Why AI Won't Replace Stock Pickers - Jack Woodruff, Candlestick & Vinoo Ganesh, Kepler

Summary

Jack Woodruff argues that AI is unlikely to change the core of long-term fundamental stock selection: investment edge still comes from developing a differentiated view by combining several pieces of evidence. AI can speed up information processing and help identify promising research areas, but people remain responsible for deciding what matters, investigating it, and acting on it.

AI may reduce routine data work and preparation time, but Woodruff expects the time savings to raise research expectations rather than simply reduce staffing. If every fund uses similar tools, basic automation becomes a competitive requirement. An edge will depend on how effectively investors combine human judgment and machines, ask questions, and uncover ideas before others do.

Finance-Specific Points

• Strategies and assets: The discussion covers long/short equities and fundamental stock selection. No specific stocks, tickers, ETFs, bonds, commodities, or cryptocurrencies are named.
• Sources of alpha: Woodruff describes an “investment mosaic” of roughly three or four confirming data points, potentially from different sources. A single data point is rarely decisive. Differentiated views and timely discovery are central to generating alpha.
• AI’s role: AI could compress some preparation from 6–8 hours to about 5 minutes and automate large-scale data work. It may also help direct analysts toward research projects worth pursuing. Woodruff does not present AI as a reliable source of stock-buy recommendations.
• Verification and risk: Woodruff says he would not rely on a model-generated number without knowing its origin and being able to verify it. He warns that data-entry errors can cost millions of dollars and argues that accuracy and auditability are essential before trusting AI outputs or using them in trading decisions. He also flags the risk of internal data leaking to vendors, citing data isolation as an important security concern.
• Competitive advantage: Arbitrage opportunities can disappear as other investors copy profitable strategies. With broadly shared models and information, advantage may come from better questions, judgment, data, and speed in connecting machine-generated leads to human research.
• Fund staffing and organization: Woodruff expects continued demand for experienced stock pickers, but potentially fewer conventional analyst roles and more investigative, project-research, or data-engineering skills. He anticipates a less traditional, more vertical structure than the pyramid model he describes, while noting that the future composition remains uncertain.
• Limits of a “no-analyst” fund: He doubts AI agents will outperform evolving human-led fundamental funds. He says that if people could be removed entirely, computing power and information access would become decisive—similar to the dynamics of quantitative investing.

Numbers and Examples

• At Citadel, 18 people covered 400 stocks; at Candlestick, Woodruff says there were 30 stock pickers.
• The conversation asks whether one strong analyst could cover 400 stocks by 2030. Woodruff emphasizes diminishing returns to time spent per stock and the continuing human bottleneck.
• In a preparation example, AI reduces the task to 5 minutes, compared with 6–8 hours previously. Woodruff expects saved time to be redirected into deeper research, industry analysis, and specialist calls, rather than necessarily reducing headcount.
• In a portfolio-coverage illustration, if a portfolio holds 20 names, covering 100 stocks means selecting 20% of the coverage universe; covering 200 means 10%; and covering 400 means 5%. Woodruff notes that broader coverage only helps if research remains sufficiently thorough.
• He contrasts strategies ranging from turnover of roughly 9,000 times per year to holding assets for five years.
• The discussion considers what a fund launched in 2027 might build or buy.

Research Framework

1. Use AI to automate routine data gathering, organization, and preparation.
2. Treat AI outputs as leads, not investment conclusions; verify the underlying numbers and sources.
3. Use the tools to identify where deeper research may be valuable.
4. Investigate those areas to develop a differentiated view, testing it against multiple, potentially independent data points.
5. Seek to identify and act on promising ideas before they become widely recognized or copied.

Disclosures and Sources

Disclosures: No explicit “not financial advice” disclaimer appears in the subtitles.

Presenters and sources: Jack Woodruff, founder and former CIO of Candlestick Capital Management and former partner and portfolio manager at Surveyor Capital, a division of Citadel; Vinoo Ganesh, CEO of Kepler. The discussion is from AI Engineer.