# Why AI Won't Replace Stock Pickers

Jack Woodruff, Candlestick Capital & Vinoo Ganesh, Kepler | AI Engineer NYC 2026 | 18:46

Source: https://www.youtube.com/watch?v=y2dyD6slKxE
Channel: AI Engineer (https://www.youtube.com/@aiDotEngineer). Summarised by AIE Talks.
Page: https://aietalks.com/talks/why-ai-wont-replace-stock-pickers
Published: 2026-10-30
Tags: finance, human-in-the-loop, security

## TL;DR
- AI will change how investors process information, but finding a divergent view will still depend on human judgment.
- Automation will reduce preparation time and push analysts toward proprietary research, field work, and projects that can produce an investment edge.
- A fully automated fundamental hedge fund may reproduce an older fund, but it will lag funds whose human teams keep improving.

## Summary
Jack Woodruff argues that AI will change the work around fundamental investing without removing the need for stock pickers. The central limit is human attention: investors still need to absorb information, decide what deserves time, and form a view that differs from the market. AI can summarize filings, process data, and direct analysts toward promising names, but the time saved will raise the amount of work expected from everyone. Woodruff places special weight on auditability. He would not trust a model number unless he could trace its source, because a single data-entry error can cost millions. He expects funds to need fewer traditional preparers and more senior investors, investigative journalists, project researchers, and data engineers. Once every fund has similar models and information, the advantage will come from asking better questions, combining tools with judgment, and finding research projects others have missed. He rejects the idea that AI agents will soon run the best fundamental funds without humans.

## Key ideas
### AI changes information processing more than stock-picking judgment
[00:53](https://www.youtube.com/watch?v=y2dyD6slKxE&t=53s)
Woodruff does not expect long-term stock picking to change much. He describes investing as a mosaic in which an investor finds a view that differs from the market. AI will change how people synthesize information and where the basic table stakes sit. It may change the traditional roles of portfolio managers and analysts, but Woodruff says the tools are still too new to predict the final structure. The investment question remains whether someone can develop a differentiated view. AI affects the preparation around that question, rather than producing the answer by itself.

### Human attention remains the limit on how many stocks one person can cover
[02:21](https://www.youtube.com/watch?v=y2dyD6slKxE&t=141s)
Woodruff says humans are still the bottleneck because investors have to get information into their heads. Investment management is also a time-management exercise. Spending all of one's time on one stock might produce deep knowledge, but leaves no time for other names, and spreading attention across more stocks creates diminishing returns. AI could help an investor cover more names or raise the baseline level of preparation. Woodruff gives the example of work that once took six to eight hours being prepared in five minutes. He expects the saved time to move into proprietary research, channel checks, and industry work.

### Automation will raise the work expected from analysts
[04:53](https://www.youtube.com/watch?v=y2dyD6slKxE&t=293s)
Woodruff does not think junior analysts will simply stop reading filings or earnings transcripts. Close coverage still requires getting that information into the investor's head, even if summaries reduce the time required. He expects the collection and processing of large datasets to be easier to outsource. Agents may also direct analysts toward the names and questions that deserve attention. When routine preparation becomes automatic, it becomes the minimum required by every competing fund. The extra time then goes into proprietary calls, industry builds, and other research that may reveal several independent signals pointing toward a divergent view.

### No model number should be trusted without an audit trail
[07:23](https://www.youtube.com/watch?v=y2dyD6slKxE&t=443s)
Woodruff says he would never sign off on a model-produced number without seeing where it came from. He connects this to his accounting background and to the need to audit every number. Human data-entry mistakes were rare, but a single mistake could cost millions of dollars. He is also unwilling to let AI tell the firm what to do or run trading algorithms if the underlying information cannot be trusted. Woodruff mentions people asking Claude to help build models, then having to audit the numbers themselves. That raises a direct question about how much time automation saves when verification remains necessary.

### Proprietary information and security must stay inside the investment process
[09:16](https://www.youtube.com/watch?v=y2dyD6slKxE&t=556s)
Woodruff says he would be nervous about exposing internal data to an AI vendor. He leaves the precise security decisions to his CTO, but says data security and isolation are essential. He describes the current phase as an automation stage in which firms assemble tools and data. The more interesting use comes later, when those tools help analysts decide where their time is most valuable. Woodruff separates that guidance from the investment output itself. Earnings previews and discounted cash flow models are preparation. Making money requires finding a divergent view, and anything proprietary to that process must be protected.

### When models become common, edge comes from choosing the right research
[11:33](https://www.youtube.com/watch?v=y2dyD6slKxE&t=693s)
Woodruff expects funds to share much of the same information and eventually many of the same models. Differences will come from how investors prompt systems, how quickly they combine machine output with human judgment, and which projects they choose to investigate. The automated work should identify something worth studying rather than tell the investor to buy a stock. An investor then builds conviction through further research while others may be looking elsewhere. Woodruff says there are unlimited possible projects, so identifying the right one early matters. If AI can complete the entire process by itself, he questions what humans would contribute.

### Future fund teams will be more vertical and more specialized
[13:29](https://www.youtube.com/watch?v=y2dyD6slKxE&t=809s)
If Woodruff launched a fund in 2027, he would still want senior stock pickers, though he might need fewer conventional analysts. He expects other roles to grow, including investigative journalists, people who can handle deep project work, and data engineers. His former structure was a pyramid, with many analysts supplying information to senior stock pickers, portfolio managers, and the CIO. He imagines a more vertical structure with more producers and makers, meaning people who make investment decisions and returns, and fewer preparers. He may still retain preparers when the fund needs more proprietary information to feed its research process.

### The no-analyst hedge fund will lag human-led funds
[17:46](https://www.youtube.com/watch?v=y2dyD6slKxE&t=1066s)
Woodruff calls the no-analyst hedge fund the finance idea everyone is discussing. He thinks AI agents might reproduce a fund that existed a year earlier, but that result would be weaker than what human-led funds develop in the following year. He does not expect humans to leave the loop in fundamental investing. Quantitative funds have operated with humans out of the loop for decades, so that would be a different case. If fundamental investing becomes fully automated, Woodruff says the contest would turn into a race for compute and information. He does not think that is the likely path.

## Notable quotes
- Jack Woodruff: "Long-term stock picking, I don't believe it's really going to change much at all." (00:53)
- Jack Woodruff: "The biggest issue is humans are still the bottleneck." (02:21)
- Jack Woodruff: "No, absolutely not." (07:23)
- Jack Woodruff: "Investment management isn't about doing earnings previews. It's not about building DCFs. It's about making money." (10:34)
- Jack Woodruff: "I just don't think you're ever going to really be able to take humans out of the loop." (18:08)

## Tools & references mentioned
- Candlestick Capital Management
- Surveyor Capital
- Citadel
- Kepler
- Claude

## Who should watch
- You are building or buying AI tools for a fundamental investment team and need to know where auditability matters.
- You manage analysts whose routine data work is becoming automated and need to rethink how their time is spent.
- You are considering an AI-only investment fund and want Woodruff's view on its limits compared with human-led funds.

## Related talks

- [The Investor's Edge Lives in Their Head](https://aietalks.com/talks/the-investors-edge-lives-in-their-head) (Anekha Sokhal, Moshi, 16:29)
- [Proving Alpha: How a Hedge Fund Measures AI's ROI](https://aietalks.com/talks/proving-alpha-how-a-hedge-fund-measures-ais-roi) (Hari Kumar, Contour Asset Management, 18:24)
- [How to Hire AI Engineers when EVERYONE is cheating with AI](https://aietalks.com/talks/how-to-hire-ai-engineers-when-everyone-is-cheating-with-ai) (Beth Glenfield, DevDay, 06:45)
- [Trading Desks to Clinical Trials: Parallels in Applied Vertical AI](https://aietalks.com/talks/trading-desks-to-clinical-trials-parallels-in-applied-vertical-ai) (Ayush Bhardwaj, Allos AI, 20:02)
- [What Technical Due Diligence Finds When the Deck Says "70% Automated"](https://aietalks.com/talks/what-technical-due-diligence-finds-when-the-deck-says-70-automated) (Soukaïna Abelhad, 06:29)