Will AI Make Markets Less Efficient?

Episode Summary

Executive Summary: The episode examines how AI, especially generative AI and language models, is reshaping quantitative investing. Osman Ali explains that QIS uses AI to process massive data sets, extract sentiment, and find alpha, but argues these tools do not eliminate market inefficiency. Instead, broader adoption may create new crowding, herd behavior, and opportunities for skilled investors with the right data, technology, and context.

Main Topics: AI in quantitative investing (Priority: 5/5): Osman Ali describes how QIS uses AI and machine learning across public markets and asset classes to analyze large data sets and improve investment decisions. Evolution from traditional sentiment analysis to LLMs (Priority: 5/5): The discussion traces sentiment analysis from bag-of-words methods in 2008 to modern fine-tuned language models that capture nuanced financial language across multiple languages. What drives stock returns today (Priority: 5/5): Ali argues that market sentiment, themes, and technicals increasingly matter more than fundamentals over a 6-12 month horizon, especially in equities. Democratization of AI and edge (Priority: 4/5): The speakers debate whether widespread access to powerful models erodes advantage; Ali says edge still comes from data, technology, and experience/context. Market efficiency vs. new inefficiencies (Priority: 5/5): Ali contends that AI may improve price discovery in some inefficient areas, but also encourage herd behavior and crowding, creating different inefficiencies. Career advice for future investors (Priority: 3/5): Ali emphasizes that future investors need data science and technology skills, but must combine them with investing judgment and context. Team structure and operating model (Priority: 3/5): Despite automation, QIS remains about a 100-person global team, highlighting that human oversight and culture remain important alongside machine-driven work.

Key Arguments: AI is not new to quantitative investing; it has long been embedded in QIS workflows, and generative AI is an evolution rather than a revolution. Language models let investors capture management, sell-side, and public sentiment with much finer granularity, including in non-English languages such as Japanese. Over a 6-12 month horizon, more than half of equity returns may be driven by market perception, themes, and sentiment rather than business fundamentals. The best investors will still need an informational edge derived from proprietary data, scalable technology, and deep investing context. Because investing is zero-sum, broad access to AI does not mean everyone can outperform; skill and differentiation still matter. AI may make some market segments more efficient by improving price discovery, but it can also increase crowding and herd behavior, creating new inefficiencies. As AI tools become more widely used by retail and institutional investors, understanding investor psychology becomes more important for predicting returns.

Data Points: QIS team history: 37 years - Osman Ali says the quantitative investment strategies team has been investing for about 37 years, dating back to the late 1980s. Sentiment analysis origin in team workflow: 2008 - Ali cites 2008 as the start of their foray into sentiment analysis using more traditional bag-of-words methods. Stocks analyzed daily: 15,000 stocks every single day - Ali describes the breadth of the team’s daily quantitative coverage. Return drivers over next 12 months: More than 50% - Ali states that for equities, more than half of expected returns over the next 12 months come from market perception/themes rather than fundamentals. Investment horizon discussed: 6 to 12 months - Ali frames the model inputs and return drivers around the near-term investing horizon. Team size: About 100 people worldwide - Ali says the QIS team remains roughly the same size despite increased automation and AI adoption. Podcast recording date: May 1, 2026 - The episode ends with a production note specifying the recording date.

Pivotal Quotes: "I think investing is a zero-sum game. I don't think everyone can outperform the market. It's mathematically impossible." — Osman Ali: Ali explains why democratized AI access does not eliminate the need for an informational edge. "What hasn't changed is the importance of investor sentiment in making investing decisions." — Osman Ali: Ali describes continuity between older sentiment tools and today’s large language models. "These tools, which you might think would converge to a more efficient market, are, in Osman's view, creating more opportunity, more alpha opportunity, less efficiency." — George Lee: Lee summarizes Ali’s contrarian view that AI may increase, not reduce, opportunities for active investors.

Implications: AI will likely reshape investing by widening access to analysis while also amplifying crowding and behavioral effects. For investors, durable edge will depend less on model access alone and more on proprietary data, judgment, and understanding how AI changes market behavior.