Individual Investing

Is the shift from keyword-based search engines to artificial intelligence (AI) as implemented with large language models (LLM) affecting typical investor behavior? In the May 2026 revision of their paper entitled “The Double-Edged Mind: How LLMs Expand Stock Market Participation Yet Strengthen Confirmation-Seeking”, Cara Damm, Kevin Bauer, Florian Hett and Loriana Pelizzon address this question via an online experiment that randomly assigns participants to groups of volunteers who have access to keyword-based search engines, an LLM-based chatbot (unlabeled Gemini 2.0 Flash) or no information filtering tools. They design the experiment with two stages:

• Stage 1: With access only to the name of the investment, participants initially choose a standard exchange-traded fund (ETF), a matched Environmental-Social-Governance (ESG) ETF or risk-free cash.
• Stage 2: They revisit their decisions after receiving access to their respective assigned information filtering tools.

At the end, participants who chose the cash alternative keep their initial investments, while those who chose an ETF receive their initial investments adjusted by a fund return. Using responses from 374 participants in the experiment, they find that:

Keep Reading

Does inclusion of artificial intelligence (AI)-generated summaries in financial news articles influence investor response to the articles? In the May 2026 revision of their paper entitled “Generative AI and Investor Processing of Financial Media”, Tony Cho, Allen Huang, Joseph Pacelli and Kristina Rennekamp examine whether and how AI-generated financial news summaries influence investor information processing in two ways:

1. They compare stock trading and return responses to articles with and without AI summaries.
2. They test article recall in  a controlled experiment in which business school graduate students read randomly assigned news articles with or without an AI summary.

Using a sample of 1,734 Wall Street Journal articles covering 158 unique firms, of which about a third include an AI summary, and associated stock return and trading data from July 2024 through June 2025, they find that:

Keep Reading

Are large language models (LLM) robust financial advisors for individuals? In their March 2026 paper entitled “AI Financial Advice: Supply, Demand, and Life Cycle Implications”, Taha Choukhmane, Tim de Silva, Weidong Lin and Matthew Akuzawa examine the personal financial advice from LLMs. They mainly use GPT-5.2 but repeat analyses using Gemini 3 Flash as a robustness check. Specifically, they:

• Construct a life cycle model of income/spending/saving/investment, with labor market shocks and asset returns calibrated to U.S. data.
• Collect questions (prompts) from a demographically representative sample of about 1,000 U.S. adults about spending and investing, including summaries of respective financial situations.
• Simulate life cycle paths of individuals for each year from ages 22 to 90 who follow two-pass advice in LLM responses to prompts from survey participants matched by age, income and employment status. The first pass solicits textual advice, and the second translates text to quantified saving, spending and asset allocation recommendations.

They consider two benchmarks: (1) the optimal behaviors for the life cycle model simulations; and, (2) substitution of survey respondent prompts with expert (academic) prompts that ask the LLM to give professional life cycle advice under modern portfolio theory, including explicit personal situations/economic assumptions. Using the specified life cycle model and LLM prompts, they find that: Keep Reading

What criteria do individual investors use when deciding which stocks to acquire? In their December 2025 paper entitled “How Investors Pick Stocks: Global Evidence from 1,540 AI-Driven Field Interviews”, Byoung-Hyoun Hwang, Don Noh and Sean Seunghun Shin report the results of 1,540 interviews moderated by artificial intelligence (AI) with actual investors across 10 countries about how they decide which stocks to buy. They partner with CoreData Research to select interviewees, with 280 from the U.S. and 140 each from Australia, Canada, France, Germany, India, Japan, South Korea, Singapore and the UK. AI-moderated interviews, initiated by sharing a link to a chatroom, maintain constant interviewer quality with no unwanted variation in personal chemistry. The AI first ask interviewees how they picked stocks in the past and then ask 10 to 15 follow-up questions to determine decision factors. Average interview is 23 minutes. The authors then: (1) conduct an AI textual analysis of each interview transcript, tracing investor reasoning from beliefs and preferences to buy decision; and, (2) compare results across investors to identify common decision factors. Based on results from 1,540 investor interviews, they find that:

Keep Reading

Should measures of long-term investment performance incorporate ways in which typical individual investors handle their portfolios over a lifetime rather than an idealized perspective such as buy-and-hold with all distributions reinvested? In his December 2025 paper entitled “Measuring Investor Outcomes”, Hendrik Bessembinder argues that investment performance measures should be realistic and discusses alternative measures of returns. Using monthly performance data for the broad U.S. stock market from the end of 1926 through 2022, he finds that: Keep Reading

How can typical investors/managers rigorously translate their views on factor/style and sector/theme exposures into a portfolio of exchange-traded funds (ETF). In their November 2025 paper entitled “Implementing Systematic Risk Premia, Factor-Based Strategies, and Sector Rotation with ETFs”, Nino Antulov-Fantulin, Petter Kolm and Mario Šikic describe a methodology for constructing systematic, long-only investment strategies for family offices and wealth managers using exchange-traded funds (ETFs). The approach, which requires no explicit return forecasts, integrates beliefs on factors and sectors by tilting the portfolio toward them while controlling tracking error relative to a selected benchmark. It employs daily covariance of ETF returns to forecast variance/risk. They illustrate the approach through three case studies in equities and fixed income. Using theory and the most recent five years of returns for selected ETFs in case studies, they find that: Keep Reading

Can a large language model (LLM) applied to social media data catalog the strategy choices, sentiment and trading behavior of retail investors? In the November 2025 revision of their paper entitled “Wisdom or Whims? Decoding Retail Strategies with Social Media and AI”, Shuaiyu Chen, Lin Peng and Dexin Zhou apply GPT-4 Turbo and BERT to StockTwits messages to classify retail investor strategies as: (1) technical analysis (TA); (2) fundamental analysis (FA); (3) other strategies (such as options trading); or, (4) no strategy. They then relate strategy classes to future stock returns and trading activity. Using StockTwits messages posted by 840,846 investors on 7,834 common stocks and associated accounting, price, trade order and financial news during January 2010 through June 2023, they find that: Keep Reading

How fast should individuals plan to grow net worth as they age? To investigate, we examine median levels of household (1) total net worth and (2) net worth excluding home equity from several vintages of U.S. Census Bureau data. We make the following head-of-household age cohort assumptions:

• “Less than 35 years” means about age 30.
• “35 to 44 years” means about age 39.
• “45 to 54 years” means about age 49.
• “55 to 64 years” means about age 59.
• “65 to 69 years” means about age 67.
• “70 to 74 years” means about age 72.
• “75 and over” means about age 78.

We calculate wealth growth between these ages as compound annual growth rates (CAGR). Using median levels of total net worth and net worth excluding home equity from 2000. 2005, 2010, 2014, 2017, 2019, 2021 and 2022 Census Bureau summary tables, we find that: Keep Reading

Do individual investors trade carefully and cautiously, or do they make snap judgments? In their March 2025 paper entitled “The Research Behavior of Individual Investors”, Toomas Laarits and Jeffrey Wurgler employ browser histories for an approximately representative sample of U.S. individual investors to address:

• How much time do they spend on stock research?
• Which sites do they use?
• On which stocks do they focus?
• How proximate is their research to their trades or to corporate events?
• What types of information do and do not attract their attention?

Using browser and trading histories, with confidential/personal data removed, from 484 volunteer households that make 2,911 individual U.S. stock trades via online brokerage accounts during January, February, March and June of 2007, they find that:

Keep Reading

Alternative trading platforms (such as Blue Ocean and Interactive Brokers) offer trading in many U.S. stocks and exchange traded products between 8PM and 4AM (nocturnal), letting U.S. retail and Asian investors trade continuously five days a week. What are the implications? In their March 2025 paper entitled “Nocturnal Trading”, Gregory Eaton, Andriy Shkilko and Ingrid Werner examine the emergence of 24-hour trading  in the last three years and its implications for financial markets. They focus on nocturnal trading market quality and price discovery. Using 24×5 trading records during 2021 through 2024, they find that:

Keep Reading