the same monitored personas under a fixed protocol. That step widens coverage when attention is absent from the public record and gives the researcher a clean separation between views and disclosure decisions. Third, the paper contributes to the literature on belief and expectation measurement. Directly elicited expectations help explain investor behavior and asset prices (Greenwood and Shleifer, 2014; Bordalo et al., 2019; Giglio et al., 2021). Our setting differs from standard surveys in two ways. The respondentsâthe digital twinsâare public financial intermediaries rather than households or institutions, and the instrument can be run daily at stock-level scale. The paper therefore shows how a survey-like design can be used to measure public-facing belief proxies for market participants who are difficult to survey directly and whose naturally observed communication is selective. Fourth, the paper contributes to work on retail investors and non-traditional information sources. Retail order flow contains information about future returns, and retail trading has become increasingly important in modern markets (Barrot et al., 2016; Boehmer et al., 2021). Episodes such as the GameStop trading frenzy show how coordinated retail attention, often facilitated by social media, can produce large and persistent deviations from fundamental values (Cookson et al., 2022). Sentiment extracted from online forums and social-media platforms also predicts cross-sectional and aggregate market behavior (Sprenger et al., 2014). Our contribution to this literature is on the interpretation of retail-facing market signals (Barber and Odean, 2008; Da et al., 2011; Cookson and Niessner, 2020; Barber et al., 2022). The same interview panel recovers aggregate market stance, disagreement, uncertainty, and stock-selection direction. It also shows that aggregation level matters: broad day-level optimism looks like market mood and carries a contrarian sign, while within-day ranking carries a positive stock-selection relation. Fifth, the paper contributes to the literature on financial text analysis, large language models, and digital twins. Financial text has been used to measure firm fundamentals, product-market relations, political risk, expected returns, and news content (Loughran and McDonald, 2011; Hoberg and Phillips, 2016; Hassan et al., 2019; Chen et al., 2022; Didisheim et al., 2026). Recent work uses large language models for return prediction, financial statement analysis, financial forecasting, and digital-twin applications.666See Lopez-Lira and Tang (2024); Kim et al. (2024); Jadhav and Mirza (2025); Xiao et al. (2025); Yao et al. (2023); Iliut a et al. (2024); He et al. (2025); Koijen and Levy (2026); Chen et al. (2025); Lehner and Lopez-Lira (2026). Our method is different from a retrospective classification of observed text. The unit of observation is an LLM-elicited response to a non-selective query, not a voluntary post or document. Moreover, the responses are generated and archived in real time, so the outcomes are traced after the interviews rather than reconstructed from archival prompts. Finally, our paper relates to the broader literature on media and financial markets. Prior research on traditional media shows that news coverage and tone are associated with trading and market outcomes (Tetlock, 2007; Tetlock et al., 2008; Engelberg and Parsons, 2011), while subsequent research studies analogous information in online forums and social-media content (Antweiler and Frank, 2004; Sprenger et al., 2014; Cookson and Niessner, 2020; Kakhbod et al., 2023). Our setting emphasizes a feature that is especially salient in social media: communication is voluntary and therefore selectively observed. Rather than measuring only the views that finfluencers choose to disclose, we repeatedly elicit standardized public-persona belief proxies under a fixed protocol, including for stocks about which the corresponding finfluencers remain silent. To our knowledge, ours is the first study to measure and evaluate stock-level public-persona belief proxies in this silent region. The remainder of the paper proceeds as follows. Section 2 develops the selective-disclosure problem and explains why digital-twin interviews are useful for measuring the silent region. Section 3 describes the monitored account pool, interview protocol, real-time implementation, and belief-proxy variables. Section 4 validates the instrument and documents the coverage gain from the silent region. Section 5 studies the stock-level information content of the interview responses, including direction, disagreement, uncertainty, and silent-region returns. Section 6 studies the broader, market-level information content of the interviews. Section 7 discusses broader applications and limitations of our research design, and Section 8 concludes. 2. Digital-Twin Interviews Under Selective Disclosure A social media post is a form of disclosure. When a finfluencer posts about a stock, the researcher observes both the content of the post and the decision to make that content public. When the same finfluencer does not post, the researcher observes silence. Silence is ambiguous. It may mean that the finfluencer has no view, that the view is uncertain or weakly held, that the post would not attract attention, or that disclosure would be costly. Thus, a post-based measure mixes the underlying view with the decision to reveal it. The disclosure decision is unlikely to be random. Finfluencers compete for attention and audience, making them more likely to post about stocks that readers are already likely to notice: firms in the news, firms with high trading activity, or firms with sharply moving prices.777This attention motive is consistent with work on online influence, influencer value, retail attention, media attention, and social diffusion (Fainmesser and Galeotti, 2021; De Veirman et al., 2017; Barber and Odean, 2008; Da et al., 2011; Fang and Peress, 2009; Engelberg and Parsons, 2011; Berger and Milkman, 2012; Muchnik et al., 2013; Vosoughi et al., 2018). Recent finfluencer evidence points in the same direction. Financial influencers are more likely to recommend stocks with strong past performance, high trading volume, and high intraday activity (Kakhbod et al., 2023; Lalwani, 2025). Disclosure and reputation models also imply that public communicators may withhold or soften views when disclosure is costly, weakly grounded, inconsistent with audience priors, or reputationally risky.888See Verrecchia (1983), Dye (1985), Scharfstein and Stein (1990), Trueman (1994), Graham (1999), Hong et al. (2000), Mullainathan and Shleifer (2005), and Gentzkow and Shapiro (2006). A public-post sample is therefore likely to overrepresent salient firms and high-conviction views. It is likely to underrepresent less familiar stocks, weaker views, and cases where the finfluencer has a view but no reason to disclose it. A scrape of public posts can measure what was said. It cannot measure the silent region created by selective disclosure. Our empirical design changes the data-generating process. Instead of waiting for finfluencers to post, we ask their digital twins each day a fixed set of questions about a set of pre-determined large-cap stocks. For each monitored finfluencer, the digital twin is constructed from observable material associated with the public persona on X: profile information, recent monitored content, and finance-relevant context. The output from daily digital-twin interviews is a standardized public-persona belief proxy for a given finfluencer. The term digital twin comes from engineering, where virtual models of physical objects or systems are used to simulate, predict, and optimize the behavior of real-world objects (Tao et al., 2018; Jones et al., 2020; Lattanzi et al., 2021; Abdelrahman et al., 2025). Digital twins of human subjects have also begun to appear in medical settings, using data such as health records and patient histories (Eadie et al., 2026; Wang et al., 2024; Angelopoulos et al., 2023; De Bartolomeis et al., 2026; Broska et al., 2025). In these studies, the conditioning material is static: twins are built at the onset of the research design from baseline medical data, health histories, demographics, and patient interviews. Our setting requires a more dynamic version of this idea because the information environment changes as markets move and as finfluencersâ accounts produce new content. In our setting, the same digital twin must, therefore, be interviewed on different dates using updated context. Our research design also differs from an emerging literature that uses LLMs to simulate stock market outcomes or predict stock returns (Mahdavi et al., 2025). That literature adopts a generic approach to return predictability: it does not attempt to construct digital twins conditioned on the public personas of actual human agents. Instead, it aggregates information from public sources to generate buy, sell, or hold recommendations, extract sentiment, construct alpha factors, or simulate markets. Alpha-GPT 2.0, for example, is a project in which LLM agents help identify, test, combine, and analyze trading signals (Yuan et al., 2024). TwinMarket creates generic synthetic traders and lets them interact in a simulated market where they recreate well-known anomalies such as fat tails, bubbles and crashes (Yang et al., 2026). Although these synthetic traders are trained on real transaction data, they are not digital twins of human traders. They are not conditioned on anyoneâs specific public persona and there is no one-to-one mapping between them and corresponding human traders.999More recently, Li et al. (2026) provide a related benchmark for LLM-based investing strategies, showing that apparent LLM trading advantages often deteriorate when evaluated over longer horizons, broader symbol universes, and with explicit controls for survivorship, look-ahead, and data-snooping biases. Their paper evaluates LLMs as trading systems; our design uses LLMs as account-conditioned interview instruments to measure public-persona belief proxies when finfluencers do not disclose a recommendation. Our empirical design and research purpose are different. We do not use the LLM as a trading engine or to simulate market interactions. We use it as an interview tool to recover public-persona belief proxies when the public channel is silent. Moreover, we conduct our research in real time. Ex post LLM prompting can create look-ahead bias because pretrained models may have been exposed to future returns, future news, or later commentary (Glasserman and Lin, 2023; Sarkar and Vafa, 2024; He et al., 2025; Lopez-Lira and Tang, 2024). Thus, our interviews are generated live, timestamped, and archived before the relevant return window begins.101010Koijen and Levy (2026) similarly emphasize the need for real-time experiments when using AI systems in asset pricing, but their study uses LLMs to explain contemporaneous stock-price reactions to observed earnings announcements and transcripts; our study uses LLMs as interview instruments to measure belief proxies generated by finfluencersâ digital twins when public disclosure is absent. Our digital twin methodology offers three unique advantages for studying selective disclosure. First, it expands coverage because we can elicit responses even when no public recommendations exist. Second, it separates disclosure from measurement: public posts reflect both a view and a decision to reveal it, while interview responses are generated on demand by a fixed protocol, whether or not the finfluencer decided to post. Third, it separates four different components of the public-persona belief proxies: stock-pick direction, disagreement, uncertainty, and market-level sentiment. Each of these four components is useful because it answers a different empirical questionâwhether a stock is favored, whether views about it are contested, whether the panel lacks grounded conviction, and whether broad market mood is high or lowâand the fact that these components predict different future outcomes further corroborates that the interviews uncover economically relevant information in the silent region. The next section describes the mechanics of our design: how we find and monitor finfluencer accounts, run the daily digital-twin interviews, and turn the responses into stock-level and macro belief-proxy variables. 3. The Interview Protocol as a Research Instrument Our research design begins with a monitored sample of finfluencer accounts on X. The sample was assembled in three stages. First, we compiled candidate accounts from public X-handle lists in third-party directories published by industry outlets such as The CFO Club, Investopedia, Investing.io, and Yahoo Finance. Second, we manually screened each candidate account against a topical inclusion rule, retaining only accounts whose primary content consisted of directional recommendations or analysis of individual stocks, indices, options, or other derivatives, and excluding real-estate, personal-finance, pure macroeconomic commentary, news-aggregator, and anonymous pump-and-dump accounts. Finally, remaining accounts were retained only if they had at least 5,0005,000 followers, met a minimum post-history threshold, and passed a content-consistency check. Our final sample contains 8181 finfluencer accounts. Importantly, these accounts were selected in June 2025, before any interviews or outcome data were collected, preventing contamination of the research experiment.111111The aim of the finfluencer selection process was not to create a census of every finance account on X, but to build a panel of accounts whose public personas regularly speak about markets and stocks. For a full description of our data-generating processâfinfluencer selection, daily interviews, timing rules, filters, and variable definitionsâsee Section A of the Internet Appendix. We then create a digital twin for each of the 8181 finfluencers. A digital twin in our experimental design is an account-conditioned interview respondent. For each finfluencer, the LLM is given information about the accountâs public persona: profile metadata, recent monitored posts, and finance-relevant context available at the time of the interview. This information gives the model a working memory of the account: what kinds of stocks the account discusses, what language it uses, what themes it emphasizes, and how it tends to frame market views. We ask each twin the same questions under a fixed protocol. Modern LLMs make this design feasible because they can use information supplied in the prompt at the time of the query, follow detailed instructions, and return account-specific answers without training a separate model for each account (Brown et al., 2020; Ouyang et al., 2022; Lewis et al., 2020). The same LLM is therefore used for all finfluencers, but the information supplied to the model differs by account and by date. Importantly, the conditioning material is updated over time as the finfluencer account produces new content and as market conditions change. The resulting responses are time-varying public-persona belief proxies of human finfluencers: they summarize what the public persona says when asked a standardized question at a known time. After creating the digital twins, we repeatedly interview them daily in real time, instructing each of them to answer as the finfluencer would answer if asked the same question at the same time. The daily interview has two main branches: stock-pick and macro. The stock-pick branch asks every digital twin for a view on stocks in the S&P 500 index, recording both a recommendation score (ranging from 0 to 100100) and a speculation score (0 to 100100).121212A recommendation score of 5050 is neutral; higher scores are more buy-leaning and lower scores are more sell-leaning. The speculation scoreâwhere a higher score indicates greater speculationâmeasures whether the twinâs recommendation score is based on specific information in the twinâs profile or is an unsupported guess. The macro branch asks every digital twin about broad market conditions, including recession risk, business conditions, current investor sentiment, stock-market direction, and expected bond-rate direction. The responses from the two interview branches produce two panels: an account Ă date Ă stock panel of stock-level views, and an account Ă date panel of broad-market views. The digital-twin interviews were conducted repeatedly from December 14, 2025 to March 15, 2026.131313Section A of the Internet Appendix provides additional details about the structure, content, and timing of these interviews. The interviews were intended to be conducted daily; however, due to operational backlogs and some missed days around holidays, the sample is almost daily. The timing of the interviews is also critical to our research design, for two reasons. First, to prevent contamination, all interviews are conducted after finfluencer selection, so the interview responses cannot affect which accounts enter the sample. Second, studies that predict stock returns using LLMs can suffer from the look-ahead bias if the model is asked today to consider an earlier market date. Our design avoids this problem by generating and archiving every digital-twin interview in real time and assigning each response to the first trading day on which an investor could have observed and acted on it. This timing convention keeps the experiment forward-looking: the sample is fixed before the interviews, and the return tests use only information that existed before the return window begins.141414Section A of the Internet Appendix provides additional details and descriptive statistics about the timing of these interviews and their relation to the return-measurement period. Table 1 provides descriptive statistics of the macro and stock-pick branches. The macro panel monitors 8181 finfluencer accounts across 5353 trading days from December 15, 2025 through March 16, 2026. The stock-pick sample uses the same 8181 accounts and contains 21,41021,410 stock-by-trading-day observations, covering 429429 unique tickers across 5050 trading days. All market data, including stock prices, returns, market capitalizations, index returns, and volatility inputs, are obtained from S&P Capital IQ, the financial-data platform of S&P Global Market Intelligence.151515The macro and stock-pick branches are timestamped and aligned separately. Macro interviews were typically run first, while stock-pick interviews were completed later, so the 5353 macro trading days and 5050 stock-pick trading days reflect branch-specific completion times rather than missing return data. We also build a third data panel from the finfluencersâ public stock recommendation posts. This is not an interview branch since the posts are voluntary disclosures written by the human finfluencers themselves. Instead, the public-post panel starts from observed posts that mention specific stocks and classifies the direction of the disclosed recommendation. We apply the same timing convention to these public posts, so a public recommendation and a digital-twin response are compared only when they are aligned to the same effective trading date. In the validation tests, this lets us compare public disclosures with digital-twin interview responses for the same account, ticker, and trading date.161616The panel of actual recommendations was created over the same period as the interview branches: December 14, 2025 to March 15, 2026. See Section A of the Internet Appendix for additional details. Table 1: Interview Sample and Analysis Coverage This table summarizes the two interview branches used in the main analyses after timing alignment and final sample filters. Both branches use the 81 finfluencer accounts observed in the shared macro/stock-pick sample; the median account has 36,118 followers in the latest macro snapshot. The macro branch collapses cleaned interviews to trading-day observations. The stock-pick branch collapses account-stock interview responses to stock-by-trading-day observations in the recurring ticker universe. Macro branch Stock-pick branch Analysis observations 53 21,410 Main observation unit Trading day Stock Ă trading day Unique stock tickers â 429 Analysis days 53 trading days 50 trading days Analysis period 2025-12-15 to 2026-03-16 2025-12-15 to 2026-03-16 From the stock-pick panel we construct six daily output variables by aggregating interview responses across the 81 digital twins. These six variables summarize the direction of the twinsâ recommendations as a group (buy, hold or sell), as well as the aggregate level of conviction, disagreement, and uncertainty prevailing among these twins. The six variables are as follows:171717Precise construction details of these variables are provided in Section A of the Internet Appendix. 1. Tilt is the average recommendation score centered around 50, so positive values are buy-leaning and negative values are sell-leaning. 2. Net Buy Share is the share of meaningful buy recommendations minus the share of meaningful sell recommendations, where meaningful recommendations are at least 15 points away from neutral (50). 3. Meaningful Tilt is the Tilt obtained by averaging only meaningful recommendations (those that are at least 15 points different from the neutral value). 4. Conviction-Weighted Meaningful Tilt multiplies Meaningful Tilt by the share of recommendations that are meaningful. This measure preserves the direction of the meaningful recommendations, but gives more weight to stock-dates where that direction is supported by a larger fraction of the interview panel.181818We view this as the cleaner intensity measure because Meaningful Tilt alone conditions on the subset of meaningful responses and therefore can be large even when only a small number of twins express a meaningful view. We report both measures for completeness. 5. Informative Polarization measures disagreement among finfluencersâ twins for recommendations that have low-speculation scores. 6. Uncertainty Score measures the absence of firm conviction among finfluencersâ twins for a particular stock. This score rises when a stock-event has fewer meaningful low-speculation responses. From the macro panel we construct four daily measures of broad market sentiment and disagreement among twins:191919Precise construction details of these variables and others are provided in Section A of the Internet Appendix. 1. Net Sentiment (3 of 7) is the share of interviewed digital twins classified as bullish minus the share classified as bearish. On any given day, a digital twin is classified bullish or bearish only when the responses to at least three of seven core questions are strong and low-speculation. The seven core questions ask about recession risk, business conditions, investor sentiment, stock-market direction, and the expected direction of interest rates. 2. Net Sentiment (3 of 6) is identical to Net Sentiment (3 of 7), except that we exclude the question about the expected direction of future interest rates. 3. Disagreement IQR (3 of 7) is the cross-sectional inter-quartile range of daily Bull scores and measures macro-level disagreement among digital twins. The Bull score is a 0â100 coding of each response to a question, with higher values meaning a more bullish market view and 50 meaning neutral. This variable is high when the digital twins give meaningfully different broad-market views on the same day. 4. Continuous Sentiment is the daily average of account-level normalized Bull scores across the seven core questions. Each answer is transformed to (Bull scoreâ50)/50(Bull score-50)/50, so the scale runs from bearish to bullish. Account-question scores are averaged by day. Then these account-day scores are averaged across accounts to get the daily average. Table 2 reports descriptive statistics for our main variables (output of digital twin interviews, future stock returns and future volatility). Statistics related to the stock-pick branch are reported in Panels A and B; those for the macro branch are reported in Panels C and D. Panel A of Table 2 summarizes interview outcomes for the stock-pick branch. In that panel, 15%15\% of twin-stock responses have meaningful recommendations, on average, and Net Buy Share averages almost 11%11\%, suggesting that finfluencersâ twins are generally optimistic. The high average Uncertainty Score, 0.930.93, reflects the fact that many twin-stock responses are neutral or speculative. Panel B summarizes stock-level market outcomes for the stock-pick branch, for the same (stock Ă trading-day) observations as in Panel A. Future excess returns are measured relative to the S&P 500 and are modestly positive on average, increasing from 0.0670.067 percentage points over one trading day to 0.6190.619 percentage points over ten trading days. These averages are small relative to the cross-sectional dispersion in returns, with standard deviations ranging from 2.302.30 to 7.237.23 percentage points. Future excess realized volatility is also positive, consistent with individual stocks being more volatile than the aggregate market index. Panel C of Table 2 summarizes interview outcomes for the macro branch. Net Sentiment averages 0.3740.374, meaning twins are bullish on the market, on average. However, bullishness is not unanimous, since the median Disagreement IQR is 2222 bull-score points, indicating that meaningful dispersion exists across twins. Panel D summarizes market outcomes for the macro branch. S&P 500 returns are negative on average over the sample. Thus, the positive average macro sentiment in Panel C does not simply reflect a rising market; the digital twins were bullish, on average, during a period of weak broad-market performance. VIX returns are positive on average, consistent with elevated market uncertainty during the sample window, including the geopolitical uncertainty surrounding the U.S.âIran conflict that began on February 28, 2026. Table 2: Summary Statistics for Main Variables This table reports summary statistics for the main belief proxies, stock-level outcomes, and market variables used in the paper. Panels A and B summarize the stock-pick branch at the stock Ă trading-day level. Panels C and D summarize the macro branch at the trading-day level. Return rows labeled Future are measured after the aligned trading day; rows labeled Lagged are measured before it. Share, return, and realized-volatility rows labeled (%) are reported in percentage points. Min and Max are raw extrema after the displayed sample filters. RV abbreviates realized volatility. Net sentiment, polarization, and uncertainty variables that are naturally bounded between zero and one are left on their native scale; Disagreement IQR is reported in bull-score points. N can differ across rows because some variables require market-data availability, nonmissing forward windows, or variable-specific construction screens. Variable Mean SD Min P25 Median P75 Max N Panel A. Stock-pick branch belief proxy variables Recommendation Count (nrâeâcn_rec) 74.06 9.42 1.00 73.00 76.00 77.00 81.00 21,410 Meaningful Rec. Share (%) 15.06 14.31 0.000 5.33 10.39 19.74 100.0 21,410 Tilt 2.95 3.38 -25.00 0.846 2.40 4.29 25.00 21,410 Net Buy Share (%) 10.78 14.23 -100.0 2.53 7.25 15.58 100.0 21,410 Meaningful Tilt 12.45 11.21 -35.00 9.00 16.58 19.00 29.78 20,823 Conv.-Weighted Meaningful Tilt 2.13 3.07 -25.00 0.456 1.42 2.96 25.00 20,823 Informative Polarization 0.210 0.329 0.000 0.000 0.000 0.438 1.00 10,536 Uncertainty Score 0.932 0.076 0.150 0.914 0.956 0.978 1.00 21,410 Panel B. Stock-level size, returns, and volatility windows Log market cap 10.82 1.23 2.16 10.02 10.64 11.46 15.37 21,410 Future Excess Return, 1d (%) 0.067 2.30 -26.95 -1.04 0.022 1.18 22.46 21,410 Future Excess Return, 5d (%) 0.268 5.09 -32.43 -2.39 0.125 2.82 42.64 21,410 Future Excess Return, 10d (%) 0.619 7.23 -36.70 -3.39 0.403 4.44 44.93 21,410 Future Excess RV, 5d (%) 4.25 2.80 0.238 2.50 3.57 5.16 39.33 21,410 Future Excess RV, 10d (%) 6.24 3.43 0.925 4.02 5.43 7.45 40.81 21,410 Panel C. Macro branch belief proxy variables Contributing Influencers per Day 76.06 3.25 63.00 75.00 77.00 78.00 80.00 53 Net Sentiment (3 of 7) 0.374 0.087 0.182 0.317 0.377 0.447 0.557 53 Disagreement IQR (3 of 7) 22.46 4.54 13.57 20.00 21.43 24.29 37.14 53 Net Sentiment (3 of 6) 0.366 0.086 0.169 0.312 0.361 0.434 0.544 53 Continuous Sentiment 0.270 0.060 0.123 0.231 0.273 0.322 0.358 53 High-Speculation Share, Q1âQ7 0.238 0.094 0.173 0.193 0.202 0.219 0.501 53 Panel D. Market return and VIX return windows Future S&P 500 Return, 1d (%) -0.040 0.779 -2.06 -0.434 -0.008 0.539 1.97 53 Future S&P 500 Return, 5d (%) -0.240 1.29 -2.45 -1.15 -0.402 0.642 3.13 53 Future S&P 500 Return, 10d (%) -0.883 1.77 -5.31 -2.45 -0.524 0.525 2.04 53 Lagged VIX Return, 5d (%) 0.936 6.29 -9.88 -2.41 -0.052 3.89 23.72 53 Lagged VIX Return, 10d (%) 1.43 9.26 -13.65 -6.53 1.29 5.87 25.94 53 4. Validation and Coverage Gain 4.A. Validation Against Observable Benchmarks: Does the Instrument Work? Before using the interview output to study market outcomes, we first ask whether the digital twins produce responses that are consistent with the public personas of the finfluencers they represent. Of course, we cannot observe a finfluencerâs private beliefs, and we do not claim to do so. Instead, we validate the interview output where validation is possible. Table 3 reports four tests that address three validation questions. First, when a finfluencer makes a public stock recommendation, does the digital-twin interview point in the same direction, ranking public buys above public sells? Second, before a finfluencerâs public recommendation appears, does the digital-twin already lean toward that later public disclosure? Third, after removing common daily market conditions, do digital-twin responses still uncover specific views of the corresponding finfluencers rather than generic knowledge of the LLM?202020Section B of the Internet Appendix provides additional evidence in support of the validation tests presented in Table 3. Table 3: Validation of Digital-Twin Interview Belief Proxies This table reports four validation tests that address three concerns about the digital-twin interview belief proxies. Panel A uses the exact-overlap sample of 4,063 same-account, same-ticker, same-date public recommendation overlaps. Sign alignment is the share of overlaps in which the interview response and public recommendation point in the same direction, either buy-buy or sell-sell. The all non-neutral row uses every overlap with a non-neutral interview response, while the meaningful low-speculation row keeps only interviews far enough from neutral and not highly speculative. The AUC row reports the area under the curve, or the probability that an interview score ranks a public buy above a public sell. Panel B asks whether interviews lean toward later public recommendations during the five prior trading days; positive signed-tilt values indicate alignment with the eventual public recommendation. Panel C uses macro questions Q1âQ7 in the 81-account interview sample and residualizes responses by removing date-question means. It reports the incremental R2R^2 from including account fixed effects, an adjacent-day same-account similarity score, and top-1 and top-5 account re-identification rates. Difference is Actual minus Benchmark. Percentage rows are in percentage points; signed tilt is in recommendation-score points; and AUC, R2R^2, and similarity are unitless. The Benchmark column reports the comparison value used for each test. The p-value column reports the validation-test p-value for the Actual-Benchmark difference. N is row-specific: the sign-alignment rows use the directional overlap subset shown in the row, the AUC row uses all exact overlaps, Panel B uses pre-disclosure validation cases, and Panel C uses residual macro-response rows or held-out account-days. For the incremental account R2R^2 row, N is the number of account-date-question macro bull-score responses used in the fixed-effect calculation, not the number of account-days. (1) (2) (3) (4) (5) Panel A. Public-Recommendation Overlap Measure Actual Benchmark Difference p-value N All non-neutral sign alignment 89.5% 50.0% 39.5 p <0.001<0.001 3,271 Meaningful low-speculation sign alignment 91.5% 50.0% 41.5 p <0.001<0.001 2,439 Buy-sell ranking AUC 0.776 0.522 0.255 <0.001<0.001 4,063 Panel B. Pre-Disclosure Belief Recovery Measure Actual Benchmark Difference p-value N Pre-disclosure signed tilt 5.32 2.60 2.72 <0.001<0.001 1,500 Matched pre-disclosure signed tilt 4.99 2.28 2.71 <0.001<0.001 1,500 Panel C. Account-Specific Belief Structure Measure Actual Benchmark Difference p-value N Incremental account R2R^2 0.338 0.000 0.338 <0.001<0.001 27,997 Adjacent-day same-account similarity 0.600 0.023 0.577 <0.001<0.001 3,900 Top-1 account re-identification 10.8% 1.2% 9.5 p <0.001<0.001 1,951 Top-5 account re-identification 35.2% 6.2% 29.0 p <0.001<0.001 1,951 Panel A of Table 3 reports the results of the first two validation tests. In this Panel, we ask the cleanest validation question: when a finfluencer account publicly recommends a stock on a given trading date, does the digital-twin interview about that stock point in the same buy or sell direction? The exact-overlap sample contains 4,0634,063 account-ticker-date observations. Among meaningful low-speculation interview responsesâresponses far enough from neutral and not highly speculativeâthe interview direction aligns with the public recommendation 91.5%91.5\% of the time, compared with a 5050-5050 benchmark. Using all non-neutral interview scores, the alignment rate is 89.5%89.5\%.212121Interview direction is determined from the digital-twin recommendation score after centering it at 5050: scores above 5050 are buy-leaning and scores below 5050 are sell-leaning. Public recommendation direction is classified from the observed public-post recommendation panel, as described in Section A of the Internet Appendix. The meaningful low-speculation subset requires interview recommendation scores at least 1515 points from neutral and speculation scores of 4040 or below. The all non-neutral measure uses any non-neutral interview score. Thus, when the public-post panel gives us a same-day benchmark, the digital-twin interview usually points in the same direction. Panel A of Table 3 also asks whether, in the exact-overlap sample, digital-twin interview scores rank publicly disclosed buy recommendations above publicly disclosed sell recommendations. This ranking test uses the area-under-the-curve (AUC) statistic. Let spublic buyDâTs^DT_public buy denote the digital-twin interview score for a randomly chosen public-buy recommendation, and define spublic sellDâTs^DT_public sell analogously for a randomly chosen public-sell recommendation. Then AâUâC=PrâĄ(spublic buyDâT>spublic sellDâT).AUC= (s^DT_public buy>s^DT_public sell). The AUC statistic is 0.7760.776. In other words, if we randomly draw one public buy and one public sell recommendation from the overlap sample, the digital-twin interview score ranks the public buy above the public sell 77.6%77.6\% of the time. As a placebo test, we shuffle the public buy and sell labels within each event date and then recompute the same AUC statistic. This benchmark AUC is 0.5220.522.222222Shuffling within event date preserves the number of public buys and sells on each date, but breaks the link between the public recommendation direction and the digital-twin interview score. Also, the difference between the actual AUC statistic and the placebo, 0.2550.255, is statistically significant (p<0.001<0.001). This second check gives the same message as the first. Public buys receive higher interview scores than public sells.232323Table A1 in the Internet Appendix provides additional validation results. Panel B of Table 3 presents the results of our third validation test. We seek to determine whether the digital twin interviews merely restate public recommendations after those recommendations have already appeared. We address this question by asking whether the digital-twin interviews already lean in the direction of later public recommendations. For each account-ticker pair, we identify the first directional public recommendation, then examine the digital-twin interviews for the same account and ticker in the five prior trading days.242424Public recommendations are first collapsed to the account-ticker-date level using the effective trading date of the public post. Here, directional means that the public recommendation score is at least 6565 for a buy or at most 3535 for a sell. Neutral public recommendations are excluded. Let t be the effective trading date of that first directional public recommendation. For an interview k trading days earlier, define signed tilt as SâiâgânâeâdâTâiâlâti,j,tâk=Di,jâ(RâeâcâSâcâoârâei,j,tâkDâTâ50),k=1,âŠ,5,SignedTilt_i,j,t-k=D_i,j(RecScore^DT_i,j,t-k-50), 23.49976ptk=1,âŠ,5, where RâeâcâSâcâoârâeDâTRecScore^DT is the digital-twin stock-pick recommendation score, Di,j=1D_i,j=1 if that first directional public recommendation is a buy, and Di,j=â1D_i,j=-1 if it is a sell. Positive values therefore mean that the earlier interview leaned toward the eventual public direction. In the five trading days before the public post, signed tilt averages 5.325.32 points, compared with 2.602.60 points for same-account, same-date placebo tickers.252525The difference between the two is statistically significant (p<0.001<0.001). We also use a matched benchmark. For each future-public ticker, this comparison uses the closest available pre-disclosure interview date and compares it with other tickers from the same account and interview date that do not receive a public recommendation over the next five trading days. The matched comparison gives a similar result: future-public tickers average 4.994.99 signed-tilt points, compared with 2.282.28 points for matched control tickers.262626Again, the difference between these two values is statistically significant (p<0.001<0.001). This result shows that even before public recommendations, digital-twin interview responses lean toward those future recommendations. Panel C of Table 3 presents the results of our fourth and final validation test. Here we seek to determine whether the interviews uncover specific views of the digital twins rather than generic knowledge of the LLM. For this test we use the macro-interview branch, where every digital twin answers the same recurring broad-market questions on the same dates. For each question on each date, we subtract the average response across twins. For example, if a twin answered macro question #1 with a score of 6060, and the average answer to question #1 across all twins that day was 5050, the residualized answer is 1010. This removes the common daily component: the part of the answer that could come from market news, broad sentiment, generic market commentary, or the wording of the question itself. We then ask whether the residualized responses still reveal which finfluencer account they came from. If the responses were primarily generic LLM output, the identity of the twin account would explain little of the remaining variation after this common component is removed. We test this first using account fixed effects.272727The fixed-effects comparison is between BâuâlâlâSâcâoârâei,q,t=Îłq,t+Ï”i,q,tBullScore_i,q,t= _q,t+ _i,q,t and BâuâlâlâSâcâoârâei,q,t=Îłq,t+αi+Ï”i,q,tBullScore_i,q,t= _q,t+ _i+ _i,q,t, where Îłq,t _q,t absorbs common date-question effects and αi _i captures account-specific structure. If the residual answers are generic, then knowing the account should add little explanatory power. Instead, adding account fixed effects raises explained variation by 0.3380.338. This means that account identity still explains a meaningful share of the interview responses after the common date-question component has been removed. Panel C also reports two more direct diagnostics. The first asks whether the same account has a stable residualized response pattern across nearby interview days. For each account on each date, we take the residualized responses to the recurring macro questions and treat them as that accountâs response pattern for the day. We then use cosine similarity to compare that pattern with the same accountâs pattern on the next interview day.282828Each account-date response vector contains the residualized bull-score responses to macro questions Q1âQ7, where residualization subtracts the event-date-by-question mean. We keep account-dates with all seven responses observed and compute cosâĄ(x,y)=xâČây/(âxâââyâ) (x,y)=x y/(\|x\|\|y\|) between an accountâs response vector and the same accountâs next observed response vector. The reported statistic is the mean across these adjacent same-account pairs. If the digital twin is not generic and is taking on the persona of a finfluencer account, we would expect the twinâs views to be auto-correlated. Indeed, we find that the similarity score is 0.6000.600 when we compare a twinâs response pattern with that same twinâs pattern on the next interview day, compared with 0.0230.023 when we compare it with a placebo account. The second diagnostic is an out-of-sample identification test. The idea is simple. If all digital twins are just repeating the same generic market commentary, then once we remove the common daily market component, it should be difficult to tell which account produced which answer. If the interviews are truly account-conditioned, then each account should leave some stable pattern in its answers. To test this, we split the data into two parts. We use the first part to build a reference profile for each account. For each account, this reference profile summarizes how that accountâs digital twin tends to answer the recurring macro questions after removing the common response on that date. We then take the remaining held-out observations, which were not used to build the reference profiles, and ask whether each held-out response pattern can be matched back to the correct account. For each held-out account-date observation, we compare its residualized response pattern to the reference profiles of all 81 accounts. The comparison uses cosine similarity again, measuring whether two response patterns point in the same direction across the macro questions. We then rank all 81 accounts from most similar to least similar. If the digital-twin answers were generic, the correct account should appear near the top of the ranking only by chance.292929With 8181 accounts, the chance benchmarks are 1/81=1.2%1/81=1.2\% for the closest match and 5/81=6.2%5/81=6.2\% for the top-five match. Instead, the correct account is the closest match 10.8%10.8\% of the time and appears in the top five matches 35.2%35.2\% of the time. The result shows that, even after removing the common daily market component, the interview outputs retain twin-specific structure. Taken together, these validation tests support the main premise of our research design. We show that outcomes from digital twin interviews line up with public recommendations when we can observe them, lean toward later recommendations before they appear, and can be traced back to a specific finfluencer even after common market conditions are removed. 4.B. The Coverage Gain from the Silent Region Having established that the digital-twin responses are consistent with the public personas of the corresponding human finfluencers, we now want to understand the magnitude and composition of the silent region. Recall that the silent region is the set of stock-events for which the digital twin interview produces a recommendation without any matching recommendation publicly posted by the corresponding human finfluencer. The evidence, presented below in Table 4, makes two simple points. First, the silent region is large. Most stock-events in the interview panel would be absent from a sample that is based exclusively on human posts, so the digital twin interviews substantially expand the set of measured stock-events. Second, public recommendations are concentrated among a smaller set of firms, especially larger, technology-heavy firms. The coverage gain, therefore, does more than just increase the number of observations. It also uncovers parts of the stock universe that public posts underrepresent. Over our sample period, the digital-twin interviews generate 21,41021,410 stock-date observations. Only 3,2513,251, or 15.2%15.2\%, coincide with a same-stock, same-date public post. The remaining 18,15918,159 stock-dates, or 84.8%84.8\% of the sample, have no same-date public post. Thus, a post-based sample would miss most of the stock-events for which we have interview responses. The silent region is large, and the interview protocol substantially expands coverage. We also ask whether finfluencersâ public recommendations are spread evenly across the recurring stock-pick universe. Table 4 shows that they are not. Panel A of Table 4 groups firms by the number of distinct public-post days during the stock-pick interview window. Sixty-one firms, or 14.2%14.2\% of the stocks in our sample, receive no public-post days at all. More broadly, 222222 firms have no more than four public-post days, so roughly half of the covered universe receives little or no public recommendation attention. At the other end, 5454 high-attention firms receive 2020 or more public-post days. These firms average 36.736.7 public-post days and 147.3147.3 public posts. Panel B of Table 4 shows that human finfluencers pay more attention to larger firms and those that are in the business equipment and technology industry. Thus, public recommendations are tilted toward large, technology-heavy firms rather than spread evenly across firms. Table 4: Public Recommendation Attention and Firm Characteristics This table describes how public stock-recommendation attention is distributed across the 429-stock universe during the interview window. The sample contains 4,063 account-stock-date events that have a same-account, same-stock, same-date interview and public post match. Panel A groups firms by the number of distinct public-post dates from December 15, 2025 through March 16, 2026; a public-post date is a date on which the firm has at least one public stock-recommendation post. Mean public days and mean public posts are per-firm averages within each bin. Mean posts per public day is the average firm-level ratio of public posts to public-post dates, computed among firms with at least one public-post date. Panel B compares low-, middle-, and high-attention firms, defined as 0â4, 5â19, and 20 or more public-post days. These groups contain 222, 153, and 54 firms. The last two columns report the high-minus-low difference and Welch two-sample t-statistic. Continuous characteristics are winsorized at the 1st and 99th percentiles. Market capitalization is reported in billions of dollars. Indicators ***, **, * denote statistical significance at the 1%, 5%, and 10% level, respectively. (1) (2) (3) (4) (5) Panel A. Public Recommendation Attention Distribution Share of Mean Mean Mean posts Public-post-day bin Firms firms public days public posts per public day 0 61 14.2% 0.0 0.0 1 52 12.1% 1.0 1.1 1.1 2 36 8.4% 2.0 2.2 1.1 3 37 8.6% 3.0 3.3 1.1 4 36 8.4% 4.0 4.6 1.2 5â9 82 19.1% 6.5 8.1 1.2 10â19 71 16.6% 13.8 20.1 1.4 20+ 54 12.6% 36.7 147.3 3.3 Panel B. Firm Characteristics by Public Recommendation Attention Low Middle High Variable 0â4 5â19 20+ High â- Low t-stat Market cap ($bn) 38.53 87.28 630.22 591.70â 5.29 Log market cap 10.29 10.99 12.62 2.33â 13.03 Listing age (years) 35.91 35.14 33.71 -2.20 -0.81 Business equipment / tech (%) 14.41 19.74 29.63 15.22â 2.27 Manufacturing (%) 9.91 9.21 12.96 3.05 0.61 Healthcare (%) 8.56 9.21 3.70 -4.85 -1.51 Other (%) 17.57 17.11 22.22 4.65 0.74 The coverage gain from digital-twin interviews is therefore not just an increase in the sample size. The key gain is that the method makes the silent part of selective disclosure measurable. Public posts tell us what finfluencers choose to reveal. Digital-twin interviews ask the same public personas the same questions under a fixed protocol, including when no public recommendation appears. This means that the interview protocol can capture views about stocks that are underrepresented in public posts. 5. Cross-Sectional Predictability as Evidence of Information Content The central question in our paper is whether the digital twin interviews contain useful information about future stock outcomes. In this section, we show that they do: interview responses predict both future excess returns and future excess realized volatility in the cross-section of stocks.303030The tests presented in this Section evaluate the informational content of twinsâ interviews. They do not evaluate the profitability of an implementable trading strategy. While we do form calendar-time portfolios, we do not study trading costs or liquidity. Our goal is to ask whether the interview variables contain economically meaningful information about future stock outcomes. The stock-pick interview branch produces three types of signals pertaining to future stock returns and volatility. Directional signals measure whether the interview responses are buy-leaning or sell-leaning for a given stock on a given date. Disagreement signals measure the extent to which low-speculation recommendations are split between buys and sells. And uncertainty variables measure whether the panel of responses lacks conviction in either direction. We study these three signals in Tables 5 through 9. 5.A. Directional Signals and Future Stock Returns We begin by studying the relation between directional signals obtained from digital twin interviews and the cross-section of future stock returns. Because event dates in our panel are spaced one day apart, return-measurement windows longer than one trading day necessarily overlap: adjacent h-day outcomes reuse many of the same daily returns. This dependence does not by itself imply biased coefficient estimates, but it induces serial correlation in the dependent variables and in the sequence of estimated cross-sectional slopes, so standard errors that ignore the overlap can overstate the level of statistical significance. We address this issue in the next two subsections. First, in Subsection 5.A.1 we perform FamaâMacBeth regressions using NeweyâWest standard errors with lag length hâ1h -1. We also perform calendar-time portfolio tests that aggregate all active event-date vintages into a daily longâshort return series, so that our statistical inference is based on the time-series variation in returns. Second, in Subsection 5.A.2, we perform staggered-vintage regressions in which we space anchor dates h trading days apart, eliminating overlap in future-return windows within each vintage. Together, these approaches assess whether the directional signal is robust to alternative treatments of overlapping outcomes. 5.A.1. FamaâMacBeth Regressions and Calendar Time Portfolios Table 5 reports the first stock-level return tests, which ask whether stocks with more buy-leaning interview responses earn higher future excess returns than other stocks. The four panels of Table 5 (A-D) each use a different directional signal as independent variable (Tilt, Net Buy Share, Meaningful Tilt, and Conviction-Weighted Meaningful Tilt). In Columns (1)â(3) of Table 5 we estimate cross-sectional return predictability using FamaâMacBeth regressions (Fama and MacBeth, 1973). Specifically, for each event date in the stock-pick branch, we estimate EâxâRâeâtj,t(h)=at,h+bt,hâDâiârâeâcâtâiâoânj,t+Δj,t(h).ExRet_j,t^(h)=a_t,h+b_t,hDirection_j,t+ _j,t^(h). Here, EâxâRâeâtj,t(h)ExRet_j,t^(h) is stock jâs future excess return over event horizon h, measured relative to the S&P 500, and Dâiârâeâcâtâiâoânj,tDirection_j,t is one of four directional signals obtained from twin interview responses, measured before the return window begins. Regression coefficients bt,hb_t,h are averaged across time (t), with standard errors computed from their intertemporal variation. These standard errors are NeweyâWest adjusted to account for serial correlation induced by overlapping return windows when the event horizon, h, exceeds one trading day (Newey and West, 1987). We report the intertemporal averages, bÂŻh b_h, of these regression coefficients for three event horizons: one, five, and ten trading days. The results show that the directional signals predict future excess returns, especially at the ten-day horizon. The clearest measure is Net Buy Share (Panel B), the share of meaningful buy recommendations minus the share of meaningful sell recommendations for a stock-date. A ten-percentage-point increase in Net Buy Share is associated with 4040 basis points higher excess return over the next ten trading days. The five-day estimate is also positive, though not statistically significant, while the one-day coefficient is close to zero. Thus, the relation is weak immediately after the interview, but stronger over the next several trading days. The other directional variables provide similar results. Conviction-Weighted Meaningful Tilt (Panel D), which multiplies the average meaningful recommendation tilt by the share of recommendations that are meaningful, also predicts higher future returns: a one-unit increase is associated with 16.916.9 basis points higher excess return over the next ten trading days.313131Since Conviction-Weighted Meaningful Tilt has a standard deviation of 3.073.07, a one-standard-deviation increase in Conviction-Weighted Meaningful Tilt is associated with 51.951.9 basis points higher excess return over the next ten trading days. Tilt and Meaningful Tilt (Panels A and C) are also positive at the ten-day horizon. These measures are constructed differently, but they tell the same story. Stocks with more positive digital-twin interview responses tend to earn higher future excess returns. Table 5: Digital-Twin Stock-Pick Direction and Future Excess Returns This table combines daily FamaâMacBeth stock-pick return-predictability regressions with calendar-time long-short portfolio tests. Columns (1)â(3) report daily FamaâMacBeth cross-sectional regressions of cumulative future excess stock returns on one digital-twin direction signal measured before the return window. Excess returns are measured relative to the S&P 500 and reported in percentage points over 1, 5, and 10 trading days. The table reports the time-series average of the daily slope estimates. Columns (4)â(6) report calendar-time portfolio tests using the same signals. On each event date, stocks are ranked by the indicated signal, the portfolio buys the top quintile and shorts the bottom quintile with equal weights within each leg, and each event-date vintage is held for the next 10 trading days, excluding the formation day. The daily calendar-time long-short return averages all active vintages on each holding date. H-L Mean and CAPM Alpha are daily percentage-point estimates while 10-day Alpha is 10 times the daily CAPM alpha. Market adjustment is estimated from a daily CAPM regression of the long-short return on the S&P 500 return. NeweyâWest standard errors are in parentheses, using lag hâ1h-1 in columns (1)â(3) and lag 9 in columns (4)â(6) to allow for serial correlation from overlapping return windows and overlapping calendar-time vintages. FamaâMacBeth coefficients are scaled to a +10 recommendation-point move in Tilt or Meaningful Tilt, a +10 percentage-point move in Net Buy Share, or a +1-unit move in Conviction-Weighted Meaningful Tilt. N/days reports stock-event observations in columns (1)â(3) and calendar holding days in columns (4)â(6). Mean adjusted R2R^2 applies only to the FamaâMacBeth columns. The sample uses 429 stocks and 81 accounts over 50 event dates from December 15, 2025 to March 16, 2026; calendar-time holding dates run from December 16, 2025 to March 30, 2026. Indicators ***, **, * denote significance at the 1%, 5%, and 10% level. (1) (2) (3) (4) (5) (6) Dep. var.: future excess return (p) Dep. var.: H-L portfolio return (p) 1 day 5 days 10 days H-L Mean CAPM Alpha 10-day Alpha Panel A. Tilt Direction signal -0.010 0.639 1.594â 0.133â 0.120â 1.203â (0.124) (0.424) (0.611) (0.061) (0.055) (0.549) N / days 21,410 21,410 21,410 71 71 71 Mean adj. R2R^2 0.009 0.013 0.013 Panel B. Net Buy Share Direction signal -0.002 0.166 0.404â 0.120â 0.108â 1.082â (0.030) (0.105) (0.152) (0.066) (0.061) (0.608) N / days 21,410 21,410 21,410 71 71 71 Mean adj. R2R^2 0.010 0.014 0.016 Panel C. Meaningful Tilt Direction signal 0.020 0.112 0.293â 0.109â 0.051 0.514 (0.034) (0.083) (0.109) (0.050) (0.041) (0.408) N / days 20,823 20,823 20,823 71 71 71 Mean adj. R2R^2 0.010 0.008 0.004 Panel D. Conviction-Weighted Meaningful Tilt Direction signal 0.000 0.065 0.169â 0.133â 0.117â 1.170â (0.013) (0.048) (0.070) (0.057) (0.052) (0.519) N / days 20,823 20,823 20,823 71 71 71 Mean adj. R2R^2 0.010 0.013 0.014 In Columns (4)â(6) of Table 5 we use calendar-time portfolios as an alternative methodology to account for the overlapping return windows. Following Mitchell and Stafford (2000), these tests aggregate the stock-level rankings into a daily high-minus-low portfolio return series and estimate performance from the time-series variation of that portfolio. The calendar-time test therefore asks a related portfolio question: do stocks ranked more favorably by the digital-twin panel outperform stocks ranked less favorably when the signal is implemented as a daily long-short portfolio? On each event date, we rank stocks by a direction signal, buy the top quintile, short the bottom quintile, and hold that event-date portfolio for the next ten trading days. On each calendar day, we then average the returns of all active event-date vintages. Column (4) reports the average daily high-minus-low return, column (5) reports the CAPM alpha from a regression of the daily high-minus-low return on the S&P 500 return, and column (6) reports the same alpha multiplied by ten to put it on a 10-trading-day scale. The calendar-time results support the same interpretation as the event-time regressions. For Net Buy Share, the high-minus-low portfolio earns an average daily return of 12.012.0 basis points and a CAPM alpha of 10.810.8 basis points per day, corresponding to a 10-day alpha of 108108 basis points. The estimates for Tilt and Conviction-Weighted Meaningful Tilt are similar and statistically significant. Meaningful Tilt has a positive high-minus-low mean return, but its CAPM alpha is smaller and not statistically significant.323232This weaker result is not surprising because Meaningful Tilt conditions only on the subset of meaningful responses and does not account for how broadly that meaningful direction is supported across the interview panel. Once the same directional information is weighted by the share of meaningful recommendations (Conviction-Weighted Meaningful Tilt), the calendar-time alpha is again positive and statistically significant. Overall, the calendar-time evidence shows that the stock-pick signal also appears in a portfolio-level design, not only in event-time cross-sectional regressions.333333Table A2 in the Internet Appendix replicates Table 5 using value weights both in the FamaâMacBeth regressions and in the calendar-time portfolios. The estimates in this table are much smaller and not statistically distinguishable from zero, indicating that the relation between interview signals and future returns is weaker among larger firms. Figure 1 presents the same directional results visually. Stocks are sorted into event-day deciles based on Net Buy Share, and the figure plots average future excess returns for each decile. The one-day panel is essentially flat, but the relation becomes positive at longer horizons. By the ten-day horizon, the top-minus-bottom decile spread is 153.3153.3 basis points. Figure 1 corroborates the statistical result from Table 5: stocks that the digital twins rank more favorably tend to have higher subsequent excess returns, especially beyond the one-day horizon. Figure 1: Digital-Twin Stock-Pick Net Buy Share and Future Excess Returns This figure sorts stock-events into event-date deciles of Net Buy Share and plots equal-weight mean cumulative future excess returns. Returns are measured relative to the S&P 500 and reported in basis points at the 1-, 3-, 5-, and 10-trading-day horizons. Deciles are formed within each event date. For each decile, the plotted value is the average return across event dates. D10-D1 is the top-minus-bottom decile spread. Shaded bands denote 90% confidence intervals computed with NeweyâWest standard errors using lag length hâ1h-1 for horizon h. The sample uses 81 accounts, the 429-stock universe, and 50 stock-pick event dates from December 15, 2025 through March 16, 2026. Table 6 asks whether the results in Table 5 survive controls for lagged stock-level excess returns and realized volatility. For each horizon h and event date t, we estimate the cross-sectional regression EâxâRâeâtj,t(h)=at,h+bt,hâDâiârâeâcâtâiâoânj,t+ÎŽt,hâLâaâgâEâxâRâeâtj,t(h)+Ξt,hâLâaâgâEâxâVâoâlj,t(h)+Δj,t(h).ExRet_j,t^(h)=a_t,h+b_t,hDirection_j,t+ _t,hLagExRet_j,t^(h)+ _t,hLagExVol_j,t^(h)+ _j,t^(h). The controls are horizon-matched: LâaâgâEâxâRâeâtj,t(h)LagExRet_j,t^(h) is the stockâs cumulative excess return over the prior h trading days, and LâaâgâEâxâVâoâlj,t(h)LagExVol_j,t^(h) is realized excess-return volatility over the same prior window. Regression coefficients (bt,hb_t,h, ÎŽt,h _t,h, and Ξt,h _t,h) are averaged across time (t), with NeweyâWest standard errors computed from the time series of daily coefficients using lag length hâ1h-1 (Newey and West, 1987). The controlled specifications support the results in Table 5. All four directional variables continue to predict the cross-section of stock returns at the ten-day horizon. For example, in Panel B, a ten-percentage-point increase in Net Buy Share is associated with 38.538.5 basis points higher future excess return at the ten-day horizon. Panel D gives the same message with Conviction-Weighted Meaningful Tilt, where a one-point increase is associated with 15.815.8 basis points higher ten-day excess return.343434A one-standard-deviation increase in Conviction-Weighted Meaningful Tilt is associated with 48.548.5 basis points higher excess return over the next ten trading days. Overall, this table shows that the directional signals are not simply restating a stockâs recent excess return or realized volatility, but continue to predict future excess returns after those controls. Table 6: Digital-Twin Stock-Pick Direction and Future Excess Returns with Controls This table repeats the daily FamaâMacBeth specifications in columns (1)â(3) of Table 5 after adding horizon-matched lagged excess return and lagged excess realized-volatility controls to each daily FamaâMacBeth cross-section. Return horizons, direction-signal scaling, NeweyâWest standard errors, N, and mean daily adjusted R2R^2 follow the Table 5 conventions; N is the total number of usable stock-event observations across daily cross-sections. Control coefficients are percentage-point return changes for a one-percentage-point increase in the control. The sample uses 429 stocks, 81 accounts in the stock-pick interview branch, and 50 event dates from December 15, 2025 through March 16, 2026. Indicators ***, **, * denote statistical significance at the 1%, 5%, and 10% level, respectively. (1) (2) (3) Dependent variable: Cumulative future excess return (p) 1 trading day 5 trading days 10 trading days Panel A. Tilt Direction signal 0.035 0.686 1.515â (0.125) (0.440) (0.805) Lagged excess return -0.029 0.031 0.061 (0.023) (0.019) (0.066) Lagged realized vol. 0.064â 0.118â 0.118 (0.034) (0.055) (0.081) N 21,410 21,410 21,410 Mean daily adj. R2R^2 0.042 0.035 0.051 Panel B. Net Buy Share Direction signal 0.008 0.175 0.385â (0.030) (0.107) (0.195) Lagged excess return -0.029 0.031 0.060 (0.023) (0.019) (0.066) Lagged realized vol. 0.063â 0.118â 0.117 (0.034) (0.055) (0.079) N 21,410 21,410 21,410 Mean daily adj. R2R^2 0.043 0.036 0.054 Panel C. Meaningful Tilt Direction signal 0.034 0.134 0.306â (0.033) (0.087) (0.172) Lagged excess return -0.030 0.028 0.060 (0.023) (0.018) (0.063) Lagged realized vol. 0.060â 0.119â 0.126 (0.034) (0.056) (0.083) N 20,823 20,823 20,823 Mean daily adj. R2R^2 0.041 0.030 0.042 Panel D. Conviction-Weighted Meaningful Tilt Direction signal 0.006 0.070 0.158â (0.013) (0.049) (0.091) Lagged excess return -0.033 0.031â 0.062 (0.023) (0.019) (0.066) Lagged realized vol. 0.058â 0.121â 0.126 (0.034) (0.054) (0.080) N 20,823 20,823 20,823 Mean daily adj. R2R^2 0.042 0.035 0.052 5.A.2. Staggered-Vintage Design The FamaâMacBeth and Calendar-Time methods reported above in Subsection 5.A.1 provide two established ways to obtain statistical inference when multi-day return windows overlap. In our sample, however, both of these methods are likely to have low statistical power. In the FamaâMacBeth tests, standard errors are obtained from the intertemporal variation in date-specific cross-sectional slope estimates. In the calendar-time tests, standard errors are obtained from the inter-temporal variation of daily longâshort portfolio returns. Both methods, therefore, draw their statistical power primarily from the time-series dimension of the data, which is relatively short in our setting compared with the much larger cross-section of stocks observed on each event date.353535See Subsection 7.B for a discussion of this paperâs limitations associated with the sample size. A pooled panel regression can potentially provide greater statistical power because it retains the full stock-level cross-section rather than reducing each date to a single slope estimate or portfolio return. The difficulty is that, at horizons longer than one trading day, adjacent dependent variables reuse many of the same daily returns. Pooling such observations without adequately addressing this mechanical overlap can understate standard errors and produce misleading inference. To address this problem, we implement a staggered-vintage design that combines the larger cross-sectional information set with non-overlapping future-return windows. Within each vintage, anchor dates are spaced h trading days apart, ensuring that the corresponding h-day return windows do not overlap; shifting the initial anchor date across vintages allows us to examine every possible calendar alignment. This design thus provides a complementary, potentially higher-power test of the signalâreturn relation without mechanically overlapping dependent-variable windows. The results are presented in Table 7. Table 7: Staggered-Vintage Stock-Pick Direction and Future Excess Returns This table reports staggered-vintage tests of whether digital-twin stock-pick direction predicts future excess returns after controlling for lagged excess returns and realized volatility. Panel A uses five-day formation and return windows and five vintages; Panel B uses ten-day windows and ten vintages. Each vintage is a separate regression sample with a different starting date and non-overlapping future-return windows. Signals and controls are measured over the formation window ending on each anchor date, and the dependent variable is the stockâs compounded excess return relative to the S&P 500 over the following return window, in percentage points. Each vintage regression includes anchor-date fixed effects and standard errors clustered by stock. The table reports the mean and range of the vintage coefficients, the number that are positive, and the number that are positive and significant at the 10% level. Mean N and adjusted R2R^2 are averaged across vintages. Because the vintages share the same calendar, the counts are descriptive. Coefficients are scaled to a +10 recommendation-point increase in Tilt or Meaningful Tilt, a +10 percentage-point increase in Net Buy Share, or a +1-unit increase in Conviction-Weighted Meaningful Tilt. The final four rows show two-sided empirical p-values from 10,000 placebo repetitions; a small value means that the observed mean coefficient is unusual under that placebo. The unrestricted placebo reassigns complete four-signal histories across stocks. The AR simulation placebo creates artificial histories with similar distributions and persistence. The size and industry placebos reassign observed histories without self-matches within initial market-capitalization quartiles or FamaâFrench 12 industries. The industry design preserves within-industry comovement in signals and returns while breaking the link between a stock and its own signal history. The sample comprises 429 stocks and 81 stock-pick accounts, with anchor dates from December 15, 2025 through March 16, 2026. (1) (2) (3) (4) Direction Signal Tilt Net Buy Share Meaningful Tilt Conv.-Wt. Meaningful Tilt Panel A. 5-Trading-Day Formation and Return Windows Mean coefficient 0.723 0.190 0.185 0.080 Vintage range [0.470, 0.886] [0.131, 0.232] [0.097, 0.251] [0.051, 0.101] Positive vintages 5/5 5/5 5/5 5/5 Positive and significant 5/5 5/5 4/5 5/5 Mean N 5,320 5,320 5,298 5,298 Mean adj. R2R^2 0.054 0.055 0.054 0.054 Placebo p-values Unrestricted 0.0006 0.0002 0.0018 0.0006 AR simulation 0.0010 0.0004 0.0128 0.0028 Size quartiles 0.0002 0.0002 0.0016 0.0002 F12 industries 0.1802 0.0642 0.7661 0.1176 Panel B. 10-Trading-Day Formation and Return Windows Mean coefficient 1.687 0.444 0.434 0.184 Vintage range [1.275, 2.133] [0.339, 0.546] [0.339, 0.495] [0.140, 0.229] Positive vintages 10/10 10/10 10/10 10/10 Positive and significant 10/10 10/10 10/10 10/10 Mean N 2,660 2,660 2,659 2,659 Mean adj. R2R^2 0.042 0.043 0.041 0.042 Placebo p-values Unrestricted 0.0004 0.0002 0.0010 0.0006 AR simulation 0.0014 0.0016 0.0210 0.0042 Size quartiles 0.0002 0.0002 0.0024 0.0002 F12 industries 0.0966 0.0268 0.4748 0.0858 The basic idea is to partition the data into several regression samples, which we call vintages. The last day on which we measure the signal and controls is the anchor date; the future-return window begins on the following trading day. Consider one five-day vintage whose first anchor is Friday, February 27, 2026. We measure signals and controls from February 23 through February 27 and predict returns from March 2 through March 6. The next anchor in that vintage is March 6, and its return window is March 9 through March 13. The two future-return windows are adjacent but do not overlap. Shifting the first anchor forward one trading day creates the next vintage. Repeating that shift produces five different starting dates at the five-day horizon and ten at the ten-day horizon. Panel A of Table 7 presents results for the five-day return measurement horizon. The construction of five-day vintages is summarized below using numbered trading days. Each row is a separate regression sample. Reading across a row shows that one return window ends immediately before the next begins. Reading down the rows shows how the starting date shifts. Vintage Anchor positions Future-return windows Vintage 1 1, 6, 11, ⊠2â6, 7â11, 12â16, ⊠Vintage 2 2, 7, 12, ⊠3â7, 8â12, 13â17, ⊠Vintage 3 3, 8, 13, ⊠4â8, 9â13, 14â18, ⊠Vintage 4 4, 9, 14, ⊠5â9, 10â14, 15â19, ⊠Vintage 5 5, 10, 15, ⊠6â10, 11â15, 16â20, ⊠Panel B of Table 7 presents results for the ten-day return measurement horizon, with ten-day vintages constructed as follows: Vintage Anchor positions Future-return windows Vintage 1 1, 11, 21, ⊠2â11, 12â21, 22â31, ⊠Vintage 2 2, 12, 22, ⊠3â12, 13â22, 23â32, ⊠⟠⟠⟠Vintage 10 10, 20, 30, ⊠11â20, 21â30, 31â40, ⊠Throughout Table 7, the dependent variable is future excess return, the independent variable is one of the four dâiârâeâcâtâiâoânâaâldirectional sâiâgânâaâlâssignals, and the controls are the same horizon-matched lagged excess-return and realized-volatility measures used in Table 6. Anchor-date fixed effects preserve the within-date comparison across stocks, and standard errors are clustered by stock. Columns (1)â(4) provide separate analyses for each of the four dâiârâeâcâtâiâoânâaâldirectional sâiâgânâaâlâssignals. Within each column, we pool the stock-level observations assigned to a vintage and estimate one regression. The variables of interest are the coefficients bv b_v of each dâiârâeâcâtâiâoânâaâldirectional sâiâgânâaâlsignal, for each vintage, v. This approach produces five coefficients at the five-day horizon (Panel A) and ten coefficients at the ten-day horizon (Panel B) for each signal. In the top half of each Table 7 panel, we report the mean and range of these bv b_v coefficients, the number that are positive, and the number that are positive and statistically significant at the 10%10\% level (using a two-tail test): Vintage 1 sampleâ¶(b^1,p1)Vintage 2 sampleâ¶(b^2,p2)âźVintage n sampleâ¶(b^n,pn)â¶mean and range of âb^vnumber of positive âb^vnumber that are positive and significant . array[]rclVintage 1 sample& &( b_1,p_1)\\ Vintage 2 sample& &( b_2,p_2)\\ & &\\ Vintage $n$ sample& &( b_n,p_n) array \ array[]cmean and range of b_v\\ number of positive b_v\\ number that are positive and significant array For these conventional regression tests, the null hypothesis is H0:bv=0H_0:b_v=0. Therefore, zero is the reference value for the vintage-specific t-statistics and for the significance counts reported in the top half of each panel. The coefficient bv b_v is positive in every vintage and statistically significant at the 10% level in nearly every one. All 2020 five-day coefficients are positive, and 1919 of them are positive and significant at the 10%10\% level (see top half of Panel A). All 4040 ten-day coefficients are positive and significant at the 10% level, and their magnitudes are also similar to the controlled FamaâMacBeth estimates (see top half of Panel B). A ten-percentage-point increase in Net Buy Share (see Column (2)) predicts an average of 19.019.0 basis points higher five-day excess return across vintages, with estimates ranging from 13.113.1 to 23.223.2 basis points. The corresponding ten-day average is 44.444.4 basis points, with a range of 33.933.9 to 54.654.6 basis points. Table 6 reports comparable estimates of 17.517.5 and 38.538.5 basis points.363636Conviction-Weighted Meaningful Tilt gives the same message. A one-unit increase predicts an average of 8.08.0 basis points at five days and 18.418.4 basis points at ten days, with vintage ranges of 5.15.1â10.110.1 and 14.014.0â22.922.9 basis points. The corresponding Table 6 estimates are 7.07.0 and 15.815.8 basis points. Thus, our results survive non-overlapping return windows and every possible starting date. The vintages represent alternative calendar alignments of the same finite sample and are not independent; the observed signals also exhibit time-series persistence. To account for these features, we use simulated placebo distributions to assess whether the staggered-vintage procedure, when applied to signals with similar empirical properties but no stock-specific link to future returns, could generate an average coefficient as large as the one observed. We construct four placebo benchmarks that preserve different features of the observed data. Three reassign complete observed signal histories among stocks, while the fourth generates artificial histories with similar distributions and persistence. For each placebo draw, we rerun the full staggered-vintage analysis and calculate the mean coefficient across vintages. Repeating this process 10,000 times produces a reference distribution against which to compare the observed mean of the bv b_v coefficients. The placebo tests use a different inferential benchmark. In the bottom half of each Table 7 panel, we compare the observed mean coefficient across vintages with the empirical distribution of mean coefficients generated by the 10,000 random draws under each placebo procedure. The reported two-sided empirical p-value measures how unusual the observed mean coefficient is relative to the corresponding placebo distribution, rather than whether it differs from zero.373737The empirical mean of the placebo distribution summarizes the center of this benchmark distribution. Thus, zero and the placebo benchmark coincide approximately when the simulated distribution is centered near zero, but they need not coincide, for example, when industry-level comovement shifts the placebo distribution to the right of zero. We examine these industry comobements in our last placebo test below. The unrestricted placebo (top line in each panelâs bottom half) reassigns each stockâs complete four-signal history across the 429429-stock universe. This preserves the observed time pattern and the relationships among the four direction measures but breaks the link between a stockâs signal history and its returns. Under this empirical distribution, the p-values of the mean bv b_v coefficients are 0.00180.0018 or smaller, suggesting that the observed directional signals do not produce similar coefficients when they are assigned arbitrarily across stocks. This result is also illustrated visually in Panel A of Figure 2, which shows eight histograms comparing the mean bv b_v coefficients in the actual data with those produced by the unrestricted placebo simulations. In each histogram, the gray bars show the empirical distribution of the 10,00010,000 placebo mean coefficients, the shaded area contains the 90%90\% confidence interval, the dashed line shows the average of the placebo coefficients, and the solid line to the right is the mean bv b_v coefficient in the actual data. Across all four measures and two event horizons, the unrestricted distributions are centered near zero, while the solid lines lie far to the right, outside the 90%90\% confidence interval. This visual separation corroborates the p-values shown in the table. Panel A. Unrestricted Placebo Panel B. Industry Placebo Figure 2: Placebo Distributions for Staggered-Vintage Return Coefficients This figure compares the mean staggered-vintage coefficients in the actual data with distributions produced by 10,000 placebo reassignments. Panel A randomly reassigns each stockâs complete four-signal history across the 429-stock universe. Panel B reassigns complete histories without self-matches among stocks in the same FamaâFrench 12 industry. Each reassignment moves the four direction measures together and is used at both return horizons. Rows correspond to the four direction measures, and columns correspond to the five- and ten-day horizons. In each histogram, gray bars show the placebo mean coefficients, the shaded area contains the middle 90% of them, the dashed line marks their mean, and the solid red line marks the mean coefficient in the actual data. Reported p-values are two-sided empirical comparisons between the observed coefficients and their placebo distributions. Coefficient scaling and the sample follow Table 7. The AR simulation placebo (second line in each panelâs bottom half) asks whether persistent signals in a short sample could mechanically produce similar coefficients. It generates artificial histories with the same overall signal distributions and approximately the same adjacent-date persistence as the observed histories.383838For Net Buy Share, the observed mean and standard deviation are 0.10780.1078 and 0.14230.1423, compared with averages of 0.10780.1078 and 0.14210.1421 across the artificial panels. The average adjacent-date correlation is 0.3610.361 in the data and 0.3590.359 in the simulations. For the other direction measures, observed average correlations range from 0.2610.261 to 0.4610.461, compared with 0.2620.262 to 0.4650.465 in the simulations. The four measures are generated separately, so the exercise reproduces each measureâs behavior rather than the exact mathematical links among them. The AR simulation p-values range from 0.00040.0004 to 0.02100.0210, indicating that persistence and the short sample do not by themselves explain the observed positive bv b_v coefficients. The Size-quartile placebo (third line in each panelâs bottom half) restricts the reassignment to stocks in the same beginning-of-sample market-capitalization quartile. This preserves broad size differences while breaking the stockâdirectional-signal match. Every p-value obtained under the Size-quartile distribution is 0.00240.0024 or smaller, suggesting that broad differences in firm size do not explain the positive bv b_v coefficients either. Finally, in the F12 industries placebo (bottom line of each panel) we reassign directional signals to other stocks using each signalâs observed history, but restrict the reassignment more sharply. We move a stockâs complete history to another stock in the same FamaâFrench 12-industry classification, using the four-digit SIC-code mapping provided in the Kenneth R. French Data Library,393939See https://mba.tuck.dartmouth.edu/pages/faculty/ken.french/data_library/det_12_ind_port.html. and no stock keeps its own history. The test therefore asks whether the actual directional signals produced by digital twins remain informative after keeping the shared industry component in place. Because signals and returns are both correlated within industries, these placebo distributions are shifted to the right of zero, as shown in Panel B of Figure 2. This shift is expected and makes the F12 industries placebo a more stringent benchmark than the preceding placebos: the observed mean coefficient must be large relative to a null distribution that already preserves industry-level comovement in both signals and returns.404040Digital-twin signal histories are correlated within industries because the interviews can express common sector and thematic views across related stocks. Returns of stocks in the same industry also move together. By reassigning histories only within industries, the placebo keeps both sources of industry comovement in place while breaking the match between a particular stock and a particular signal history. Nonetheless, even relative to this more stringent benchmark, the observed mean bv b_v coefficients remain statistically significant at the 10% level for the ten-day horizon in the case of three of our four directional signals: Tilt, Net Buy Share, and Conviction-Weighted Meaningful Tilt (Columns (1), (2), and (4) of Panel B). The mean bv b_v coefficient obtained with Net Buy Share also preserves its significance at the 10% level for the five-day horizon (Column (2) of Panel A). For example, for a ten-percentage-point increase in Net Buy Share, excess returns obtained under the industry-placebo distribution are 11.411.4 basis points at the five-day horizon and 25.325.3 basis points at ten-day horizon, compared with observed coefficients of 19.019.0 and 44.444.4 basis points. The corresponding p-values are 0.06420.0642 and 0.02680.0268. Taken together, the placebo tests conducted with the Unrestricted, AR simulation, and Size-quartiles methods show that the statistical significance of bv b_v estimates is not readily explained by arbitrary signal-to-stock matching, broad size differences, or persistent signals in a short sample. The F 12 industries placebo tests show that some predictability is shared within industries, as expected when the interviews elicit sector and thematic views. Yet, after preserving within-industry correlation in both signals and returns, Net Buy Share remains statistically significant at both five- and ten-day horizons, as do ten-day Tilt and Conviction-Weighted Meaningful Tilt. Thus, while industry comovement appears to contribute to the signal-return relation, it does not fully account for it. 5.B. The Silent Region Return Signal We now return to the silent region, which consists of stock-date pairs that are not specifically covered by human finfluencers. We ask whether cross-sectional return predictability in the silent region is different from predictability resulting from actual posts. We show that return predictability is much stronger in the silent region. Recommendations made by digital twins that are associated with public posts have weaker predictability. Table 8 compares public recommendations with the additional coverage created by the interview protocol. The table follows the same controlled FamaâMacBeth design as Table 6, and each panel again uses one directional signal measure. The difference is that each panel now separates three sources of information. The first row uses the public-post signal on stock-dates that have both a public post and an interview signal for the same stock on the same date. The second row uses the digital-twin interview signal on those same overlap stock-dates. The third row uses the digital-twin interview signal on stock-dates with no same-date public post. Because the signal definitions are fixed across public posts and interviews, differences across rows reflect differences in the information used to build the signal: observed public recommendations, interview responses on stock-dates that overlap with public posts, or interview responses in the no-post region.414141The four direction measures are constructed the same way for public posts and interviews. Recommendations are mapped to the same 0â100 scale, neutral is 50, buy and sell indicators use the same 65/35 cutoffs, and the same formulas produce Tilt, Net Buy Share, Meaningful Tilt, and Conviction-Weighted Meaningful Tilt. The difference is support. Public-post signals use only observed public recommendations, so an overlap stock-date is often based on a single public account-stock-date event; in the overlap sample, the median public stock-date has one such event. As a result, public Net Buy Share is often a discrete value such as â1-1, 0, or 11, whereas the interview signal averages across the digital-twin panel responses for the same stock-date. The three rows ask three separate questions. The top row (public-signal-overlap) asks whether the visible recommendation predicts returns. The middle row (interview-signal-overlap) holds fixed the publicly visible stock-date and asks whether the broader interview panel adds information on that same stock-date. The bottom row (interview-signal-no-post) asks whether the interview protocol recovers directional information in the silent region. The results show two clear patterns. First, public recommendations do not predict future excess returns in the overlap sample. Across all four signal definitions, the public-signal coefficients are small and statistically insignificant. Second, the interview signal is strongest in the silent region. For stock-dates with no same-date public post, the interview coefficients are positive across three of the four direction measures and become larger at longer horizons. A ten-percentage-point increase in interview Net Buy Share is associated with 2424 basis points higher five-day excess return and 5050 basis points higher ten-day excess return. The Tilt and Conviction-Weighted Meaningful Tilt rows again show the same broad pattern, while Meaningful Tilt is statistically insignificant. Together, the results show that the silent region does more than increase coverage. It is where the interview design recovers its strongest directional return signal. The interview-signal estimates on the overlap row answer a narrower question. When a stock is already publicly discussed, does the broader interview panel add anything beyond the public post? On overlap dates, the public channel reveals that at least one account chose to post about the stock, while the interview protocol adds contemporaneous views from many accounts that did not post. Most interview-overlap coefficients are not statistically distinguishable from zero, so this evidence should not be overstated. The main exceptions are Net Buy Shareâwhere a ten-percentage-point increase in interview Net Buy Share is associated with 3131 basis points higher ten-day excess returnâand Tilt. This suggests that broader interview coverage may add some information even on stock-dates with public posts, but the evidence is much weaker than in the no-post region. Table 8: Coverage Gain, Public Recommendations, and Interview Signals This table compares observed public recommendations with the additional stock coverage created by the interview protocol. The dependent variable is cumulative future excess stock return relative to the S&P 500, in percentage points. Each panel reports controlled daily FamaâMacBeth regressions for one signal definition in three samples: public-post signals on exact public-post/interview stock-date overlaps, interview signals on those same overlaps, and interview signals on stock-dates with no same-date public post. Public and interview signals use the same 0â100 recommendation-score scale, neutral score of 50, buy/sell cutoffs of 65/35, meaningful cutoff of 15 points from neutral, and aggregation formula. Public-signal rows are observed-disclosure aggregates and may be based on few posts; interview-signal rows are digital-twin panel aggregates. All rows use 429 stocks and 81 accounts (65 have public stock-recommendation posts). The interview panel contains 21,410 stock-events: 3,251 (15.2%) coincide with a same-stock, same-date public recommendation post, and 18,159 (84.8%) have no same-date public post. Return columns report 1-, 5-, and 10-trading-day horizons. N reports usable stock-event observations after requiring the signal, future return, and horizon-matched lagged excess-return and realized-volatility controls. Coefficient scales are +10 percentage points for Net Buy Share, +10 recommendation points for Tilt and Meaningful Tilt, and +1 unit for Conviction-Weighted Meaningful Tilt. NeweyâWest standard errors with lag hâ1h-1 are shown beneath estimates. Indicators ***, **, * denote significance at the 1%, 5%, and 10% level. (1) (2) (3) (4) Dependent variable: Cumulative future excess return (p) N 1 trading day 5 trading days 10 trading days Panel A. Tilt Public Signal (Overlap) 3,251 -0.027 0.008 0.077 (0.063) (0.111) (0.170) Interview Signal (Overlap) 3,251 -0.043 0.411 1.072â (0.170) (0.440) (0.605) Interview Signal (No Post) 18,159 0.190 1.017â 2.087â (0.147) (0.455) (0.906) Panel B. Net Buy Share Public Signal (Overlap) 3,251 -0.008 -0.003 0.003 (0.011) (0.025) (0.016) Interview Signal (Overlap) 3,251 -0.008 0.134 0.312â (0.041) (0.100) (0.132) Interview Signal (No Post) 18,159 0.044 0.242â 0.503â (0.035) (0.114) (0.218) Panel C. Meaningful Tilt Public Signal (Overlap) 2,476 0.014 0.030 -0.113 (0.056) (0.134) (0.185) Interview Signal (Overlap) 3,225 0.129 0.021 0.238 (0.092) (0.166) (0.187) Interview Signal (No Post) 17,589 0.025 0.149 0.310 (0.032) (0.092) (0.202) Panel D. Conviction-Weighted Meaningful Tilt Public Signal (Overlap) 2,476 0.001 0.005 -0.003 (0.006) (0.015) (0.019) Interview Signal (Overlap) 3,225 -0.001 0.033 0.074 (0.017) (0.043) (0.046) Interview Signal (No Post) 17,589 0.024 0.109â 0.239â (0.017) (0.056) (0.109) 5.C. Disagreement, Uncertainty, and Future Volatility We now turn to evaluating disagreement and uncertainty signals obtained from our digital twin interviews and ask whether they have different implications for future returns and volatility. This question is important for the paperâs contribution. In public posts, silence is hard to interpret. A stock may be missing because views are split, because conviction is low, or because no one chose to post. Digital-twin interviews reduce this problem in two ways. They ask the same accounts about the same stock universe on each event date, and they ask each account to report both its directional view and its uncertainty about that view. That design lets us separate clear disagreement from broad uncertainty. Unlike the preceding subsection (5.B), the tests here use all stock-dates in the interview panel, not only the silent region, and the state variables are constructed solely from interview responses. Our analysis is summarized in Table 9, which uses the same controlled FamaâMacBeth design as the return-predictability tables above. For each outcome Y, horizon h, and event date t, we estimate the cross-sectional regression Yj,t(h)=at,hY+bt,hYâSâtâaâtâej,t+ÎŽt,hYâLâaâgâEâxâRâeâtj,t(h)+Ξt,hYâLâaâgâEâxâVâoâlj,t(h)+Δj,tY,(h).Y_j,t^(h)=a_t,h^Y+b_t,h^YState_j,t+ _t,h^YLagExRet_j,t^(h)+ _t,h^YLagExVol_j,t^(h)+ _j,t^Y,(h). Here, Yj,t(h)Y_j,t^(h) is either stock jâs future excess return or future excess realized volatility over horizon h. The state variable Sâtâaâtâej,tState_j,t is either Informative Polarization or Uncertainty Score, measured from the digital-twin stock-pick interviews before the outcome window begins. The controls, LâaâgâEâxâRâeâtj,t(h)LagExRet_j,t^(h) and LâaâgâEâxâVâoâlj,t(h)LagExVol_j,t^(h), are measured over the matching pre-event horizon. The table reports the average of the daily slope estimate, bt,hb_t,h, with NeweyâWest standard errors computed from the time series of daily coefficients using lag length hâ1h-1 (Newey and West, 1987). Informative Polarization is a proxy for disagreement. This variable measures whether digital twins with clear, low-speculation views about a stock disagree with each other. Informative Polarization is high when these informative interview recommendations are split between buys and sells, and it is zero when they all point in the same direction. Thus, the variable captures disagreement among informed, non-neutral views, not general uncertainty. Uncertainty Score is a proxy for the broader concept of uncertainty. This variable is high when recommendations are mostly neutral, when non-neutral recommendations are speculative, or when meaningful recommendations lack supporting information. It is low when many digital twins give clear buy or sell recommendations that are grounded in the supplied account-specific material. Both state variables are normalized to a scale from zero to one, and the table reports coefficients for a 0.100.10-unit increase in these state variables.424242Section A of the Internet Appendix provides additional details on how these two state variables were constructed and normalized to the range from zero to one. Panel A of Table 9 shows the results using Future Excess Returns as the dependent variable. Informative Polarization is negatively related to future returns across the reported horizons. The ten-day coefficient implies that a ten-percentage-point increase predicts 27 basis points lower future excess return. This negative relation is broadly consistent with theoretical and empirical research linking disagreement to lower subsequent returns. In Miller (1977), investorsâ inability to register pessimistic views causes prices to place greater weight on relatively optimistic valuations, implying lower subsequent returns. Diether et al. (2002) document a similar empirical relation using analyst forecast dispersion. Because our measure captures disagreement among digital-twin interview responses rather than analyst forecasts, we view this comparison as suggestive rather than direct. The evidence is weaker for Uncertainty Score. Its ten-day coefficient is also negative, but it is not statistically significant. This is not too surprising because uncertainty captures the speculative or noncommittal nature of responses, rather than disagreement between informative bullish and bearish views. Panel B of Table 9 shows the results using Future Excess Volatility as dependent variable. This Panel provides the most relevant results from Table 9. Informative Polarization predicts higher future volatility at every horizon. A ten-percentage-point increase predicts 1212 basis points higher excess realized volatility at the ten-day horizon. This is what we would expect from a disagreement variable. When clear, low-speculation recommendations are split between the buy and sell sides, the stockâs future price path becomes more volatile. In contrast, Uncertainty Score has the opposite pattern. A ten-percentage-point increase predicts 2727 basis points lower excess realized volatility at the ten-day horizon, with the same negative and significant sign at the shorter horizons. The contrast between these two state variables is a key contribution of our paper. The positive sign on Informative Polarization shows that our digital twin interviews recover not only directional informational content but also content about future volatility. When twins have strong, diverging opinions about a given stock, the stockâs returns are more volatile in the future. This result further corroborates the validity of our experimental design in that it provides a different setting in which interviews obtained from digital twins reveal value-relevant information. Uncertainty Score, on the other hand, captures weak convictions among digital twins. A stock-date pair with a high Uncertainty Score means that the digital twinsâ directional views are mostly neutral, speculative, or unsupported. That corresponds to a low-attention state. A low-attention stock should have fewer opinion-driven trades, thus lower future realized volatility. This interpretation is consistent with models in which investor attention increases volatility by increasing the amount of information processed into prices. In Andrei and Hasler (2015), time-varying attention to past dividends generates time-varying stock-market volatility. Our Uncertainty Score appears to capture the opposite state: a lack of grounded attention or conviction in the interview panel, which is followed by less volatile returns in the future. Table 9: Digital-Twin Stock-Pick Disagreement, Uncertainty, and Future Outcomes This table reports daily FamaâMacBeth regressions of future outcomes on digital-twin stock-pick responses before the outcome window. Panel A uses cumulative future excess returns, measured relative to the S&P 500. Panel B uses future excess realized volatility. Outcomes are reported in percentage points over 1, 5, and 10 trading days. Coefficients are percentage-point changes in the outcome for a +0.10 increase in the independent variable. Informative Polarization captures disagreement among meaningful low-speculation recommendations. Uncertainty Score is one minus average certainty coverage, where meaningful recommendations receive certainty weight 1âspeculation/1001-speculation/100 and non-meaningful responses receive zero weight. All rows use 429 stocks and 81 accounts. Each daily cross-section includes horizon-matched controls for lagged excess return and lagged excess realized volatility. NeweyâWest standard errors with lag hâ1h-1 are shown beneath estimates. Informative Polarization requires at least three meaningful low-speculation recommendations, so those rows use 10,536 stock-event observations over 50 event days; Uncertainty Score rows use 21,410 stock-event observations over 50 event days. All rows also require nonmissing future outcomes and matching lagged controls. The sample runs from December 15, 2025 through March 16, 2026. Indicators ***, **, * denote significance at the 1%, 5%, and 10% level. (1) (2) (3) Panel A. Cumulative Future Excess Returns Dependent variable: Cumulative future excess return (p) 1 trading day 5 trading days 10 trading days Informative Polarization -0.038â -0.093â -0.270â (0.017) (0.045) (0.063) Uncertainty Score 0.007 -0.198 -0.378 (0.071) (0.259) (0.419) Panel B. Future Excess Realized Volatility Dependent variable: Future excess realized volatility (p) 1 trading day 5 trading days 10 trading days Informative Polarization 0.038â 0.100â 0.117â (0.010) (0.020) (0.034) Uncertainty Score -0.105â -0.198â -0.266â (0.034) (0.068) (0.058) 5.D. Informational Content of Digital Twin Interviews Taken together, the results in Section 5 show that the stock-pick branch of the digital-twin interviews recovers informative signals about future stock returns and volatility. Directional signals, such as Net Buy Share, predict the cross-section of stock returns, especially in the silent region where the human finfluencer does not make a public recommendation. Disagreement among digital twins predicts lower future returns and higher future volatility, consistent with the idea in Miller (1977) and the related analyst disagreement literature that followed. Uncertainty behaves differently. Our Uncertainty Score is high when the panel gives mostly neutral, speculative, or weakly supported recommendations. It therefore captures the absence of attention or absence of grounded conviction, not fundamental uncertainty in the usual asset-pricing sense.434343Our use of the term uncertainty differs from the traditional distinction between risk and uncertainty in economics. Since Knight (1921), risk typically refers to situations in which the relevant probability distribution is known or can be measured, while uncertainty refers to situations in which the probability distribution itself is unknown. Ellsberg (1961) provides an illustration of this distinction, showing that individuals treat known probabilities differently from unknown or ambiguous probabilities. Our Uncertainty Score is different. It does not measure uncertainty about the true probability distribution of stock returns. It measures the absence of grounded conviction, or the lack of attention, in the digital-twin panel: recommendations are neutral, speculative, or weakly supported by account-specific material. The fact that high uncertainty predicts lower future volatility is consistent with the attention-based interpretation: stocks that draw little attention from the digital twins have lower future volatility, consistent with the predictions of Andrei and Hasler (2015). These results further support the validity of our research design. Digital twins do not merely produce generic market commentary. Their responses contain stock-level information that maps into different future outcomes: direction predicts returns, disagreement predicts volatility, and uncertainty predicts stability. In this sense, the twins behave like account-conditioned representations of finfluencersâ public personas. And, their interviews produce structured responses that are useful precisely because they are tied to the public-facing accounts we study. This Sectionâs main contribution, therefore, is about measuring the silent region under selective disclosure. Our predictability tests are not designed to prove that finfluencers are good stock pickers or that an investor could earn positive abnormal returns by following the twinsâ recommendations. Additional evidence, including out-of-sample validation, would be required to make that claim. The result here is narrower, but central to the paper. Digital-twin interviews produce stock-level public-persona belief proxies with relevant economic content, particularly in cases where public disclosures do not exist. They identify which stocks the panel favors, which stocks divide the panel, and which stocks draw weak conviction. This is the kind of information our research design was intended to recover. 6. Macro Beliefs and Future Market Returns Section 5 studies the stock-level information contained in the stock-pick branch of the digital-twin interviews. We now turn to the macro branch, which asks the same digital twins about broad market conditions, including recession risk, business conditions, investor sentiment, market direction, and interest rates. These questions and their responses let us ask whether the interviews also recover market-level belief proxies. We estimate predictive regressions of the form RâeâttM,(h)=αh+ÎČhâMâaâcârâoâSâiâgânâaâlt+ÎhâCâoânâtârâoâlâst(h)+Δt(h),Ret_t^M,(h)= _h+ _hMacroSignal_t+ _hControls_t^(h)+ _t^(h), where RâeâttM,(h)Ret_t^M,(h) is the future cumulative S&P 500 return over horizon h, and MâaâcârâoâSâiâgânâaâltMacroSignal_t is a macro belief proxy, either Sentiment or Disagreement, measured before the return window begins. The controls are horizon-matched lagged S&P 500 returns and lagged VIX returns. Standard errors are NeweyâWest adjusted with lag length hâ1h-1. This test is distinct from the stock-level tests in Section 5. There, the tests ask whether stocks favored by the digital-twin panel outperform other stocks on the same date. Here, the analysis asks whether market-level beliefs on a given date predict future broad-market returns. Thus, the two branches of the interview protocol speak to different objects, with the stock-pick branch measuring relative stock preference and the macro branch measuring aggregate market beliefs. Table 10 first asks whether average macro Sentiment predicts future market returns. The table reports three sentiment measures. Panel A uses Net Sentiment 3 of 7, defined as the share of interviewed digital twins classified as bullish minus the share classified as bearish. A twin is classified as bullish when at least three of the seven core macro responses are strongly bullish and low-speculation, and bearish when at least three are strongly bearish and low-speculation. Panel B uses the analogous measure based on six macro questions, excluding the interest-rate question. Panel C uses a continuous sentiment score that averages the bullishness of the macro responses.444444Section 3 describes these variables. Section A of the Internet Appendix gives the exact construction of all sentiment variables. Table 10: Digital-Twin Macro Sentiment and Future Market Returns This table reports controlled predictive regressions of future cumulative S&P 500 returns on macro sentiment from the macro interview branch. Returns are measured in percentage points over 1, 5, and 10 trading days. Each specification includes horizon-matched lagged S&P 500 returns and lagged VIX returns as controls. Panels AâC identify the sentiment measure. Coefficients are percentage-point changes in future S&P 500 returns for a +1-unit increase in the sentiment measure. NeweyâWest standard errors are shown beneath each estimate, with lag length hâ1h-1 for horizon h. N reports trading-day observations. The sample uses the 81-account macro interview branch and contains 53 trading days from December 15, 2025 through March 16, 2026. Indicators ***, **, * denote statistical significance at the 1%, 5%, and 10% level, respectively. (1) (2) (3) Dependent variable: Future S&P 500 return (p) 1 trading day 5 trading days 10 trading days Panel A. Net Sentiment (3 of 7) Sentiment 0.991 3.090 3.330 (1.247) (3.067) (3.030) Lagged return -0.233 -0.588â -0.280 (0.194) (0.177) (0.363) Lagged VIX return -0.013 -0.116â -0.140â (0.036) (0.034) (0.049) N 53 53 53 Adj. R2R^2 -0.025 0.245 0.459 Panel B. Net Sentiment (3 of 6) Sentiment 1.385 3.442 4.276 (1.248) (2.938) (2.689) Lagged return -0.247 -0.589â -0.246 (0.195) (0.175) (0.353) Lagged VIX return -0.015 -0.114â -0.131â (0.036) (0.033) (0.046) N 53 53 53 Adj. R2R^2 -0.014 0.251 0.468 Panel C. Continuous Sentiment (Quadratic) Sentiment -1.910 15.707 27.572 (13.033) (17.919) (31.567) Sentiment2 6.007 -7.785 -24.224 (25.109) (36.885) (68.233) Lagged return -0.201 -0.619â -0.369 (0.228) (0.133) (0.286) Lagged VIX return -0.007 -0.071â -0.089â (0.043) (0.026) (0.050) N 53 53 53 Adj. R2R^2 -0.050 0.388 0.549 Across all three measures of Sentiment and three investment horizons, the Sentiment coefficients are not statistically distinguishable from zero. Thus, in this sample, average macro sentiment from the digital-twin panel does not reliably predict future broad-market returns. Table 11 next asks whether Disagreement among the digital twinsâ macro views predicts future market returns. This parallels the stock-pick analysis in Table 9, where Informative Polarization measures disagreement across stock-level recommendations and predicts lower future stock returns. The macro setting asks the analogous market-level question. Two days can have similar average sentiment, but on one day the twins may largely agree, while on another day their views may be far apart. We measure aggregate Disagreement using Disagreement IQR, the interquartile range of account-level bull scores.454545Bull scores translate each macro response to a common 0â100100 scale, oriented so that higher values are more bullish for the stock market. The account-level composite bull score averages those responses within an account-day. Section A of the Internet Appendix gives the exact construction of this variable. Panel A reports controlled specifications with Disagreement as the main independent variable. Panel B adds Net Sentiment 3 of 7 as a second independent variable, to ask whether Disagreement contains information beyond the average bullishness of the panel. In both panels, the regressions include horizon-matched lagged S&P 500 returns and lagged VIX returns as controls. The results show a negative relation between disagreement and future market returns. In Panel A, a 1010-point increase in the bull-score Disagreement measure predicts 77.177.1 basis points lower S&P 500 returns over the next five trading days and 86.486.4 basis points lower returns over the next ten trading days. The ten-day coefficient remains negative and statistically significant after adding Net Sentiment 3 of 7 as a control in Panel B. Thus, the macro branch does not show reliable return predictability from average sentiment, but it does show that greater dispersion in macro beliefs is followed by weaker broad-market returns. Table 11: Digital-Twin Macro Disagreement and Future Market Returns This table reports controlled predictive regressions of future cumulative S&P 500 returns on macro disagreement from the macro interview branch. Returns are measured in percentage points over 1, 5, and 10 trading days. The disagreement measure is Disagreement IQR (3 of 7). Each specification includes horizon-matched lagged S&P 500 returns and lagged VIX returns as controls. Panel A reports disagreement-only specifications. Panel B adds Net Sentiment (3 of 7) to the same controlled specification. Disagreement coefficients and standard errors are scaled to a 10 bull-score-point increase in the disagreement measure. Sentiment coefficients are percentage-point changes in future S&P 500 returns for a +1-unit increase in sentiment. NeweyâWest standard errors are shown beneath each estimate, with lag length hâ1h-1 for horizon h. N reports trading-day observations. The sample uses the 81-account macro interview branch and contains 53 trading days from December 15, 2025 through March 16, 2026. Indicators ***, **, * denote statistical significance at the 1%, 5%, and 10% level, respectively. (1) (2) (3) Dependent variable: Future S&P 500 return (p) 1 trading day 5 trading days 10 trading days Panel A. Disagreement with Lagged Market Controls Disagreement -0.069 -0.771â -0.864â (0.212) (0.452) (0.510) Lagged return -0.212 -0.558â -0.362 (0.197) (0.157) (0.265) Lagged VIX return -0.012 -0.116â -0.146â (0.039) (0.026) (0.039) N 53 53 53 Adj. R2R^2 -0.036 0.281 0.482 Panel B. Disagreement and Sentiment with Lagged Market Controls Disagreement 0.136 -0.761 -0.863â (0.356) (0.512) (0.444) Sentiment 1.502 0.089 0.013 (2.026) (2.786) (2.651) Lagged return -0.229 -0.559â -0.362 (0.195) (0.165) (0.310) Lagged VIX return -0.011 -0.115â -0.146â (0.037) (0.026) (0.050) N 53 53 53 Adj. R2R^2 -0.043 0.267 0.471 Figure 3 shows the macro predictability patterns visually. Panels A and C of Figure 3 sort days into quintiles of Net Sentiment. The highest-sentiment quintile beats the lowest-sentiment quintile by 140.4140.4 basis points over five days (Panel A) and 297.1297.1 basis points over ten days (Panel C). However, as shown above, this pattern is not statistically significant in a more formal setting. Panels B and D of Figure 3 sort days into quintiles of Disagreement IQR. The highest-disagreement quintile underperforms the lowest-disagreement quintile by 106.8106.8 basis points over five days (Panel B) and by 260.5260.5 basis points over ten days (Panel D), a relation that is statistically significant in the more formal setting. Figure 3: Digital-Twin Macro Sentiment, Disagreement, and Future Market Returns This figure sorts trading days into quintiles of the indicated macro belief proxy and plots average future S&P 500 returns, reported in basis points. Panels A and C use Net Sentiment (3 of 7). Panels B and D use Disagreement IQR (3 of 7). The top row reports 5-trading-day future returns, and the bottom row reports 10-trading-day future returns. The sample uses 81 accounts and 53 trading days from December 15, 2025 through March 16, 2026. Q5-Q1 is the top-minus-bottom quintile spread. Shaded bands denote 90% confidence intervals for quintile means, computed with NeweyâWest standard errors using lag length hâ1h-1 for horizon h. Taken together, these results show that the interviews in the macro branch are informative, but not because average sentiment forecasts market returns. The level of bullishness in the digital-twin panel is not reliably related to future S&P 500 returns. Instead, the stronger signal is disagreement across twins. Days with more dispersed macro beliefs are followed by weaker broad-market performance, even after controlling for average sentiment, lagged market returns, and lagged VIX returns. This pattern complements the stock-pick evidence: the interviews recover useful belief proxies not only about which stocks are relatively favored, but also about when market-level views are more contested. In this sample, belief dispersion, rather than average optimism, is the macro belief proxy with predictive content. The results in this section are consistent with the broader asset-pricing evidence that market timing is difficult, even when cross-sectional return predictability is present. Prior work documents predictable cross-sectional return patterns, including size, value, and momentum (Fama and French, 1992; Jegadeesh and Titman, 1993), while the market-timing literature emphasizes the difficulty of reliably forecasting broad market movements (Goyal and Welch, 2008; Goyal et al., 2024). The stronger contribution of our paper remains the stock-level measurement result: the digital-twin interviews recover economically meaningful variation across stocks, especially in the silent region where public recommendations are absent. 7. Discussion 7.A. Applications Beyond Financial Social Media The selective-disclosure problem is not unique to financial influencers on social media. Many social-science settings involve agents who choose what to reveal, when to reveal it, how strongly to express it, and when to remain silent. The design in this paper offers a way to study that problem. Researchers can build digital twins of human agents from public statements, ask fixed questions in real time, validate the output where public statements provide a benchmark, and then conduct experiments using the added coverage from the silent region. The goal is not to replace public statements but to measure beliefs alongside them. The main benefit of this research design is that the resulting panel includes both disclosed views and silent-region belief proxies, so it can be used to study a broader range of views. In the finance literature, sell-side analysts are one natural extension of our research. These analysts disclose ratings and price targets, and their public personas can be observed through report language, conference-call questions, and public commentary. Prior work shows that analyst recommendations and revisions can affect prices and predict subsequent stock returns (Womack, 1996; Barber et al., 2001; Jegadeesh and Kim, 2006; Sorescu and Subrahmanyam, 2006). Still, analysts choose which firms to emphasize, which risks to highlight, and when to update a recommendation. A digital-twin design could help researchers measure the views implied by an analystâs public-report persona across a wider range of covered firms and risks, especially in cases where the public reports are silent. The same logic applies to executives, fund managers, financial journalists, market commentators, and financial advisers. Beyond finance, the same measurement problem appears in polling, political communication, labor-market disclosure, journalism, central banking, and other public-facing settings.464646For related work on survey nonresponse, response bias, and digital-twin polling, see Keeter et al. (2017), Kennedy and Hartig (2019), Groves and Peytcheva (2008), Brick and Williams (2013), Mercer and Lau (2023), Claassen and Ryan (2025), and Cerina and Duch (2025). For strategic issue emphasis in politics, see Petrocik (1996), Sniderman and Theriault (2004), and Riker (1996). For labor-market signaling, disclosure, and audit-study designs, see Spence (1973), Pallais (2014), Kessler et al. (2019), Kuhn and Shen (2013), and Bertrand and Mullainathan (2004). The details differ, but the common problem is that public records reveal, in part, an agentâs decision to share a particular view, not the full set of views held by that agent. A digital-twin design could help researchers compare what public agents say voluntarily with what these same agents would say when questioned through their corresponding digital twins. Then, researchers could study both what is voluntarily disclosed and what becomes visible only because of the digital twin interviews. 7.B. Scope, Limitations, and Future Research The prospective nature of our research design imposes a practical constraint on scale. Although the stock-pick panel is substantial in the cross-sectional dimensionâcontaining 21,410 stock-date observations across 429 stocksâit is generated from 81 finfluencer personas over 50 event dates, while the macro panel spans 53 event dates. Building this panel and conducting repeated daily interviews of the digital twinsâparticularly the account-by-stock interviewsârequired substantial API usage, token consumption, and associated expenditures. These resource demands constrained the duration of the data-collection period. At the same time, our main stock-level directional findings are robust across different empirical approaches: FamaâMacBeth regressions with NeweyâWest inference, Calendar-Time portfolios, and staggered-vintage regressions with non-overlapping return windows and placebo benchmarks. These complementary designs do not substitute for a longer sample, but they reduce the concern that the findings reflect one particular estimator or treatment of overlapping outcomes. Future research can extend the design across longer periods, additional personas and platforms, broader stock universes, and different market regimes. We hope that such extensions will corroborate and refine our findings while further establishing digital-twin interviews as a measurement instrument for studying selective disclosure. Importantly, digital-twin interviews are measurement instruments, not claims about hidden views. They require transparent conditioning material, fixed protocols, real-time archives, and validation benchmarks. Researchers should be especially careful when the underlying material is private, sensitive, or not clearly part of a public persona. The goal is not to infer private beliefs of human agents. It is to create comparable records in settings where public disclosure is incomplete. Used carefully, this approach can help social scientists study a common problem: important information is often unobserved not because it does not exist, but because someone chose not to disclose it. 8. Conclusion This paper starts from a simple measurement problem. Public financial posts in social media are selective disclosures. They reveal what finfluencers choose to say, not the full set of views that these finfluencers could have expressed. Silence, therefore, is ambiguous. A research design based on actual public posts can measure disclosed content, but it cannot observe the silent region where no post appears. We address that problem with repeated, real-time digital-twin interviews of finfluencers. The interviews use a fixed protocol, are timestamped before the relevant return windows, and produce public-persona belief proxies for the overall market and for a broad set of stocks. We do not claim that our method recovers the private beliefs of these finfluencers. Our claim is narrower, but important. Standardized responses from digital-twin interviews let us study both the views that appear in public posts and the belief proxies that become visible only because the interview protocol asks. The evidence supports the premise of our research design. Digital-twin interview responses align with public recommendations, lean toward later public recommendations even before they appear, and retain account-specific structure after common market conditions are removed. The silent region is large, accounting for 84.8%84.8\% of the stock-pick panel, and the stock-level interview responses carry valuable economic content. Stocks with more buy-leaning digital-twin responses earn higher future excess returns. Stocks with greater disagreement have higher future volatility, while stocks that lack attention have more stable price patterns in the future. At the aggregate level, average macro sentiment does not reliably predict broad-market returns, but disagreement in macro beliefs predicts weaker future market performance. Our main contribution is measurement under selective disclosure. The paper is not a claim that digital twins reveal private beliefs, and it is not primarily a claim that finfluencers can pick stocks. Instead, this paper shows that standardized, real-time digital-twin interviews can make the silent region of a public persona visible. In settings where public communication is selective, we show that what is not said can also be measured. References Abdelrahman et al. (2025) Abdelrahman, Mahmoud, Edgardo Macatulad, Binyu Lei, Matias Quintana, Clayton Miller, and Filip Biljecki, 2025, What is a digital twin anyway? deriving the definition for the built environment from over 15,000 scientific publications, Building and Environment 274, 112748. Adams et al. (2023) Adams, Travis, Andrea Ajello, Diego Silva, and Francisco VĂĄzquez-Grande, 2023, More than words: Twitter chatter and financial market sentiment, Technical Report FEDS 2023-034, Board of Governors of the Federal Reserve System, Accessed 2026-03-01. Andersson et al. (2019) Andersson, Charlotte, Andrew D Johnson, Emelia Benjamin, Daniel Levy, and Ramachandran Vasan, 2019, 70-year legacy of the Framingham Heart Study, Nature Reviews Cardiology 16, 687â698. Andrei and Hasler (2015) Andrei, Daniel, and Michael Hasler, 2015, Investor attention and stock market volatility, Review of Financial Studies 28, 33â72. Angelopoulos et al. (2023) Angelopoulos, Anastasios N, Stephen Bates, Clara Fannjiang, Michael I Jordan, and Tijana Zrnic, 2023, Prediction-powered inference, Science 382, 669â674. Antweiler and Frank (2004) Antweiler, Werner, and Murray Z. Frank, 2004, Is all that talk just noise? The information content of internet stock message boards, Journal of Finance 59, 1259â1294. Baker and Wurgler (2006) Baker, Malcolm, and Jeffrey Wurgler, 2006, Investor sentiment and the cross-section of stock returns, Journal of Finance 61, 1645â1680. Barber et al. (2001) Barber, Brad, Reuven Lehavy, Maureen McNichols, and Brett Trueman, 2001, Can investors profit from the prophets? Security analyst recommendations and stock returns, Journal of finance 56, 531â563. Barber et al. (2022) Barber, Brad M., Xing Huang, Terrance Odean, and Christopher Schwarz, 2022, Attention-induced trading and returns: Evidence from Robinhood users, Journal of Finance 77, 3141â3190. Barber and Odean (2008) Barber, Brad M., and Terrance Odean, 2008, All that glitters: The effect of attention and news on the buying behavior of individual and institutional investors, Review of Financial Studies 21, 785â818. Barberis et al. (1998) Barberis, Nicholas, Andrei Shleifer, and Robert Vishny, 1998, A model of investor sentiment, Journal of Financial Economics 49, 307â343. Barrot et al. (2016) Barrot, Jean-NoĂ«l, Ron Kaniel, and David Sraer, 2016, Information and the trading behavior of retail investors, Review of Financial Studies 29, 2667â2701. Berger and Milkman (2012) Berger, Jonah, and Katherine L Milkman, 2012, What makes online content viral?, Journal of marketing research 49, 192â205. Bertrand and Mullainathan (2004) Bertrand, Marianne, and Sendhil Mullainathan, 2004, Are Emily and Greg more employable than Lakisha and Jamal? A field experiment on labor market discrimination, American Economic Review 94, 991â1013. Boehmer et al. (2021) Boehmer, Ekkehart, Charles M. Jones, and Xiaoyan Zhang, 2021, Retail trading and market quality, Journal of Finance 76, 1029â1064. Bollen et al. (2011) Bollen, Johan, Huina Mao, and Xiao-Jun Zeng, 2011, Twitter mood predicts the stock market, Journal of Computational Science 2, 1â8. Bordalo et al. (2019) Bordalo, Pedro, Nicola Gennaioli, Rafael La Porta, and Andrei Shleifer, 2019, Diagnostic expectations and stock returns, Journal of Finance 74, 2839â2874. Brick and Williams (2013) Brick, J Michael, and Douglas Williams, 2013, Explaining rising nonresponse rates in cross-sectional surveys, The ANNALS of the American Academy of Political and Social Science 645, 36â59. Broska et al. (2025) Broska, David, Michael Howes, and Austin van Loon, 2025, The mixed subjects design: Treating large language models as potentially informative observations, Sociological Methods & Research 54, 1074â1109. Brown et al. (2020) Brown, Tom B., Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, Sandhini Agarwal, Ariel Herbert-Voss, Gretchen Krueger, Tom Henighan, Rewon Child, Aditya Ramesh, Daniel M. Ziegler, Jeffrey Wu, Clemens Winter, Christopher Hesse, Mark Chen, Eric Sigler, Mateusz Litwin, Scott Gray, Benjamin Chess, Jack Clark, Christopher Berner, Sam McCandlish, Alec Radford, Ilya Sutskever, and Dario Amodei, 2020, Language models are few-shot learners, Advances in neural information processing systems 33, 1877â1901. Cerina and Duch (2025) Cerina, Roberto, and Raymond Duch, 2025, The 2024 US presidential election PoSSUM poll, PS: Political Science & Politics 58, 286â297. Chen et al. (2025) Chen, Shuaiyu, T. Clifton Green, Huseyin Gulen, and Dexin Zhou, 2025, What does chatgpt make of historical stock returns? extrapolation and miscalibration in llm stock return forecasts, Purdue University Working Paper . Chen et al. (2022) Chen, Yifei, Bryan T. Kelly, and Dacheng Xiu, 2022, Expected returns and large language models, Journal of Finance forthcoming. Claassen and Ryan (2025) Claassen, Ryan L, and John Barry Ryan, 2025, Biased polls: Investigating the pressures survey respondents feel: Rl claassen, jb ryan, Acta Politica 60, 821â843. Cookson et al. (2022) Cookson, J. Anthony, Zhaoran Lu, and Marina Niessner, 2022, The gamestop short squeeze, Review of Asset Pricing Studies 12, 1â40. Cookson and Niessner (2019) Cookson, J. Anthony, and Marina Niessner, 2019, When saving is gambling, Journal of Finance 74, 1887â1939. Cookson and Niessner (2020) Cookson, J. Anthony, and Marina Niessner, 2020, Why donât we agree? Evidence from a social network of investors, Journal of Finance 75, 173â228. Da et al. (2011) Da, Zhi, Joseph Engelberg, and Pengjie Gao, 2011, In search of attention, Journal of Finance 66, 1461â1499. Daniel et al. (2023) Daniel, Kent, Alexander Klos, and Simon Rottke, 2023, The dynamics of disagreement, Review of Financial Studies 36, 2431â2467. De Bartolomeis et al. (2026) De Bartolomeis, Piersilvio, Javier Abad, Guanbo Wang, Konstantin Donhauser, Raymond Duch, Fanny Yang, and Issa Dahabreh, 2026, Efficient randomized experiments using foundation models, Advances in Neural Information Processing Systems 38, 159470â159509. De Long et al. (1990) De Long, J. Bradford, Andrei Shleifer, Lawrence H. Summers, and Robert J. Waldmann, 1990, Noise trader risk in financial markets, Journal of Political Economy 98, 703â738. De Veirman et al. (2017) De Veirman, Marijke, Veroline Cauberghe, and Liselot Hudders, 2017, Marketing through instagram influencers: the impact of number of followers and product divergence on brand attitude, International journal of advertising 36, 798â828. Didisheim et al. (2026) Didisheim, Antoine, Bryan T. Kelly, Mohammad Pourmohammadi, and Hanqing Tian, 2026, The inefficient pricing of news,, National Bureau of Economic Research working paper. Diether et al. (2002) Diether, Karl B., Christopher J. Malloy, and Anna Scherbina, 2002, Differences of opinion and the cross section of stock returns, Journal of Finance 57, 2113â2141. Dye (1985) Dye, Ronald A, 1985, Disclosure of nonproprietary information, Journal of accounting research 123â145. Eadie et al. (2026) Eadie, Ashley L, Holly Fernandez Lynch, Naomi Scheinerman, and Ravi B Parikh, 2026, The arrival of digital twins and in silico trials in drug development, Nature Medicine 1â5. Ellsberg (1961) Ellsberg, Daniel, 1961, Risk, ambiguity, and the savage axioms, Quarterly Journal of Economics 75, 643â669. Engelberg et al. (2018) Engelberg, Joseph, Arzu Ozoguz, and Shenghua Wang, 2018, The social transmission of information in financial markets, Review of Financial Studies 31, 3849â3879. Engelberg and Parsons (2011) Engelberg, Joseph E., and Christopher A. Parsons, 2011, The causal impact of media in financial markets, Journal of Finance 66, 67â97. Espeute and Preece (2024) Espeute, Serena, and Rhodri G. Preece, 2024, The finfluencer appeal: Investing in the age of social media, Technical report, CFA Institute Research and Policy Center. Fainmesser and Galeotti (2021) Fainmesser, Itay P, and Andrea Galeotti, 2021, The market for online influence, American Economic Journal: Microeconomics 13, 332â372. Fama and French (1992) Fama, Eugene F., and Kenneth R. French, 1992, The cross-section of expected stock returns, Journal of Finance 47, 427â465. Fama and French (2008) Fama, Eugene F, and Kenneth R French, 2008, Dissecting anomalies, Journal of Finance 63, 1653â1678. Fama and MacBeth (1973) Fama, Eugene F., and James D. MacBeth, 1973, Risk, return, and equilibrium: Empirical tests, Journal of Political Economy 81, 607â636. Fang and Peress (2009) Fang, Lily, and Joel Peress, 2009, Media coverage and the cross-section of stock returns, The journal of finance 64, 2023â2052. Financial Industry Regulatory Authority (2025) Financial Industry Regulatory Authority, 2025, Social media-influenced investing, FINRA report, Published December 11, 2025. Gentzkow and Shapiro (2006) Gentzkow, Matthew, and Jesse M Shapiro, 2006, Media bias and reputation, Journal of political Economy 114, 280â316. Giglio et al. (2021) Giglio, Stefano, Matteo Maggiori, Johannes Stroebel, and Stephen Utkus, 2021, Five facts about beliefs and portfolios, American Economic Review 111, 1481â1522. Glasserman and Lin (2023) Glasserman, Paul, and Caden Lin, 2023, Assessing look-ahead bias in stock return predictions generated by gpt sentiment analysis, arXiv preprint arXiv:2309.17322 . Gordillo MartĂnez (2024) Gordillo MartĂnez, Lizeth, 2024, The impact of the Social Media Sentiment Index on S&P 500 returns, The AnĂĄhuac Journal https://doi.org/10.36105/theanahuacjour.2024v24n1.08. Goyal and Welch (2008) Goyal, Amit, and Ivo Welch, 2008, A comprehensive look at the empirical performance of equity premium prediction, Review of Financial Studies 21, 1455â1508. Goyal et al. (2024) Goyal, Amit, Ivo Welch, and Athanasse Zafirov, 2024, A comprehensive 2022 look at the empirical performance of equity premium prediction, Review of Financial Studies 37, 3490â3557. Graham (1999) Graham, John R, 1999, Herding among investment newsletters: Theory and evidence, The Journal of Finance 54, 237â268. Greenwood and Shleifer (2014) Greenwood, Robin, and Andrei Shleifer, 2014, Expectations of returns and expected returns, Review of Financial Studies 27, 714â746. Grossman and Stiglitz (1980) Grossman, Sanford J., and Joseph E. Stiglitz, 1980, On the impossibility of informationally efficient markets, American Economic Review 70, 393â408. Groves and Peytcheva (2008) Groves, Robert M, and Emilia Peytcheva, 2008, The impact of nonresponse rates on nonresponse bias: a meta-analysis, Public opinion quarterly 72, 167â189. Hassan et al. (2019) Hassan, Tarek A., Stephan Hollander, Laurence van Lent, and Ahmed Tahoun, 2019, Firm-level political risk: Measurement and effects, Quarterly Journal of Economics 134, 2135â2202. Hayek (1945) Hayek, Friedrich A., 1945, The use of knowledge in society, American Economic Review 35, 519â530. He et al. (2025) He, Songrun, Linying Lv, Asaf Manela, and Jimmy Wu, 2025, Chronologically consistent large language models, arXiv preprint 2502.21206. Hoberg and Phillips (2016) Hoberg, Gerard, and Gordon Phillips, 2016, Text-based network industries and endogenous product differentiation, Journal of Political Economy 124, 1423â1465. Hong et al. (2000) Hong, Harrison, Jeffrey D Kubik, and Amit Solomon, 2000, Security analystsâ career concerns and herding of earnings forecasts, The Rand journal of economics 121â144. Hong et al. (2004) Hong, Harrison, Jeffrey D. Kubik, and Jeremy C. Stein, 2004, Social interaction and stock-market participation, Journal of Finance 59, 137â163. Hong and Stein (2007) Hong, Harrison, and Jeremy Stein, 2007, Disagreement and the stock market, Journal of Economic Perspectives 21, 109â128. Hong and Stein (1999) Hong, Harrison, and Jeremy C. Stein, 1999, A unified theory of underreaction, momentum trading, and overreaction in asset markets, Journal of Finance 54, 2143â2184. Horton et al. (2023) Horton, John J, Apostolos Filippas, and Benjamin S. Manning, 2023, Large language models as simulated economic agents: What can we learn from homo silicus?, Technical report, National Bureau of Economic Research. Iliut a et al. (2024) Iliut a, Miruna-Elena, Mihnea-Alexandru Moisescu, Eugen Pop, Anca-Daniela Ionita, Simona-Iuliana Caramihai, and Traian-Costin Mitulescu, 2024, Digital twinâa review of the evolution from concept to technology and its analytical perspectives on applications in various fields, Applied Sciences 14, 5454. Jadhav and Mirza (2025) Jadhav, Aakanksha, and Vishal Mirza, 2025, Large language models in equity markets: Applications, techniques, and insights, Frontiers in Artificial Intelligence 1608365. Jegadeesh and Kim (2006) Jegadeesh, Narasimhan, and Woojin Kim, 2006, Value of analyst recommendations: International evidence, Journal of Financial Markets 9, 274â309. Jegadeesh and Titman (1993) Jegadeesh, Narasimhan, and Sheridan Titman, 1993, Returns to buying winners and selling losers: Implications for stock market efficiency, Journal of Finance 48, 65â91. Jones et al. (2020) Jones, David, Chris Snider, Aydin Nassehi, Jason Yon, and Ben Hicks, 2020, Characterising the digital twin: A systematic literature review, CIRP journal of manufacturing science and technology 29, 36â52. Kakhbod et al. (2023) Kakhbod, Ali, Seyed Mohammad Kazempour, Dmitry Livdan, and Norman Schuerhoff, 2023, Finfluencers, Swiss Finance Institute Research Paper. Keeter et al. (2017) Keeter, Scott, Nick Hatley, Courtney Kennedy, and Arnold Lau, 2017, What low response rates mean for telephone surveys, Pew Research Center 15, 1â39. Kennedy and Hartig (2019) Kennedy, Courtney, and Hannah Hartig, 2019, Response rates in telephone surveys have resumed their decline, Pew Research Center 27. Kessler et al. (2019) Kessler, Judd B, Corinne Low, and Colin D Sullivan, 2019, Incentivized resume rating: Eliciting employer preferences without deception, American Economic Review 109, 3713â3744. Kim et al. (2024) Kim, Alex G., Maximilian Muhn, and Valeri V. Nikolaev, 2024, Financial statement analysis with large language models, Working paper. Knight (1921) Knight, Frank H., 1921, Risk, Uncertainty and Profit (Houghton Mifflin, Boston, MA). Koijen and Levy (2026) Koijen, Ralph S.J., and Bradford Lynch Levy, 2026, Assessing the benefits of optimized agentic AI systems for asset pricing, Available at SSRN 6474601. Kuhn and Shen (2013) Kuhn, Peter, and Kailing Shen, 2013, Gender discrimination in job ads: Evidence from china, Quarterly Journal of Economics 128, 287â336. Lalwani (2025) Lalwani, Vaibhav, 2025, Finfluencer recommendations, Economics Letters 112511. Lattanzi et al. (2021) Lattanzi, Luca, Roberto Raffaeli, Margherita Peruzzini, and Marcello Pellicciari, 2021, Digital twin for smart manufacturing: A review of concepts towards a practical industrial implementation, International Journal of Computer Integrated Manufacturing 34, 567â597. Lehner and Lopez-Lira (2026) Lehner, Sebastian, and Alejandro Lopez-Lira, 2026, Chatgpt as a time capsule: The limits of price discovery, arXiv preprint arXiv:2604.21433 . Lewis et al. (2020) Lewis, Patrick, Ethan Perez, Aleksandra Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich KĂŒttler, Mike Lewis, Wen tau Yih, Tim RocktĂ€schel, Sebastian Riedel, and Douwe Kiela, 2020, Retrieval-augmented generation for knowledge-intensive nlp tasks, Advances in neural information processing systems 33, 9459â9474. Li et al. (2026) Li, Weixian Waylon, Hyeonjun Kim, Mihai Cucuringu, and Tiejun Ma, 2026, Can llm-based financial investing strategies outperform the market in long run?, in Proceedings of the 32nd ACM SIGKDD Conference on Knowledge Discovery and Data Mining V.1, KDD â26 (Association for Computing Machinery, New York, NY, USA). Lopez-Lira and Tang (2024) Lopez-Lira, Alejandro, and Yuehua Tang, 2024, Can ChatGPT forecast stock price movements? Return predictability and large language models, arXiv preprint 2304.07619. Loughran and McDonald (2011) Loughran, Tim, and Bill McDonald, 2011, When is a liability not a liability? Textual analysis, dictionaries, and 10-Ks, Journal of Finance 66, 35â65. Ludwig et al. (2024) Ludwig, Jens, Sendhil Mullainathan, and Ashesh Rambachan, 2024, Large language models: An applied econometric framework, Annual Review of Economics 18. Mahdavi et al. (2025) Mahdavi, Sedigheh, Pradeep Kumar Joshi, Lina Huertas Guativa, Upmanyu Singh, et al., 2025, Integrating large language models in financial investments and market analysis: A survey, arXiv preprint 2507.01990. Mahmood et al. (2014) Mahmood, Syed S, Daniel Levy, Ramachandran S Vasan, and Thomas J Wang, 2014, The Framingham Heart Study and the epidemiology of cardiovascular disease: A historical perspective, Lancet 383, 999â1008. Mercer and Lau (2023) Mercer, Andrew, and Arnold Lau, 2023, Comparing two types of online survey samples, Pew Research Center 40â50. Miller (1977) Miller, Edward M., 1977, Risk, uncertainty, and divergence of opinion, Journal of Finance 32, 1151â1168. Mitchell and Stafford (2000) Mitchell, Mark L., and Erik Stafford, 2000, Managerial decisions and long-term stock price performance, Journal of Business 73, 287â329. Muchnik et al. (2013) Muchnik, Lev, Sinan Aral, and Sean J Taylor, 2013, Social influence bias: A randomized experiment, Science 341, 647â651. Mullainathan and Shleifer (2005) Mullainathan, Sendhil, and Andrei Shleifer, 2005, The market for news, American economic review 95, 1031â1053. Newey and West (1987) Newey, Whitney K., and Kenneth D. West, 1987, A simple, positive semi-definite, heteroskedasticity and autocorrelation consistent covariance matrix, Econometrica 55, 703â708. NĂșñez-Mora and Mendoza-Urdiales (2023) NĂșñez-Mora, JosĂ© Antonio, and RomĂĄn A. Mendoza-Urdiales, 2023, Social sentiment and impact in US equity market: An automated approach, Social Network Analysis and Mining 13. Ontario Securities Commission (2025) Ontario Securities Commission, 2025, OSC research uncovers concerns about finfluencersâ power of persuasion. Ouyang et al. (2022) Ouyang, Long, Jeff Wu, Xu Jiang, Diogo Almeida, Carroll L. Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, John Schulman, Jacob Hilton, Fraser Kelton, Luke Miller, Maddie Simens, Amanda Askell, Peter Welinder, Paul Christiano, Jan Leike, and Ryan Lowe, 2022, Training language models to follow instructions with human feedback, Advances in neural information processing systems 35, 27730â27744. Pallais (2014) Pallais, Amanda, 2014, Inefficient hiring in entry-level labor markets, American Economic Review 104, 3565â3599. Park et al. (2024) Park, Joon Sung, Carolyn Q. Zou, Aaron Shaw, Benjamin Mako Hill, Carrie Cai, Meredith Ringel Morris, Robb Willer, Percy Liang, and Michael S. Bernstein, 2024, Generative agent simulations of 1,000 people, arXiv preprint arXiv:2411.10109 52. Petrocik (1996) Petrocik, John R, 1996, Issue ownership in presidential elections, with a 1980 case study, American Journal of Political Science 825â850. Ranco et al. (2015) Ranco, Gabriele, Darko Aleksovski, Guido Caldarelli, Miha GrÄar, and Igor MozetiÄ, 2015, The effects of twitter sentiment on stock price returns, PLoS ONE 10, e0138441. Reis (2006) Reis, Ricardo, 2006, Inattentive consumers, Journal of Monetary Economics 53, 1761â1800. Riker (1996) Riker, William H., 1996, The Strategy of Rhetoric: Campaigning for the American Constitution (Yale University Press). Sarkar and Vafa (2024) Sarkar, Suproteem K., and Keyon Vafa, 2024, Lookahead bias in pretrained language models, Working Paper, University of Chicago. Scharfstein and Stein (1990) Scharfstein, David S, and Jeremy C Stein, 1990, Herd behavior and investment, The American economic review 465â479. Shleifer (2000) Shleifer, Andrei, 2000, Inefficient Markets: An Introduction to Behavioral Finance (Oxford University Press). Sims (2003) Sims, Christopher A., 2003, Implications of rational inattention, Journal of Monetary Economics 50, 665â690. Sniderman and Theriault (2004) Sniderman, Paul M., and Sean M. Theriault, 2004, The structure of political argument and the logic of issue framing, Studies in Public Opinion: Attitudes, Nonattitudes, Measurement Error, and Change 3, 133â65. Sorescu and Subrahmanyam (2006) Sorescu, Sorin, and Avanidhar Subrahmanyam, 2006, The cross section of analyst recommendations, Journal of Financial and Quantitative Analysis 41, 139â168. Spence (1973) Spence, Michael, 1973, Job market signaling, Quarterly Journal of Economics 87, 355â374. Sprenger et al. (2014) Sprenger, Timm O., Philipp G. Sandner, Andranik Tumasjan, and Isabell M. Welpe, 2014, Tweets and trades: The information content of stock microblogs, European Financial Management 20, 926â957. Sul et al. (2017) Sul, Hong Kee, Alan R Dennis, and Lingyao Yuan, 2017, Trading on Twitter: Using social media sentiment to predict stock returns, Decision Sciences 48, 454â488. Tao et al. (2018) Tao, Fei, He Zhang, Ang Liu, and Andrew YC Nee, 2018, Digital twin in industry: State-of-the-art, IEEE Transactions on industrial informatics 15, 2405â2415. Tetlock (2007) Tetlock, Paul C., 2007, Giving content to investor sentiment: The role of media in the stock market, Journal of Finance 62, 1139â1168. Tetlock et al. (2008) Tetlock, Paul C., Maytal Saar-Tsechansky, and Sofus Macskassy, 2008, More than words: Quantifying language to measure firmsâ fundamentals, Journal of Finance 63, 1437â1467. Trueman (1994) Trueman, Brett, 1994, Analyst forecasts and herding behavior, The review of financial studies 7, 97â124. U.S. Securities and Exchange Commission Investor Advisory Committee (2024) U.S. Securities and Exchange Commission Investor Advisory Committee, 2024, Recommendations regarding the protection of investors in their interactions with finfluencers. Vaswani et al. (2017) Vaswani, Ashish, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N. Gomez, Ćukasz Kaiser, and Illia Polosukhin, 2017, Attention is all you need, Advances in neural information processing systems 30, I. Verrecchia (1983) Verrecchia, Robert E, 1983, Discretionary disclosure, Journal of accounting and economics 5, 179â194. Vosoughi et al. (2018) Vosoughi, Soroush, Deb Roy, and Sinan Aral, 2018, The spread of true and false news online, science 359, 1146â1151. Wang et al. (2024) Wang, Yue, Tianfan Fu, Yinlong Xu, Zihan Ma, Hongxia Xu, Bang Du, Yingzhou Lu, Honghao Gao, Jian Wu, and Jintai Chen, 2024, Twin-gpt: digital twins for clinical trials via large language model, ACM Transactions on Multimedia Computing, Communications and Applications . Womack (1996) Womack, Kent L, 1996, Do brokerage analystsâ recommendations have investment value?, Journal of Finance 51, 137â167. Xiao et al. (2025) Xiao, Mengxi, Zihao Jiang, Lingfei Qian, Zhengyu Chen, Yueru He, Yijing Xu, Yuecheng Jiang, Dong Li, Ruey-Ling Weng, Min Peng, Jimin Huang, Sophia Ananiadou, and Qianqian Xie, 2025, Retrieval-augmented large language models for financial time series forecasting, arXiv preprint 2502.05878. Yang et al. (2026) Yang, Yuzhe, Yifei Zhang, Minghao Wu, Kaidi Zhang, Yunmiao Zhang, Honghai Yu, Yan Hu, and Benyou Wang, 2026, Twinmarket: A scalable behavioral and social simulation for financial markets, in The Thirty-ninth Annual Conference on Neural Information Processing Systems. Yao et al. (2023) Yao, Jun-Feng, Yong Yang, Xue-Cheng Wang, and Xiao-Peng Zhang, 2023, Systematic review of digital twin technology and applications, Visual Computing for Industry, Biomedicine, and Art 6, 10. Yuan et al. (2024) Yuan, Hang, Saizhuo Wang, and Jian Guo, 2024, Alpha-gpt 2.0: Human-in-the-loop ai for quantitative investment, arXiv preprint arXiv:2402.09746 . Internet Appendix âTalking to Digital Twins: Selective Disclosure and Belief Measurement in Financial Social Mediaâ A. Data Appendix This section of the Internet Appendix describes the data-generating process outlined in Section 3. It covers the selection of the finfluencer panel, the daily digital-twin interviews, both the stock-pick and macro interview branches, the public-post comparison panel, the timing rules, and the variable dictionary used in the empirical tests. A.1. Finfluencer Panel Selection This project is designed to study public personas on X that post finance-related content. We do not attempt to construct a census of all finance accounts on X. Instead, we construct a fixed panel of accounts that regularly discuss markets and stocks, have a meaningful public audience, and provide enough observable content to support repeated digital-twin interviews. The panel was fixed before the interviews began, so interview responses could not affect which accounts entered the sample. Our initial approach to building the panel was to identify finfluencers algorithmically, by combining a finance-specific word list with a list of exchange-listed ticker symbols (cashtags). An account was flagged as a candidate finfluencer when its recent posts contained a sufficiently high density of finance vocabularyâdrawn from the Loughran and McDonald (2011) financial dictionariesâtogether with frequent references to tickers. The appeal of this approach was scale and objectivity: in principle it could screen a very large number of accounts with little manual judgment and could be reproduced by other researchers. In practice, however, the lexicon-plus-ticker screen produced too many false positives to support a credible persona panel. Two sources of noise were especially severe. First, ticker symbols frequently coincide with common English words and abbreviations, so ordinary posts were misclassified as stock commentary. Cashtags such as $IT, $ALL, $ON, $A, $CAR, $NOW, $REAL, and $GOOD match everyday language at least as often as they identify the underlying equities (Gartner, Allstate, ON Semiconductor, Agilent, Avis, ServiceNow, The RealReal, and Gladstone Commercial, respectively), so casual posts were repeatedly flagged as ticker references. Second, finance vocabulary is shared with many non-investing domains: words such as âshare,â âinterest,â âcredit,â âbond,â âsecurity,â âmargin,â âstock,â and âoptionsâ appear routinely in posts about retail inventory, personal relationships, online gaming, sports betting, and promotional giveaways. As a result, the automated screen swept in news bots and headline aggregators, corporate and investor-relations handles, cryptocurrency and giveaway-promotion accounts, and general-interest accounts that merely used finance-sounding languageânone of which represent the kind of market-commentary persona our design requires. Because the method selected accounts on the basis of token co-occurrence rather than on whether an account maintains a recognizable, finance-oriented public persona, its precision was low, and the flagged set required so much manual pruning that the automation offered little practical advantage. We therefore pivoted to curated, human-compiled lists of finance-focused accounts. The sample was assembled in three stages. First, we compiled candidate accounts from third-party online directories and public lists, including The CFO Club, Investopedia, Investing.io, and Yahoo Finance.474747The source pages were consulted between April 2025 and June 2025. The full list, with hyperlinks, is as follows: The CFO Club; Investopedia; Feedspot, Investing X Influencers; Feedspot, Trading X Influencers; LendInvest; Agility PR; Investing.io; Traders Union; Bullish Bears; Asktraders; Securities.io; OptionsHawk; Real Trading; Yahoo Finance; and r/Trading. The consulted sources span trading generalists, long-horizon investors, wealth managers, technical analysts, day traders, market-sentiment commentators, options and derivatives traders, self-identified bullish and bearish investors, and chart-based technical analysts. In the second stage, we manually screened each candidate account and its post history from December 2024 through June 2025 against a topical inclusion rule. To be retained, an account had to post primarily about tradable financial instrumentsâindividual stocks, equity indices, options, futures, or other derivativesâwith content that took the form of recommendations, directional calls, or technical analysis of specific tickers. We excluded accounts whose principal content was real estate, personal finance advice, pure macroeconomic commentary without tradable recommendations, generic financial-news aggregation, aggressive ticker promotion, paid-signal advertising, course or lecture sales, mentoring offers, or similar commercial content. In the third stage, we applied an additional manual screen to the remaining candidates. Accounts were retained if they had at least 5,0005,000 followers, a sustained posting history of at least 1010 posts since profile creation so there is sufficient information to support repeated digital-twin interviews, and language and topical content consistent with a financial persona rather than a general social-media personality. The final panel contains 8181 finfluencer accounts on X that were observed in both interview branches of our study. The panel was finalized in June 2025, before any interviews were conducted.484848The main sample contains the 8181 X accounts observed in both interview branches. The macro interview branch contains one additional account, for a total of 8282 accounts, while the stock-pick branch contains the 8181 accounts used in the main analysis. Because the stock-level analysis requires accounts to appear in the stock-pick branch, and to keep the sample consistent across the paper, the main analysis uses the 8181 accounts observed in both branches. All tests that can be estimated with all 8282 accounts are quantitatively similar when the 82nd account is retained. The X handles of the 8181 accounts used in our study are listed below. These are the public personas that supply the account-conditioned respondents (digital twins) for the macro and stock-pick interview panels. 1OptionsTrading abnormalreturns AdeptMarket anymantrading AswathDamodaran BezosCrypto Blackopstocks BreakoutStocks BrianStutland ChandlerTrading Chariot_Invest ContrarianShort CrossStocks Darkminer71 DataDInvesting davidmoadel daytradesignals DCDOWORK DevotedDividend dirtcheapstocks Dividend_Dr DKellerCMT dmdsplyinvestor DV_Situations ElliottForecast GlobalStockPick gurgavin hiddensmallcaps I_Am_The_ICT ideahive InvestorsLive ivanhoff2 Jake__Wujastyk LindaRaschke LizAnnSonders MarketMovesMatt MasteredTrader MichaelGoodwell OMillionaires OnlyOTrades OptionAlpha optionscjp OptionsDepth OptionsHawk OptiontradinIQ profitly RealDayTrading ReturnsJourney SeekingAlpha Selling4Premium sentimentrader seth_fin SethCL StockMKTNewz StockOptionCole StocksOnSpaces StocksToTrade Stocktwits SuburbanDrone SwaggyStocks TAftermath2020 TheAlphaThought TheStalwart TheTradingTank timothysykes traderstewie TradesTrey tradewithprof Trading0secrets TradingwithFun1 TrendSpider TSXtrad3r ukarlewitz value_invest12 ValueStockGeek ValueWolf VertiCallAlgo VROStocks vsourbh ZacksResearch zerohedge A.2. Daily Digital-Twin Interviews For each X account in the panel, we create an account-conditioned digital twin. The digital twin is an LLM respondent conditioned on the accountâs public persona.494949We retain the model identifier for every digital-twin interview response. In the retained analysis samples (after timing and deduplication), nearly all responses were generated with gpt-5-mini-2025-08-07; gpt-5-nano-2025-08-07 was used only during a brief interval, corresponding to eight event dates in each interview branch. Model choice reflected API availability and cost considerations, not any review or analysis of the interview responses. All queries were issued programmatically through OpenAIâs API rather than the consumer web (ChatGPT) interface, allowing us to automate and reproduce the prompts at scale. API access is a paid service, and the volume of queries required for this study entailed substantial usage charges to OpenAI. The conditioning material includes profile metadata and recent posts. As the account produces new public content and as market conditions change, the conditioning material is updated. The interview questions remain fixed for the core measures. This design gives us a repeated panel of public-persona belief proxies. Before launching the main interview protocol, we conducted two daily pilot studies, from August 1, 2025 through August 31, 2025, and again from September 29, 2025 to December 14, 2025. During these pilot periods, we queried the digital twins about aggregate market conditions and about content appearing in the finfluencersâ contemporaneous public posts. The pilots were designed to evaluate feasibility rather than to produce the main empirical results. They allowed us to estimate the cost of repeated real-time interviews, identify operational constraints, and refine the workflow used in the main experiment. We began our main experiment on December 15, 2025. The interviews were conducted almost daily through March 14, 2026.505050A few days are missing from this period because of ordinary operational issues such as backlogs. The alignment rules described below in Subsection A.4 explain how we handled these irregularities. The interview protocol has two layers of instructions. The first level gives the model the account profile, recent posts, and permission to retrieve clearly relevant market context through web search. When using web search, the prompt tells the model to prioritize reputable finance-focused sources and to treat social-media or forum content mainly as sentiment. The second level of the prompt then asks for one structured answer to each question. Three excerpts from the protocol capture these instructions: âSelect the most likely response based strictly on the provided profile data. The chosen response must be the most accurate representation of the profile.â âYou may also use web search to retrieve relevant, up-to-date information about the specific tickers, sectors, asset classes, strategies, and macro themes that this profile is likely to care about.â âYou must give an answer for every question while maintaining the persona and perspective of the financial influencer profile provided.â Together, these instructions bound the interview responses. The model can use outside market context when it is clearly relevant, but the selected response must still be supported by the account material or by clearly relevant retrieved information. If the sources are uncertain or conflicting, the model is told to acknowledge that uncertainty. Within the macro branch, questions are structured with forced responses: each response includes a substantive answer, a selected symbol/category, an explanation, and a speculation score. The speculation score is not a confidence score about the future state of the market. It records how much useful account information supports the selected response.515151The prompt defines speculation on a 0 to 100100 scale: 0â2020 is low speculation, 2121â4040 is moderate-low, 4141â6060 is moderate, 6161â8080 is moderate-high, and 8181â100100 is high. This scale matters in the analysis because the headline macro signal treats answers with speculation scores of 4040 or below as sufficiently grounded in the accountâs public persona to contribute to strong bullish or bearish classifications. The prompt also asks the model to preserve a structured response format so the answers can be analyzed consistently. The first macro question illustrates the structure: Question: What is your best estimate of the probability that the U.S. economy will enter a recession in the next 12 months? A1) 0 to 20% (Very unlikely) A2) 21 to 40% (Unlikely) A3) 41 to 60% (About as likely as not) A4) 61 to 80% (Likely) A5) 81 to 100% (Very likely) Required response fields: question: [Question text] explanation: [Features of the supplied information that support the selected response and speculation score] symbol: [Selected symbol, such as A1] category: [Full text of selected response category] speculation: [Speculation score from 0 to 100] The raw files produced from every daily interview retain the explanation and category fields. In our macro analysis, we use the selected symbol and speculation score for the seven core questions listed below. We then translate the selected symbols into question-specific bull scores, oriented so that higher values are more bullish. For example, on the recession question above, an answer of A4, corresponding to a 6161 to 80%80\% recession probability, is translated into a bull score of 3030. 1. What is your best estimate of the probability that the U.S. economy will enter a recession in the next 12 months? 2. Considering business conditions in the country as a whole, do you think that during the next 12 months we will have good times financially, bad times, or something in between? 3. In the next six months, do you expect business conditions to be better, worse, or the same? 4. In your view, is overall investor sentiment currently bearish, neutral, or bullish? 5. Do you expect the direction of the stock market over the next six months to be up, unchanged, or down? 6. In the next six months, do you expect U.S. stock market indices to rise, stay about the same, or fall? 7. In the next 12 months, do you expect interest rates on U.S. bonds to rise, stay about the same, or fall? Within the stock-pick branch we used the same account-conditioning protocol as we did in the macro branch, but asked a much larger set of questions. Instead of asking each digital twin for a broad market view, we asked them for a separate view on each stock in the covered universe. As a result, this interview branch was significantly more expensive to run.525252We did not begin the Stock-Pick branch until after the conclusion of the two pilot studies because we required additional time to define the scope of our research (studying the finfluencersâ silent region), obtain cost estimates for accessing OpenAIâs API at a much larger scale, and identify funding sources. After aligning trading days to interview timestamps, the cleaned stock-pick files run from December 15, 2025 through March 16, 2026.535353The last interview was on March 14, 2026, corresponding to the trading day on March 16, 2026. The first stock-pick runs were intentionally broad: we asked each digital twin for reflections on every ticker in the Russell 4,0004,000. That breadth was useful for learning what the process could support, but it was too costly to run as a daily panel. We therefore shifted the interviews to a narrower set of 429429 large-cap stocks that are a subset of the Russell 4,0004,000 index. In all of the analyses in the paper, we use this smaller, recurring universe of stocks rather than the broad Russell 4,0004,000 universe we interviewed for a small number of days. The resulting stock-pick sample contains 429429 recurring tickers, 21,41021,410 stock-by-trading-day observations, and 5050 trading days. The complete list of the 429 tickers follows. A AAPL ABBV ABNB ABT ADBE ADI ADM ADP ADSK AEE AEP AES AFL AIG AIZ AJG AKAM ALB ALGN ALL AMAT AMD AME AMGN AMP AMZN ANET AOS APA APD APH APO APP ATO AVGO AVY AWK AXON AXP AZO BA BAC BALL BAX BBY BDX BEN BIIB BK BKNG BKR BLDR BLK BMY BR BRO BSX BX C CAG CAH CARR CAT CBRE CDNS CDW CEG CF CFG CHD CHRW CHTR CI CINF CL CLX CMCSA CME CMG CMI CMS CNC CNP COF COIN COO COP COST CPB CPRT CRL CRM CRWD CSCO CSGP CSX CTAS CTRA CTSH CTVA CVS CVX D DAL DASH D DDOG DE DECK DELL DG DGX DHI DHR DIS DLTR DOV DOW DPZ DRI DTE DUK DVA DVN DXCM EA EBAY ECL ED EFX EIX EL ELV EME EMR EOG EPAM EQT ERIE ES ETR EVRG EW EXC EXPD EXPE F FANG FAST FCX FDS FDX FE FFIV FICO FIS FISV FITB FOXA FSLR FTNT FTV GD GDDY GE GEHC GEN GILD GIS GL GLW GM GNRC GOOG GOOGL GPC GPN GS GWW HAL HAS HBAN HCA HD HIG HII HLT HOLX HON HOOD HPE HPQ HRL HSIC HSY HUBB HUM HWM IBKR IBM ICE IDXX IEX IFF INCY INTC INTU IP IQV IR ISRG IT ITW J JBHT JBL JKHY JNJ JPM K KDP KEY KEYS KHC KKR KLAC KMB KMI KO KR KVUE L LDOS LEN LH LHX LII LKQ LLY LMT LNT LOW LRCX LULU LUV LVS LW LYV MA MAR MAS MCD MCHP MCK MCO MDLZ MET META MGM MHK MKC MLM MMC M MNST MO MOH MOS MPC MPWR MRK MRNA MS MSCI MSFT MSI MTB MTCH MTD MU NDAQ NDSN NEE NEM NFLX NI NKE NOC NOW NRG NSC NTAP NTRS NUE NVDA NVR NWS NWSA ODFL OKE OMC ON ORCL ORLY OTIS OXY PANW PAYC PAYX PCAR PCG PEG PEP PFE PFG PG PGR PH PHM PKG PLTR PM PNC PNW PODD POOL PPG PPL PRU PSX PTC PWR PYPL QCOM REGN RF RJF RL RMD ROK ROL ROP ROST RSG RTX SBUX SCHW SHW SJM SMCI SNA SNPS SO SPGI SRE STLD STT STZ SWK SWKS SYF SYK SYY T TAP TDG TDY TECH TER TFC TGT TJX TMO TMUS TPL TPR TRGP TRMB TROW TRV TSCO TSLA TSN TTD TTWO TXN TXT TYL UAL UBER UHS ULTA UNH UNP UPS URI USB V VLO VLTO VMC VRSK VRSN VRTX VST VTRS VZ WAB WAT WBD WDAY WDC WEC WFC WM WMB WMT WRB WSM WST WYNN XEL XOM XYL YUM ZBH ZBRA ZTS The stock-pick prompt includes stock-by-stock questions. A representative stock-level question and response template is: Question: Based on your general knowledge of market conditions, indicate whether you would BUY, HOLD, or SELL (SHORT SELL) the stock ticker: [COMPANY NAME (TICKER)]. Required response fields: question: [COMPANY NAME (TICKER)] explanation: [Detailed explanation for recommendation] recommendation: [Value selected] confidence: [Value selected] speculation: [Speculation score selected] expected holding period: [Number of days selected] primary catalyst type: [Earnings/Macro/Product/Regulatory/Flow-Technical/Other] The prompt defines the recommendation field on a 0â100100 scale, where 0 is a very strong sell recommendation and 100100 is a very strong buy recommendation. It describes 4040â6060 as hold, 6060â8080 as moderate buy, and 8080 or above as strong buy; values below 4040 are sell-leaning. Confidence is also reported on a 0â100100 scale. Speculation follows the same interpretation as in the macro interview: higher values mean the supplied account material contains less useful information for the stock-specific answer. The raw daily stock-pick files are extremely wide because each account-day contains fields for many tickers. We reshape those files into a long panel with one row per account, interview, and ticker. The compact cleaned panel keeps the recommendation and speculation fields; the long explanations remain in the raw files and are not used in the main stock-pick analysis. When the model supplies a text recommendation instead of a number, we map SELL or SHORT to 2525, HOLD to 5050, and BUY to 7575 before any stock-date measures are built. For each stock and event date, we then aggregate across the interviewed digital twins. The basic direction measure is the average recommendation score centered at 5050. A recommendation is treated as meaningful only if it is at least 1515 points away from neutral, so scores of 6565 or above are meaningful buys and scores of 3535 or below are meaningful sells. Informative Polarization uses the subset of meaningful recommendations with speculation scores of 4040 or below, and we require at least three such recommendations before reporting it. Uncertainty Score is high when a stock-event has little directional conviction: for example, when many responses are near neutral, speculative, or both. A.3. Public-Post Comparison Panel We also construct a public-post comparison panel. This panel does not include digital twins but rather the 8181 human finfluencers who make actual public posts. We identify posts that mention specific stocks and classify the direction of the disclosed recommendation. The panel gives us an observable benchmark for digital twin interview data. When a human finfluencer publicly recommends a ticker on a given trading date, we can compare that recommendation with the digital-twin interview response for the same finfluencer, ticker, and date. When no matching public stock-recommendation post exists for a stock-event, the digital twin interview panel measures the silent region: the part of the belief space that cannot be observed from public recommendations alone. We use the public-post panel for validation and for the paperâs selective-disclosure comparisons. A.4. Timing Alignment and Sample Filters The purpose of our empirical tests is to ask whether information contained in the digital-twin interviews predicts later market outcomes. We therefore impose a strict, conservative timing rule to align interviews in market-time. As such, the return window used to evaluate interview responses begins only after the interview has been assigned to an effective trading date. This prevents an interview from being evaluated using returns that were already realized, or partly realized, when the interview response was generated. We align interviews to market time using a 3:30 p.m. U.S. Eastern Time (ET) cutoff; a notably conservative cutoff considering that trading continues until 4:00 p.m. each day. If an interview occurs on a trading day before 3:30 p.m. ET, we keep its calendar date and map that date to the corresponding trading date (e.g., an interview completed on a Tuesday before 3:30 p.m. ET is mapped to that day, Tuesday). If an interview occurs on a trading day at or after 3:30 p.m. ET, we first shift it forward by one calendar day and then map the resulting date to the next trading date (e.g., an interview completed on a Tuesday after 3:30 p.m. ET is mapped to the next day, Wednesday). Interviews that occur on non-trading days are mapped to the next trading date. The mapped trading date is the event date used in the analysis.545454We use 3:30 p.m. ET, rather than the market close, as a conservative buffer against very late interviews being paired with same-day closing prices or end-of-day information. Relative to a 4:00 p.m. cutoff, this choice has little effect. In the macro branch, 3838 of 7,1077,107 interview records before de-duplication map differently. The final daily macro dataset has the same 5353 event dates under either cutoff, and only one daily signal row changes. On that date, January 23, 2026, continuous sentiment changes from 0.3270.327 to 0.3230.323. In the stock-pick branch, 9090 of 7,4317,431 interview runs map differently before de-duplication. After keeping one run per account-event date, the only substantive difference is that a 4:00 p.m. cutoff would add one March 5, 2026 stock-pick event-date cross-section based on only nine interviews completed between 3:57 p.m. and 3:59 p.m. ET. Existing stock-event signal rows are unchanged across the two timing conventions. Returns are then measured over close-to-close trading-day windows following the event date. If an interview is assigned to event date t, the one-day return is measured from the close of trading on t to the close of trading on the next S&P 500 trading day. More generally, the h-day return is the cumulative close-to-close return over the next h trading days. Thus, an interview conducted at 2:00 p.m. ET on a Tuesday is assigned to Tuesday, and its one-day return is measured from Tuesdayâs close to Wednesdayâs close. An interview conducted at 3:45 p.m. ET on that same Tuesday is shifted forward, assigned to Wednesday, and evaluated using returns beginning after Wednesdayâs close. A Sunday interview is assigned to the next trading day, typically Monday, and its one-day return is measured from Mondayâs close to Tuesdayâs close. When more than one interview response is available for the same account and event date, we retain only the latest response that maps to that event date. In the stock-pick branch, this de-duplication is applied at the account-event-date level before forming stock-level recommendation measures, so a single account does not receive multiple weights on the same event date. In the macro branch, the same logic retains one account-level macro response per event date before forming daily macro belief measures.555555Here, âlatestâ means the latest available run after sorting by the source run date and the observed interview timestamp, not the latest intended target date. A small number of interview runs were completed in catch-up batches because computational-resource and processing constraints prevented every scheduled run from being executed on its target calendar night. We apply the event-date rule to the observed interview timestamp. When multiple runs for the same account map to the same event date, the latest observed run is retained. Figure A1 summarizes the timing of the interviews that enter the analysis after applying the timing rule and account-date de-duplication. The figure shows that very few interviews occur near the market close. In the macro branch, only 0.8%0.8\% of interviews occur between 3:00 p.m. and 4:00 p.m. ET on trading days. In the stock-pick branch, 2.3%2.3\% occur between 3:00 p.m. and 4:00 p.m. The broader timing pattern reflects the sequencing of the interview pipeline. Macro interviews are typically launched after the market close and completed overnight, so most macro interviews occur after close or before the next open. The stock-pick branch is run after the macro branch, so fewer stock-pick interviews complete during the evening; instead, many complete before the open or during regular trading hours on the next trading day. Figure A1: Timing of Digital-Twin Interviews The figure reports the timing distribution for macro and stock-pick interviews during the shared sample period. Non-trading day covers interviews occurring between 00:00 and 23:59 ET on weekends or S&P 500 market holidays. The remaining categories use actual ET clock time on S&P 500 trading days: pre-open is before 9:30am, regular hours are 9:30amâ3:00pm, the late-market bucket is 3:00pmâ4:00pm, and after close is 4:00pm or later. We apply the same market-time convention to public recommendations, using the public-post timestamp. A public recommendation posted at or before 3:30 p.m. ET keeps its calendar date before being mapped to a trading date. A public recommendation posted after 3:30 p.m. ET is first shifted forward by one calendar day and then mapped to the next S&P 500 trading date. If multiple public recommendations by the same account about the same stock map to the same public event date, we collapse them to one account-stock-event-date public signal. The collapsed observation averages the public recommendation scores and related post-level measures, while retaining the number of posts and the first and last mention times. This keeps the interview and public-post samples aligned under the same market-time convention and the same account-stock-event-date unit of observation. The remaining filters are simple sample-construction choices. First, the paper uses the 8181 accounts observed in both the macro and stock-pick branches. Second, an account-day observation in the macro branch must contain enough usable core-question information to support the daily belief measures. In practice, it must have at least four observed speculation scores among the seven core questions, and a macro trading day must have at least five contributing finfluencers. Third, the stock-pick analysis uses the recurring 429-ticker universe described above. A.5. Variable Definitions The tables and tests in the paper use variables built from the digital-twin interviews and public-post data. The variable dictionary below records the construction, unit, and observation unit for the main variables used in the paper, including the macro sentiment and disagreement measures, stock-pick direction and uncertainty measures, public-post comparison variables, and market return and volatility outcomes.565656An event date is the trading date assigned to an interview response or public post after applying the market-time alignment rule. It may differ from the calendar date on which the interview or post occurred. Throughout the paper, event date refers to this aligned trading date. A stock-event is a stock-by-event-date observation. Data Appendix Variable Definitions Variable Construction and interpretation Unit Sample unit Stock-pick interview panel Recommendation score Parsed 0â100 stock-pick recommendation score for an account-stock interview response. The neutral value is 50. Higher values are more buy-leaning and lower values are more sell-leaning. When a response is available only as text, we map sell or short to 25, hold to 50, and buy to 75 before aggregation. Used to construct Tilt, Net Buy Share, Meaningful Tilt, Conviction-Weighted Meaningful Tilt, Informative Polarization, Uncertainty Score, and the public-overlap validation measures. Score points Account Ă stock Ă event date Stock-pick speculation score Parsed 0â100 speculation score for a stock-pick response. Higher values mean the stock-level answer is less grounded in the supplied account material. A meaningful recommendation is informative when this score is observed and at most 40. Used to construct Informative Polarization, Uncertainty Score, and the low-speculation validation filters. Score points Account Ă stock Ă event date Recommendation Count Number of parsed account-stock recommendations contributing to a stock-event observation. This count is reported in the summary statistics and is the denominator for Meaningful Recommendation Share, Net Buy Share, and Uncertainty Score. Count Stock Ă event date Meaningful Recommendation Share Share of parsed recommendations with |rec_scoreâ50|â„15|rec\_score-50|â„ 15 for the stock-event. This records how much of the recommendation set moves materially away from neutral and is used to construct Conviction-Weighted Meaningful Tilt. Share in [0,1][0,1] Stock Ă event date Tilt Average recommendation score for the stock-event, centered at 50: rec_scoreÂŻiâtâ50 rec\_score_it-50. Positive values mean the recommendation set leans buy-side. Rec. points Stock Ă event date Net Buy Share Meaningful buy share minus meaningful sell share for the stock-event. Meaningful buys have recommendation scores at least 65, and meaningful sells have scores at most 35. Both shares are divided by the total number of parsed recommendations for the stock-event, not only by meaningful recommendations. Net share in [â1,1][-1,1] Stock Ă event date Meaningful Tilt Average recommendation score centered at 50, using only recommendations with |rec_scoreâ50|â„15|rec\_score-50|â„ 15. This removes the neutral center of the raw recommendation distribution. Used to construct Conviction-Weighted Meaningful Tilt. Rec. points Stock Ă event date Conviction-Weighted Meaningful Tilt Meaningful Tilt multiplied by Meaningful Recommendation Share. The measure is defined when the stock-event has at least one meaningful recommendation. High values require both directional strength and many recommendations away from neutral. Tables sometimes shorten the label to Conviction-Weighted Tilt. Index Stock Ă event date Informative Polarization 4Ăbuy shareiâtiânâfĂsell shareiâtiânâf4Ăbuy share^inf_itĂsell share^inf_it for the stock-event, where the buy and sell shares are measured within the informative subset. That subset requires both meaningful direction and a speculation score at most 40. The factor of 4 rescales the index so it reaches one when informative buys and sells are evenly split. The measure is zero when informative recommendations are all on one side and rises toward one when informative recommendations are sharply split. The paper requires at least three informative recommendations. Index in [0,1][0,1] Stock Ă event date Uncertainty Score Stock-event index that is low when many responses give clear, grounded buy or sell views and high when responses are mostly neutral, weak, or speculative. Each meaningful recommendation receives a grounding weight of 1âspeculation/1001-speculation/100; non-meaningful recommendations and meaningful recommendations without a parsed speculation score receive weight zero. The score is one minus the average grounding weight across all parsed recommendations. Index in [0,1][0,1] Stock Ă event date Market capitalization Market capitalization for stock i, matched to the event date. Used for value-weighted stock-pick estimates and for the public-attention firm-characteristic comparison, where it is reported in billions of dollars. Dollars Stock Ă event date or firm log(Market Cap) Natural logarithm of event-date market capitalization for stock i. Reported in the summary statistics and used as a firm-size characteristic in the public-attention comparison. Log dollars Stock Ă event date or firm Excess Return (h days) Cumulative excess stock return relative to the S&P 500 over the next hâ1,5,10hâ\1,5,10\ trading days, excluding the event date. The measure sums daily stock-minus-S&P 500 log returns over the window and converts the cumulative excess log return to simple-return form. Percentage points Stock Ă event date Excess Realized Volatility (h days) Square root of the sum of squared daily stock-minus-S&P 500 log returns over the next hâ1,5,10hâ\1,5,10\ trading days, excluding the event date. The reported measure is multiplied by 100. Percentage points Stock Ă event date Macro interview panel Bull score A 0â100 coding of each macro answer, with higher values always meaning a more bullish market view and 50 meaning neutral. We map the ordered response categories to scores separately for each question. When a category gives a range, we use its midpoint before orienting the score. For example, a 61â80% recession-probability answer is coded as 30 because high recession risk is bearish for stocks. Used to construct Net Sentiment, Continuous Sentiment, Disagreement IQR, and the macro validation diagnostics. Score points Account Ă question Ă day Speculation score A 0â100 score recorded with each macro answer. Higher values mean the answer is less supported by the account material supplied to the digital twin. The low-speculation screen used for Net Sentiment and Continuous Sentiment keeps answers with scores at most 40; High-Speculation Share counts answers with scores at least 60. The availability of these scores is also used to retain macro account-days. Score points Account Ă question Ă day Net Sentiment (3 of 7) Daily bullish share minus daily bearish share across retained account-days, using Q1âQ7. An account-day is bullish if at least three answers have Bull score at least 70 and Speculation score at most 40. It is bearish if at least three answers have Bull score at most 30 and Speculation score at most 40. Mixed account-days stay in the denominator but are not counted as bullish or bearish. Net share in [â1,1][-1,1] Trading day Net Sentiment (3 of 6) Same construction as Net Sentiment (3 of 7), but using Q1âQ6 only. This six-question version drops Q7, the question about the expected direction of interest rates on U.S. bonds over the next 12 months. Net share in [â1,1][-1,1] Trading day Continuous Sentiment Daily average of account-level normalized Bull scores across Q1âQ7. Each answer with an observed Bull score and Speculation score at most 40 is transformed to (Bull scoreâ50)/50(Bull score-50)/50, so the scale runs from bearish to bullish. The account-day score requires at least three valid core questions, and the daily variable averages those account-day scores. Some regressions also include the square of this variable to allow a curved sentiment-return relation. Index in [â1,1][-1,1] Trading day Disagreement IQR (3 of 7) Interquartile range, across retained accounts on a trading day, of account-level composite Bull scores. The account-level composite averages available Q1âQ7 Bull scores for that account-day. This spread is computed from Bull-score levels, not from the bullish and bearish classification flags used in Net Sentiment. Higher values mean macro views are more dispersed across accounts. Bull-score points Trading day High-Speculation Share, Q1âQ7 Daily average, across retained account-days, of the share of Q1âQ7 Speculation scores that are at least 60. This variable is reported in the summary statistics. Share in [0,1][0,1] Trading day Contributing Influencers per Day Number of distinct panel accounts retained on a trading day after the macro interview is aligned to the trading calendar, de-duplicated, and required to have at least four observed speculation scores among Q1âQ7. This count can be below 81 because some account-days are missing or filtered. Count Trading day S&P 500 Return (h days) Cumulative S&P 500 simple return over horizon h, built by summing daily S&P 500 log returns and converting the sum back to a simple return. Forward returns use the next h trading days after the aligned interview date and exclude the event day. Lagged returns use the previous h trading days. The main horizons are hâ1,5,10hâ\1,5,10\. Percentage points Trading day S&P 500 Realized Volatility (h days) Square root of the sum of squared S&P 500 daily log returns over horizon h. The macro realized-volatility controls use lagged windows. For the one-day horizon, the control is the absolute value of the previous dayâs S&P 500 log return; for the 5- and 10-day horizons, it uses the prior h trading days. The reported measure is multiplied by 100. Percentage points Trading day VIX Return (h days) Cumulative VIX simple return over horizon h, built by summing daily VIX log returns and converting the sum back to a simple return. The macro return regressions use horizon-matched lagged VIX returns for hâ1,5,10hâ\1,5,10\ as the VIX-based control. Percentage points Trading day Public-post comparison variables Public Net Buy Share Public-post version of Net Buy Share. Public stock-recommendation posts are assigned to an effective trading date from the mention timestamp, using the 3:30 p.m. Eastern Time cutoff. When an account has multiple posts about the same stock on the same effective date, those posts are collapsed to one account-stock-event-date public signal by averaging the public recommendation scores and related post-level measures. Public Net Buy Share is meaningful public buys minus meaningful public sells, divided by total account-level public recommendations for the stock-event. Net share in [â1,1][-1,1] Stock Ă event date Digital-Twin Interview Net Buy Share Net Buy Share from the stock-pick interview branch, used alongside Public Net Buy Share in the exact-overlap and no-public-post comparisons. Net share in [â1,1][-1,1] Stock Ă event date Public direction Buy or sell direction of an observed public recommendation. Public buys have recommendation scores at least 65, and public sells have scores at most 35. Neutral or conflicting public directions are excluded when a validation test needs a buy/sell benchmark. Used in the public-overlap sign-alignment, AUC, and signed-tilt validation tests. Buy/sell sign Account Ă stock Ă event date Public-post days Number of distinct effective public-post dates for a firm during the stock-pick interview window from December 15, 2025 through March 16, 2026, using public recommendations from the stock-pick interview account set. Used to form the public-attention distribution. Count Firm Public posts Number of observed public stock-recommendation posts for a firm during the stock-pick interview window, using the same effective-date public-post sample as Public-post days. Public-attention distribution rows report per-firm means within public-post-day bins. Count Firm Posts per public day Firm-level ratio of public stock-recommendation posts to public-post days, computed for firms with at least one public-post day. Ratio Firm Public attention bins Firm groups based on Public-post days during the stock-pick interview window. The distribution bins are 0, 1, 2, 3, 4, 5â9, 10â19, and 20+ public-post days; the characteristic comparison groups are low attention (0â4), middle attention (5â19), and high attention (20+). Category Firm Firm characteristics Firm-level variables used in the public-attention comparison. Table 4 reports market capitalization, log market capitalization, CRSP listing age, and selected industry indicators. Market capitalization is matched from the stock-level market data used in the return-linked stock-pick panel. Listing age and industry classifications are built from CRSP monthly data as of December 31, 2024. Continuous characteristics are winsorized at the pooled 1st and 99th percentiles. Mixed Firm A.6. Capital IQ and CRSP Data This section lists the market-data variables used to construct the return, volatility, size, and firm-characteristic variables in the paper. From S&P Capital IQ, we use adjusted close price levels, Pi,dP_i,d, for U.S. equities and selected market series, where i represents the specific stock or macro-level series, and d is the trading day.575757The exact function used is CIQ(.,IQ_CLOSEPRICE_ADJ,.). This function generates daily prices that are adjusted for dividends and stock splits. The equity file contains a Capital IQ security identifier, company name, exchange-ticker identifier, exchange, ticker label, and a panel of daily adjusted close prices. We convert these adjusted close prices to daily log returns (log denotes the natural logarithm), ri,d=logâĄ(Pi,d/Pi,dâ1),r_i,d= (P_i,d/P_i,d-1), setting returns to missing when either the current or prior adjusted close price is missing, zero, or negative. These daily log returns are used to construct future stock returns, stock excess returns relative to the S&P 500, and realized-volatility measures. We also use Capital IQ to obtain closing prices for the S&P 500 and daily values of the volatility index, VIX. The S&P 500 series provides the trading calendar, market returns, and the benchmark return used to compute stock excess returns. The VIX series provides the VIX-return controls used in the macro predictive regressions. Finally, we use stock-level market capitalization from the same market-data files to construct event-date market capitalization, log market capitalization, and value-weighted robustness tests. From CRSP585858CRSP refers to Morningstarâs Center for Research in Security Prices, formerly a product of the University of Chicago., we use monthly security data to construct the firm characteristics in Table 4. We match tickers to CRSP securities, use the first observed CRSP date to compute listing age as of December 31, 2024, and use SIC codes to form industry indicators based on the FamaâFrench 12-industry classification. The table reports listing age and selected industry shares, along with market capitalization and log market capitalization from the return-linked stock-level market data. Continuous firm characteristics are winsorized at the pooled 1st and 99th percentiles. B. Validation and Public-Disclosure Appendix In this section of the Internet Appendix we present, in Table A1, additional tests to assess the robustness of the results presented in Table 3 (Section 4) of the main paper. Recall that Table 3 shows that the digital-twin interview outputs look like account-conditioned public-persona responses, not like unconditional, generic market commentary. Table A1 extends the public-recommendation overlap test beyond that result. Specifically, in Table 3 of the main paper we use exact same-account, same-stock, same-date overlaps. Here, in Table A1, we ask whether the result also holds when public recommendations are allowed to fall on nearby trading dates. It does. In the exact-date sample, meaningful low-speculation interview directions align with public recommendations 91.5%91.5\% of the time. In the narrow one-trading-day window, alignment is still 90.5%90.5\%, and the area-under-the-curve (AUC) is 0.7420.742. These nearby-window checks support the same conclusion as Table 3: when public recommendations provide an observable benchmark, the digital-twin interview responses usually point in the same direction and rank public buys above public sells. Table A1: Additional Validation of Digital-Twin Interview Belief Proxies This table provides additional detail for the public-recommendation overlap validation in Table 3. It compares digital-twin stock-pick interview responses with public recommendations made by the same account about the same stock in the core 429-stock universe. Exact-date matches require the interview and public recommendation to have the same effective trading date. Narrow-window matches also allow the public recommendation to fall one trading day before or after the interview date. Forward-window matches allow the public recommendation to fall on the same trading date or one trading day after the interview date. Sign alignment and AUC are the same validation measures used in Table 3. Sign alignment uses the meaningful low-speculation interview subset, and AUC is the probability that the interview score ranks a public buy above a public sell. Benchmark is the null or placebo benchmark used in Table 3. Percentage rows are in percentage points, AUC is unitless, and N is row-specific. The p-value column reports the sign-test or AUC randomization p-value for the Actual-Benchmark difference. The public-overlap validation branch uses the shared 81-account analysis sample. (1) (2) (3) (4) (5) Measure Actual Benchmark Difference p-value N Exact-date matches: Sign alignment 91.5% 50.0% 41.5 p <0.001<0.001 2,439 Exact-date matches: AUC 0.776 0.522 0.255 <0.001<0.001 4,063 Narrow-window matches: Sign alignment 90.5% 50.0% 40.5 p <0.001<0.001 5,385 Narrow-window matches: AUC 0.742 0.511 0.231 <0.001<0.001 9,797 Forward-window matches: Sign alignment 90.1% 50.0% 40.1 p <0.001<0.001 3,825 Forward-window matches: AUC 0.725 0.509 0.215 <0.001<0.001 7,056 C. Stock-Pick Results Appendix In this section of the Internet Appendix we report two robustness checks for the stock-pick analysis presented in Section 5 of the main paper. The main text shows that stocks with more favorable digital-twin stock-pick signals earn higher future excess returns in equal-weighted daily FamaâMacBeth regressions, Calendar-Time portfolios, and Staggered-Vintage regressions (with appropriate controls in each case). Here, we ask whether this result is driven by larger firms or by the repeated use of overlapping return windows. Table A2 repeats the analysis in Table 5 using event-date market capitalization weights. Columns (1)â(3) estimate the daily FamaâMacBeth regressions by weighted least squares, and columns (4)â(6) value-weight stocks within the high- and low-signal legs of the calendar-time portfolios. The value-weighted estimates are much smaller than the equal-weighted estimates and are not statistically distinguishable from zero. The main stock-pick result is therefore not a large-cap effect. It is stronger in the equal-weighted cross-section, where smaller and larger stock-events receive comparable weight. Table A2: Value-Weighted Digital-Twin Stock-Pick Direction and Future Excess Returns This table is the value-weighted analogue of Table 5. Columns (1)â(3) report daily FamaâMacBeth cross-sectional WLS regressions of cumulative future excess stock returns on one digital-twin direction signal measured before the return window. Stock-events are weighted by event-date market capitalization. Excess returns are measured relative to the S&P 500 and reported in percentage points over 1, 5, and 10 trading days. The table reports the time-series average of the daily slope estimates. Columns (4)â(6) report calendar-time portfolio tests using the same signals. On each event date, stocks are ranked by the indicated signal, the portfolio buys the top quintile and shorts the bottom quintile, stocks are value-weighted within each leg using event-date market capitalization, and each event-date vintage is held for the next 10 trading days, excluding the formation day. The daily calendar-time long-short return averages all active vintages on each holding date. H-L Mean and CAPM Alpha are daily percentage-point estimates while 10-day Alpha is 10 times the daily CAPM alpha. Market adjustment is estimated from a daily CAPM regression of the long-short return on the S&P 500 return. NeweyâWest standard errors are in parentheses, using lag hâ1h-1 in columns (1)â(3) and lag 9 in columns (4)â(6). Coefficients are scaled to a +10 recommendation-point move in Tilt or Meaningful Tilt, a +10 percentage-point move in Net Buy Share, or a +1-unit move in Conviction-Weighted Meaningful Tilt. N/days reports stock-event observations in columns (1)â(3) and calendar holding days in columns (4)â(6). Mean weighted adjusted R2R^2 applies only to the FamaâMacBeth columns. The sample uses 429 stocks and 81 accounts over 50 event dates from December 15, 2025 to March 16, 2026; calendar-time holding dates run from December 16, 2025 to March 30, 2026. Indicators ***, **, * denote significance at the 1%, 5%, and 10% level. (1) (2) (3) (4) (5) (6) Dep. var.: future excess return (p) Dep. var.: H-L portfolio return (p) 1 day 5 days 10 days H-L Mean CAPM Alpha 10-day Alpha Panel A. Tilt Direction signal -0.181 -0.197 -0.478 0.016 0.011 0.107 (0.140) (0.488) (0.667) (0.092) (0.097) (0.968) N / days 21,410 21,410 21,410 71 71 71 Mean wtd. adj. R2R^2 0.051 0.053 0.042 Panel B. Net Buy Share Direction signal -0.041 -0.032 -0.108 0.012 0.008 0.078 (0.033) (0.120) (0.187) (0.094) (0.099) (0.986) N / days 21,410 21,410 21,410 71 71 71 Mean wtd. adj. R2R^2 0.042 0.047 0.041 Panel C. Meaningful Tilt Direction signal -0.003 -0.027 0.043 -0.003 -0.038 -0.378 (0.059) (0.177) (0.232) (0.081) (0.086) (0.859) N / days 20,823 20,823 20,823 71 71 71 Mean wtd. adj. R2R^2 0.025 0.025 0.016 Panel D. Conviction-Weighted Meaningful Tilt Direction signal -0.018 -0.025 -0.054 0.020 0.012 0.118 (0.014) (0.049) (0.067) (0.084) (0.088) (0.879) N / days 20,823 20,823 20,823 71 71 71 Mean wtd. adj. R2R^2 0.055 0.059 0.045 Table A3 provides an alternative way to address correlation among dependent-variable observations arising from overlapping return windows. In the main text we use Calendar-Time portfolios and daily FamaâMacBeth regressions with Newey-West standard errors to determine whether digital twinsâ recommendations can predict returns at the ten-day investment horizon. We also use the staggered-vintage design to exclude overlapping return windows. Here, in Table A3, we also address the overlapping-window problem by partitioning trading days into non-overlapping blocks, where a block is a pre-defined period of time ranging from two to five trading days. Within each block, we average a stockâs directional signal and compute the lagged return and volatility controls. We then use those block variables to predict the same stockâs excess return in the next block. This gives a block-level FamaâMacBeth test in which each cross-section uses only information from the prior block and adjacent dependent variables do not reuse the same return days. Table A3: Non-Overlapping Stock-Pick Direction and Future Excess Returns with Controls This table is a non-overlapping-period robustness companion to Table 6. Each period-specific cross-section regresses next-period stock excess returns on the prior-period direction signal and two controls: the stockâs formation-period excess return and formation-period excess realized volatility. The controls are computed over the same non-overlapping formation block as the signal and are included in the regressions but not reported. The signal is averaged within stock and formation period, then used to predict the same stockâs cumulative excess return over the next non-overlapping return period. Unlike Table 6, whose forward return windows overlap and therefore use NeweyâWest lags of hâ1h-1, this design partitions trading dates into non-overlapping blocks; standard errors are therefore NeweyâWest with lag 0 on the period-specific FamaâMacBeth slope series. Two-day blocks are MonâTue, WedâThu, and Friday-only blocks with Friday excess log returns multiplied by 2; three-day blocks are MonâWed blocks scaled by 2/32/3 and ThuâFri blocks unscaled; five-day blocks run Wednesday through the following Tuesday. Coefficients are percentage-point changes in next-period excess returns for a +10 recommendation-point increase in Tilt or Meaningful Tilt, a +10 percentage-point increase in Net Buy Share, or a +1-unit increase in Conviction-Weighted Meaningful Tilt. Stars use a Tâ1T-1 t-reference, where T is the number of formation-period slopes. N is the number of stock-period observations. The sample uses 429 stocks and 81 stock-pick accounts, with formation periods from December 15, 2025 through March 13, 2026. Indicators ***, **, * denote significance at the 1%, 5%, and 10% level. (1) (2) (3) Dep. var.: next non-overlapping excess return (p) 2-day blocks 3-day blocks 5-day WedâTue Panel A. Tilt Direction signal 0.261 0.250 1.105â (0.218) (0.198) (0.536) N 14,976 10,293 5,148 Formation periods 35 24 12 Mean adj. R2R^2 0.052 0.045 0.034 Panel B. Net Buy Share Direction signal 0.061 0.060 0.271â (0.052) (0.048) (0.133) N 14,976 10,293 5,148 Formation periods 35 24 12 Mean adj. R2R^2 0.052 0.045 0.036 Panel C. Meaningful Tilt Direction signal 0.083 0.095â 0.209 (0.067) (0.053) (0.126) N 14,525 10,190 5,138 Formation periods 35 24 12 Mean adj. R2R^2 0.052 0.044 0.033 Panel D. Conviction-Weighted Meaningful Tilt Direction signal 0.026 0.026 0.115â (0.023) (0.022) (0.059) N 14,525 10,190 5,138 Formation periods 35 24 12 Mean adj. R2R^2 0.051 0.045 0.036 The results presented in Table A3 point in the same direction as those presented in Table 6 and Table 7 of the main paper. The evidence is strongest in the five-day Wednesday-to-Tuesday specification. In that column, a ten-percentage-point increase in Net Buy Share predicts 27.127.1 basis points higher next-period excess return. Tilt and Conviction-Weighted Meaningful Tilt are also positive and statistically significant in the five-day block specification. The results of shorter block specifications are generally positive but do not attain statistical significance at conventional levels. These results show that the stock-pick direction signal is not a large-cap effect and is not driven by the correlation in dependent variables resulting from overlapping daily event windows. The evidence is strongest in equal-weighted stock selection, where smaller and larger stock-events receive comparable weight, and it continues to appear when return windows are forced to be non-overlapping within each block or vintage.