When Jim Simons died on May 10, 2024, the obituaries struggled to decide what to call him.
He was, by training and temperament, a serious mathematician. He earned his PhD at Berkeley in 1961 at the age of 23. He held faculty positions at MIT and Harvard. He chaired the math department at Stony Brook University. He co-developed something called Chern-Simons theory in differential geometry — work that initially looked like pure mathematics, but which later turned out to be foundational in theoretical physics, quantum field theory, and string theory. The “Simons” in Chern-Simons is his.
He also spent four years working as a codebreaker at the U.S. National Security Agency during the Vietnam War, until he was fired in 1968 for publicly opposing the war in a New York Times letter.
By any honest accounting, he was one of the most accomplished mathematicians of his generation.
He was also, by the time he died, almost certainly the richest mathematician in human history. His personal net worth was around $31 billion. He had given roughly $6 billion more to philanthropy through the Simons Foundation.
He made it by trusting mathematicians over financial experts.
This piece is about that story — what he actually did, why it worked, and why the Medallion Fund is the single most important case study any senior executive can read in 2026.
The numbers
In 1982, Simons founded a hedge fund called Renaissance Technologies. In 1988, he launched its flagship Medallion Fund.
The numbers that followed are, candidly, almost impossible to write without sounding like a finance influencer exaggerating.
According to data compiled by Gregory Zuckerman in his 2019 book The Man Who Solved the Market, and corroborated by Cornell Capital Group, Institutional Investor, and multiple independent academic sources:
- The Medallion Fund delivered an average annual gross return of approximately 66% from 1988 to 2018 — a thirty-year window.
- After fees (which were extraordinary by industry standards — 5% fixed plus 44% performance, vs. the industry standard 2% and 20%), investors still earned approximately 39% net per year.
- $100 invested in Medallion in 1988 became approximately $2.1 million by 2018, net of all fees.
- The same $100 invested in the S&P 500 over the same window became approximately $1,900.
- The Medallion Fund has produced over $100 billion in cumulative profits.
- In 2008, while the S&P 500 lost more than 38%, Medallion almost doubled.
- From 2001 to 2013, its lowest annual return was +21%.
By comparison: Warren Buffett’s Berkshire Hathaway has compounded at roughly 20% annually over its entire history. That is itself a generational performance. Medallion’s net returns are roughly double Buffett’s, sustained over thirty years.
Zuckerman’s verdict, in The Man Who Solved the Market: Buffett, Soros, Lynch, Cohen, Dalio — all of them fall short.
By any defensible measure, Simons is the most financially successful trader in modern financial history.
He did it as a mathematician, with mathematicians, refusing to hire anyone who came from the financial industry.
The hiring philosophy
This is the part that most retellings of the Simons story underweight, and it’s the part that matters most for what comes next.
Renaissance Technologies has approximately 310 employees. Roughly 90 of them — a third of the entire firm — hold PhDs in mathematics, physics, computer science, astronomy, or other quantitative scientific disciplines. The firm explicitly avoids hiring people from Wall Street.
Peter Brown, the current CEO, came from IBM. He was, before joining Renaissance, working on speech recognition with Jeff Hinton — the man later called “the godfather of AI.” Brown joined Renaissance in 1993, along with his IBM colleague Bob Mercer, after Simons doubled their salaries to lure them away. Mercer would eventually become co-CEO before Brown.
Brown’s articulation of the firm’s philosophy, in a 2023 Goldman Sachs podcast, was direct: “The company was founded by scientists, is owned by scientists, run by scientists, and we employ scientists.”
The standard objection from finance professionals — that mathematicians don’t understand markets and would therefore be terrible traders — Renaissance has inverted from the beginning. Their working assumption: it is much easier to teach a brilliant mathematician how markets work than to teach mathematics to someone whose career has been built on market intuition.
This is the foundational disagreement between Renaissance and almost every other financial institution. Wall Street believes the rare and valuable input is market judgment — the accumulated intuition of someone who has been in the room for thirty years. Renaissance believes the rare and valuable input is mathematical pattern recognition — the capacity to find statistically robust signals in noisy data.
The thirty-year compounded results suggest who was right.
The methodology
The Renaissance approach, stripped to its essentials, contains six principles. None of them are secret. All of them have been described, in public, by Simons or his successors over the past two decades.
One: data first, models second. Simons described the firm’s philosophy in roughly these terms: we don’t start with theories about how markets work. We start with data. We look for patterns. We test whether those patterns replicate. If they replicate thousands of times, we trade on them.
Two: find small statistical edges and compound them across millions of decisions. Most investment strategies look for big asymmetric bets. Renaissance does the opposite. The Medallion Fund executes 150,000 to 300,000 trades daily. The mathematics of compounding small edges across that many decisions is the actual source of the returns.
Three: automate execution and eliminate emotional bias. Every Medallion trade is executed by automated systems. No human discretion. No emotional override. The way to eliminate that source of error is to eliminate the option to override the system.
Four: hire scientists, distrust domain experts. Renaissance bets that verifiable claims compound and unverifiable claims don’t. A scientist’s claims are open to verification. A traditional financial expert’s claims are often grounded in pattern recognition built over years of experience — which sounds valuable but is structurally hard to falsify.
Five: trust the math even when it disagrees with the room. Renaissance built thirty years of compounded outperformance on the practice of doing the reverse of what most enterprises do: when the senior person’s instinct disagrees with the math, the math wins by default.
Six: take the math seriously enough to act on it, even when you can’t fully explain it. Renaissance trades on signals it can’t always explain in human-comprehensible terms. The mathematics says the signal is real. The pattern replicates. Wall Street has historically refused to act on signals it can’t tell a story about. Renaissance acts on them.
Why this matters in 2026
For forty years, in the narrow domain of financial markets, Simons and his team were doing — at small organizational scale — exactly what every serious AI-native company is now doing, at massive scale, across every domain that has data.
Look at the six principles above and translate them out of finance.
Start with data, not theories. This is the foundational principle of every modern machine learning system. You don’t tell the model how to recognize a cat. You show it ten million images and let it find the pattern itself.
Find small statistical edges and compound them. This is the foundational principle of every recommendation system, every search ranker, every personalization engine in the modern technology stack. Tiny improvements in click-through rates, compounded across billions of decisions, produce trillion-dollar companies.
Automate execution, eliminate emotional bias. This is the foundational principle of every automated decision system from supply chain optimization to algorithmic credit underwriting.
Hire scientists, distrust domain experts. This is the foundational principle of every AI-first company that has ever competed against a legacy incumbent. The incumbent has the domain experts. The AI-first competitor has the data scientists. The latter wins, structurally, over a long enough time horizon.
Trust the math even when it disagrees with the room. This is the foundational principle of every enterprise that has successfully deployed AI at scale. The hardest decision is the moment when the model disagrees with the senior executive’s intuition.
Take the math seriously when you can’t fully explain it. This is now the central interpretability debate in modern AI. We act on the outputs of systems whose internal reasoning we cannot fully describe. Renaissance has been doing this since the late 1980s.
The Medallion Fund is the longest-running, best-documented case study in the world for what an organization built around these six principles looks like over a thirty-year horizon.
It works.
It works very well.
It works in defiance of almost every organizational instinct of the people who weren’t running it.
What enterprises are still getting wrong
In 2026, almost every major enterprise will tell you it is undergoing an AI transformation. Almost every senior leader I talk to has rolled out a strategy deck about it.
Almost none of them have done what Simons did.
Look at how a typical enterprise is still organized in 2026.
Senior strategic roles are held by people whose career credentials are MBA-shaped, not data-shaped. Most boards of directors have zero members with deep technical or quantitative training.
Decision-making at the executive committee level is driven by judgment, with data brought in as supporting evidence. When the data and the judgment disagree, the judgment usually wins. The data is reframed as “directional but not conclusive.”
Human discretion is preserved as the mark of seniority. A senior leader who can’t override the recommendation engine isn’t really senior. The status of the role is bound up with the latitude to disagree with the system.
Big transformational strategic bets are favored over compounding small edges. The CEO who launches the bold initiative gets the magazine cover. The CEO who improves a thousand small decisions by 3% each gets no magazine cover, even though that CEO is structurally producing more value.
Domain experts are protected as carriers of irreplaceable judgment. Twenty-year veterans of the industry are held up as the people who really understand the business. The thirty-year-old data scientist who points out that the twenty-year veteran’s “instinct” is actually a set of statistically unjustified anchoring biases is, in most enterprises, not the one whose career gets accelerated.
This is the 1980s Wall Street model, applied to 2026 enterprises.
It is the model Simons spent forty years quietly destroying, with thirty-year compounded returns to prove it.
And it is the model that almost no enterprise has yet been willing to dismantle, even when their CEOs talk publicly about being “AI-first.”
The executive leadership translation
If you are a senior leader in 2026, the Medallion Fund forces a specific, uncomfortable question:
What part of my organization is currently structured around someone’s accumulated intuition rather than someone’s verifiable pattern recognition?
The honest answer is: most of it.
Strategy. Hiring. Capital allocation. Pricing. Product roadmap. M&A. Marketing creative direction. Sales territory design. Executive succession.
In almost every enterprise, almost all of these are still decided by senior leaders’ professional judgment, with data used to confirm or season the conclusions rather than to drive them.
The Medallion principle would invert this. Wherever the math is good enough, the decision belongs to the math. Human judgment is reserved for the cases where the math is genuinely uncertain.
This is, in my experience working with executives, the hardest organizational shift any senior team can undertake. Not because the technical part is hard. But because the identity part is hard.
If your value as a senior leader has been built on intuition, experience, taste, and “feel” — and the company you work for is now telling you that these qualities are increasingly being outperformed by quantitative systems — what exactly is your value going forward?
This is the unspoken question underneath every enterprise AI transformation. It is rarely asked out loud, because asking it out loud is professionally dangerous.
But it is the question Simons answered in 1988 by hiring mathematicians instead of traders. And it is the question every enterprise is now being forced to answer.
What to actually do
Audit your decision rights. For the top twenty decisions your organization makes regularly, ask honestly: who owns this decision? Is it owned by someone whose value comes from accumulated intuition, or by someone whose value comes from verifiable pattern recognition?
Hire scientists into senior roles, not just analyst roles. Renaissance’s structural insight was that the scientists are the senior leadership, not their support staff.
Build compounding small edges, not bet-the-company moves. Stop looking for the bold transformational AI bet. Look for the hundred small decisions your organization makes every day where AI can improve performance by 3%.
Defend the data when it disagrees with the room. When a model output disagrees with a senior executive’s instinct, the default in most enterprises is to “investigate further” — which is usually a polite way of overriding the data while preserving the executive’s authority. The Medallion discipline is the opposite.
Accept that some of your senior leaders won’t make the transition. Renaissance’s structural advantage was that it never had to retrain a Wall Street veteran into a scientist. It hired scientists from the beginning. Most enterprises are trying to retrain leaders whose entire professional identity is bound up with intuition-based decision-making.
The closing thought
There is a sentence Simons used in interviews that has stayed with me as the cleanest articulation of his philosophy.
“We don’t start with models. We start with data.”
Ten words.
It is the entire intellectual move that produced the most successful investment fund in modern financial history. It is the entire intellectual move that powers every modern AI system. And it is the entire intellectual move that almost no traditional enterprise has been willing to make at the senior level.
Most enterprises in 2026 still start with models — strategy frameworks, leadership theories, executive intuitions about how the world works. Then they look for data to support the model. When the data contradicts the model, the model usually wins.
The Medallion Fund — and increasingly, every successful AI-native organization — starts with the data. The models come later, derived from what the data actually shows, not from what someone in the room assumed it should show.
Simons died as the richest mathematician in history.
He didn’t get there by being the smartest mathematician.
He got there by being willing to build an organization that took mathematics seriously enough to act on it — and by doing it forty years before everyone else realized they would have to.
The enterprises that learn from him will be the ones the next decade is built around.
The ones that admire him from a distance, write internal memos quoting his returns, and then go back to running themselves on senior intuition will be the ones the next decade quietly replaces.
The world has changed. The leaders who notice will be the ones the next decade is built around.
