In the early 1980s, Richard Dennis and William Eckhardt ran one of the most famous experiments in investment history.
Dennis, a legendary commodities trader, believed trading could be taught. Eckhardt was more skeptical. The argument was not merely about trading technique. It was about talent. Are great traders born, or can you select people with the right traits, give them a repeatable process, and turn them into exceptional market operators?
To test the idea, they recruited a group of people from outside the traditional finance elite, trained them in a systematic futures-trading strategy, and gave them capital. The trainees became known as the Turtle Traders. Their system was not mystical. It was rules-based trend following: identify breakouts, size positions by volatility, cut losses, let winners run, and avoid emotional override. The published "Original Turtle Trading Rules" describe a complete system that minimized discretionary judgment and emphasized process discipline.
The Turtle story has become legendary, and like many legends, it is probably cleaner in the retelling than it was in reality. The experiment was not an academic study. It had no randomized control group, no public participant-level dataset, and no independent scientific audit. But the core story is credible: Dennis and Eckhardt did run the program, and several former Turtles went on to become successful systematic traders.
The enduring lesson is not that anyone can become a great portfolio manager. That would be too simple. The deeper lesson is that some forms of investment skill can be decomposed, trained, tested, and scaled outside the traditional pedigree pipeline.
That idea is now becoming much more important.
For decades, the selection of portfolio managers has been dominated by pedigree. Elite university. Elite bank. Elite hedge fund. Famous mentor. Strong references. Familiar résumé. The implicit assumption is that investment talent is best found by looking upstream at credentials.
But the Turtle experiment suggested another possibility: perhaps the better approach is to look directly at behavior under uncertainty. Can this person follow a process? Can they size risk? Can they tolerate drawdowns? Can they keep acting rationally when the feedback loop is noisy and emotionally punishing?
That question is even more relevant today because AI is starting to lower the technical barriers to investment research in the same way it has lowered the barriers to software engineering.
The next generation of portfolio managers may not all come from Goldman, Morgan Stanley, Citadel, Point72, or Harvard Business School. Some will. But many may come from the founder ecosystem, AI labs, open-source communities, quant competitions, data-science platforms, and nontraditional backgrounds that would historically never have been considered "PM material."
The old talent filter is starting to look outdated.
From pedigree to proof-of-work
WorldQuant offers one of the clearest modern versions of the Turtle idea.
WorldQuant's thesis is often summarized as: talent is global, opportunity is not. The firm built platforms and competitions that allow participants around the world to create "alphas"—mathematical signals designed to predict future price movements. Its BRAIN platform describes itself as a place to "Learn, Earn and Grow" in quantitative finance, and WorldQuant has used competitions such as the Global Alphathon and International Quant Championship as both a research engine and a talent-discovery pipeline.
The scale is remarkable. WorldQuant said its 2022 Global Alphathon attracted more than 14,000 participants from more than 100 countries, with prizes and potential full-time or internship opportunities. Reuters later reported that WorldQuant's 2025 International Quant Championship reached a record 80,000 university participants, double the prior year, with winners from South Korea, India, Kenya, and Taiwan. Reuters also reported that Igor Tulchinsky said many teams were using AI and language models to help them build trading models, read documents, evaluate ideas, and run simulations.
That is not a small detail. It is the whole point.
WorldQuant is not selecting talent only by asking, "Where did you go to school?" It is asking, "Can you produce useful signals in a constrained research environment?" That is a much better test of relevant ability.
The now-famous anecdote of a farmer in rural China winning or performing well in a WorldQuant alpha competition is powerful because it breaks the mental model of what financial talent is supposed to look like. Even if the individual had technical education, the story still matters: the person was far outside the traditional Wall Street pipeline, yet the platform surfaced relevant ability.
That is the shift: from credential-based selection to performance-based discovery.
The Turtle Traders were selected, trained, and observed. WorldQuant industrialized that idea with software, data, simulations, and global reach. AI will take it further.
Good PMs and good founders look more similar than people think
The traditional view treats founders and portfolio managers as very different archetypes.
The founder builds. The PM allocates. The founder creates value. The PM prices value. The founder operates a company. The PM operates a book.
Those differences are real. But at the level of temperament and decision-making, the overlap is striking.
Good founders and good portfolio managers both operate in environments where the future is unknowable, feedback is noisy, and the cost of self-deception is high.
They both need to:
- Make decisions before they have complete information
- Distinguish good decisions from lucky outcomes
- Manage downside while preserving upside
- Stay calm through drawdowns
- Size bets intelligently
- Cut failing ideas without ego
- Double down when real pull appears
- Learn from weak signals
- Avoid ruin
A founder's "portfolio" is not made of public equities. It is made of runway, product bets, hires, customer segments, distribution channels, investor relationships, and strategic options.
But the underlying logic is similar. A founder is constantly asking: where should I allocate scarce resources under uncertainty? Which bet deserves more capital, time, and attention? Which bet should be killed? When is weak evidence enough to continue, and when is it a mirage?
That is portfolio management.
The best founders are not reckless risk-takers. They are risk shapers. They take ambiguity and convert it into experiments, milestones, and asymmetric opportunities. The best PMs do something similar. They take uncertainty and convert it into positions, hedges, risk limits, and expected value.
This is why founders may be a more natural future source of PM talent than the industry currently recognizes.
Poker is not an accident
It is also not an accident that poker has long appealed to both investors and founders.
Poker is a compact simulation of decisions under uncertainty. It combines probability, psychology, incomplete information, emotional regulation, and bet sizing. The point is not that poker makes someone a good investor or founder. It does not. The point is that people who are drawn to poker often enjoy the same kind of thinking required in startups and public markets.
Poker teaches a distinction that most people struggle with: a good decision can lose, and a bad decision can win.
That distinction is central to investing. A stock can go up after a bad thesis. A good short can lose money because the market stays irrational longer than expected. A correct view can be too early. A wrong view can be bailed out by liquidity.
It is also central to startups. A founder can make a good strategic decision and still fail because timing, distribution, or capital markets go against them. Another founder can make a poor decision and be rewarded temporarily by hype.
Poker attracts people who are comfortable with this separation between process and outcome. That is the same psychological muscle required by founders and PMs.
The common liking for poker is therefore not just a cultural quirk. It reveals a shared appetite for probabilistic, adversarial, high-variance decision-making.
The early signs are already visible
The next question is whether founders and technical talent actually want to enter investment management.
There are early signs that they do.
Apoorva Mehta, the co-founder of Instacart, recently launched Abundance, a hedge fund reportedly using AI agents to handle much of the investment process, including idea generation, research, stock selection, bet sizing, and execution. Bloomberg described the firm as being built by a small team of quantitative researchers, engineers, and AI experts, with thousands of bots scouring the internet for trade ideas and conducting research.
One case is not a trend. But the direction is notable. Mehta is not simply becoming an angel investor or venture capitalist, which would be the normal post-founder path. He is treating investment management as a system-building problem.
That is the important distinction.
The same pattern appeared, in a different form, at DeepMind. Reporting around "Project Mario" described an internal effort by DeepMind researchers to explore trading and hedge-fund-like applications using AI. Whether or not that project was commercially viable, the fact that frontier AI researchers were drawn to markets is telling. Markets are data-rich, adversarial, measurable, and directly monetizable. For a certain type of technical mind, they are an irresistible arena.
WorldQuant shows the talent-pipeline side. Mehta's Abundance shows the founder side. DeepMind's Project Mario shows the AI-researcher side. None of these alone proves a mass migration. Together, they suggest that the boundary between technology company, AI lab, quant research platform, and investment firm is becoming porous.
Investment management is starting to look less like an apprenticeship profession and more like a software-native, AI-native systems problem.
AI changes the talent equation
The most important reason this shift may become inevitable is AI.
Historically, public-market investing had high barriers to entry. Some were regulatory and capital-related, but many were technical and institutional:
- Access to data
- Ability to code
- Ability to build research infrastructure
- Ability to parse filings, transcripts, news, and alternative data
- Ability to run backtests
- Ability to monitor portfolios
- Ability to generate research at scale
- Ability to translate ideas into executable positions
AI lowers many of these barriers.
Just as AI coding tools allow non-coders or semi-technical founders to build software prototypes, AI research tools will allow nontraditional investors to build investment workflows that previously required analysts, engineers, data teams, and expensive infrastructure.
A founder who could not previously build a research platform may now be able to assemble one with AI agents, data APIs, coding assistants, document intelligence, and simulation tools.
A domain expert who could not code may soon be able to express an investment hypothesis in natural language, have an AI system translate it into testable features, run historical analysis, identify relevant filings, compare peers, and generate a candidate portfolio.
A technical researcher who does not come from finance may be able to use AI to understand accounting, market structure, regulatory filings, and portfolio construction faster than before.
This does not mean investing becomes easy. Lower barriers increase competition. More people entering the arena may compress obvious edges. AI can also increase overfitting, false confidence, and copycat strategies.
But it does mean the old selection mechanism becomes less defensible.
If AI allows more people to do the technical work, then the scarce trait shifts from "can this person access the machinery?" to "does this person have good judgment, original insight, emotional discipline, and the ability to design a repeatable process?"
That favors nontraditional PMs.
It favors founders.
Why founders are especially well positioned
Founders are natural candidates for this new PM archetype because they are already trained in high-uncertainty resource allocation.
They are used to building systems from scratch. They are used to making decisions with incomplete information. They are used to operating without institutional permission. They are used to raising capital, telling a story, recruiting talent, and iterating quickly.
A traditional PM may know markets better. That still matters enormously.
But an AI-native founder may be better at building the machine around the investment process: the data layer, agent workflows, research interface, evaluation harness, feedback loops, and user experience. As investment management becomes more automated, that system-building capability becomes more valuable.
The PM of the future may not look like a lone stock picker with a Bloomberg terminal. The PM of the future may look more like a founder of an intelligent capital-allocation system.
That person may combine:
- Market judgment
- Product intuition
- Technical leverage
- AI orchestration
- Risk discipline
- Capital formation ability
This is a different talent profile from the traditional analyst-to-PM path.
The opportunity: infrastructure for nontraditional PMs
If this thesis is right, then the biggest opportunity is not merely to become one of these new PMs.
The bigger opportunity is to build the platform that enables them.
Because if thousands of nontraditional PMs emerge, they will need infrastructure:
- Research tooling
- AI agents for filings, transcripts, macro, and news
- Backtesting environments
- Portfolio construction tools
- Risk management
- Compliance workflows
- Investor reporting
- Attribution
- Capital formation
- Model evaluation
- Data access
- Performance verification
- Fund administration integrations
Today, the hedge fund industry is still structurally optimized for established managers, institutional allocators, and large funds. It is not designed for a world where a brilliant founder, engineer, domain expert, or independent researcher can use AI to build an investable strategy from outside the traditional pipeline.
That gap is a generational venture opportunity.
The analogy is not just Robinhood for trading. It is not just Bloomberg for individuals. It is not just Kaggle for finance. It is something broader: an AI-native operating system for discovering, evaluating, launching, and scaling new investment managers.
WorldQuant built part of this for quant alphas. But the opportunity is much larger if extended to fundamental research, thematic investing, event-driven strategies, private/public market crossover analysis, and founder-led investment theses.
The winning platform would not simply provide tools. It would provide trust.
That means answering the questions allocators care about:
- Is this strategy real or overfit?
- Is the PM skilled or lucky?
- Is the process repeatable?
- Is the risk controlled?
- Is the performance auditable?
- Is the edge capacity-constrained?
- Is the manager operationally sound?
- Can capital be allocated safely?
The platform that solves those problems becomes more than software. It becomes market infrastructure for a new class of investment talent.
The old model will not disappear, but it will lose monopoly power
Pedigree will not stop mattering. Elite institutions still produce talented people. Existing hedge funds still train analysts and PMs. Reputation still matters when managing other people's money.
But pedigree will no longer be enough, and eventually it may no longer be the default filter.
The better filter is proof-of-work.
Can you generate differentiated insight? Can you test it? Can you express it as a repeatable process? Can you manage risk? Can you survive variance? Can you attract capital? Can you compound?
That is the same question Dennis and Eckhardt were asking in their own way with the Turtles. It is the same question WorldQuant asks through alpha competitions. It is the same question poker asks every hand. It is the same question founders face every day.
The tools have changed. The principle has not.
Conclusion
The Turtle Traders showed that investment talent could be trained and tested outside the traditional finance pipeline. WorldQuant showed that alpha research talent could be discovered globally through platforms, simulations, and competitions. Poker explains why founders and investors often share the same psychological orientation: probabilistic thinking, emotional control, asymmetric payoffs, and comfort with noisy feedback.
Now AI is lowering the technical barriers that once kept many talented outsiders away from investment management.
That makes the emergence of nontraditional PMs not just possible, but likely. Some will be founders. Some will be engineers. Some will be domain experts. Some will come from places the traditional industry would never have searched.
The future of investment management will not be purely pedigree-driven. It will be increasingly platform-driven, AI-enabled, and proof-of-work based.
The generational opportunity is to build the infrastructure that lets this new class of PMs form, prove themselves, raise capital, and manage money responsibly.
The next great portfolio manager may not come from the usual résumé stack.
They may look more like a Turtle, a WorldQuant participant, a poker player, or a founder with an AI research system.