What Jim Simons Taught Us About Investment: Lessons from the Mathematician Who Beat Wall Street

Jim Simons died in May 2024 at 86. His Medallion Fund generated 66% average annual returns over 34 years—the best track record in investment history. Not 66% total. 66% per year.

That number still doesn’t make sense to me. $100 invested in 1988 turned into $398.7 million by 2018 after fees. The same $100 in the S&P 500 would’ve grown to roughly $2,000.

But here’s what gets me: Simons wasn’t a finance guy. He was a mathematician who won the Oswald Veblen Prize in 1976, one of the highest honors in geometry. He left academia at 40, applied code-breaking techniques to markets, and built something that shouldn’t exist according to efficient market theory.

I’ve spent the past week reading everything I could find about Simons. Not because I think anyone can replicate Medallion’s returns—they can’t. The fund is closed to outside investors and employs 90 PhDs working on a single unified model that took decades to build.

I wanted to understand what his career teaches us about investing. What can someone running a traditional portfolio learn from a mathematician who treated markets like cipher codes?

Turns out, quite a bit.

Simons hired physicists, astronomers, and speech recognition experts. He actively avoided MBAs and Wall Street professionals. When the firm needed a breakthrough in 1993, he recruited Robert Mercer and Peter Brown from IBM’s speech recognition group—people who’d never traded a stock.

This wasn’t contrarian hiring for the sake of being different. Simons believed markets were pattern recognition problems, not economic forecasting exercises. He wanted people who could prove theorems, not people who could explain why the Fed’s next move mattered.

The critical hire was Leonard Baum in the early 1980s, creator of the Baum-Welch algorithm used in speech recognition. Baum brought hidden Markov models to finance—mathematical tools designed to find patterns in noisy data. That’s exactly what price movements are.

Here’s what I think matters for investors: Simons didn’t try to understand why markets moved. He looked for what actually happened in the data. No macro forecasts. No earnings projections. No views on whether a company was well-managed.

Renaissance processed terabytes of data daily, searching for anomalies that repeated. As Simons put it: “We don’t start with models. We start with data. We look for things that can be replicated thousands of times.”

Robert Mercer once admitted he didn’t know Chrysler no longer existed as a standalone company—the model just signaled when to buy or sell. That’s pure quant investing. No fundamental research required.

I’m not suggesting fundamental analysis is worthless. I use it. But Simons proved there’s another way to generate alpha: find statistical edges that repeat, execute them systematically, and don’t let narrative override data.

Simons repeated this in every interview: his biggest contribution wasn’t the math, it was hiring great people.

Renaissance employed 300-400 people total versus 2,000-5,000 at competitors. Roughly 90 PhDs among 150-200 research staff. Median tenure ran 14-16 years compared to 2-3 years typical in finance.

The compensation structure was unusual: 5% management fee and 44% performance fee, versus the typical 2% and 20%. Yes, Renaissance charged more. But employees got a massive cut of that 44%. The firm’s 401(k) plans could invest directly in Medallion through a special IRS exemption. Combined with lifetime NDAs and non-compete agreements, leaving meant walking away from extraordinary wealth.

Simons also insisted everyone see what everyone else was working on. In 1995, when Mercer and Brown implemented the “one model” system, all researchers began working on a single unified model across all asset classes. Discoveries in currency markets benefited equity trading. Breakthroughs in commodities improved fixed income strategies.

This goes against how most hedge funds operate. Typical structure: portfolio managers compete against each other, guard their best ideas, and get fired if they underperform.

Renaissance did the opposite. Complete transparency. Collaborative research. Long-term incentives. Pay people so well they’d never leave.

The lesson here isn’t “hire PhDs” or “pay 44% performance fees.” Most of us can’t do either. But the principle holds: if you can afford to work with people better than you at specific tasks—research, risk management, execution—that’s probably your highest-return investment.

I’ve seen investors waste money on Bloomberg terminals they barely use while refusing to pay for quality research or proper tax advice. That’s backwards. Simons built the world’s most profitable fund by surrounding himself with people who knew things he didn’t.

During the March 2020 selloff, markets dropped 34% in 23 days—the fastest bear market in history. Most funds got crushed. Medallion made money.

How? I don’t know the specifics because Renaissance doesn’t publish details. But I know their philosophy: they don’t predict market direction. They identify patterns that statistically repeat over thousands of trades.

When Medallion returned 152% gross (82% net) during the 2008 financial crisis, they weren’t predicting Lehman’s collapse. When they delivered 128% gross during the 2000 tech bubble, they weren’t forecasting which dotcoms would fail.

They were exploiting short-term anomalies—price discrepancies that existed for hours or days, not months or years. High-frequency statistical arbitrage across thousands of positions. Win rate probably wasn’t that high on individual trades. But with enough trades and proper position sizing, the law of large numbers takes over.

Here’s where this gets relevant for traditional investors: you don’t need to predict the future to make money. You need repeatable processes that work more often than they fail.

I think about this with my own investing. I don’t know if we’re headed into recession. I don’t know if inflation will spike or crater. I don’t know if the Fed will cut rates three times or zero times this year.

What I do know: quality companies with pricing power and low debt tend to outperform during volatility. Rebalancing a diversified portfolio back to target weights forces you to buy low and sell high. Dollar-cost averaging removes timing risk.

None of that requires predicting anything. It’s just pattern recognition—strategies that have worked across multiple market cycles.

Simons took this to an extreme with pure quant methods. Most investors can’t do that. But we can stop pretending we need to forecast the future and instead focus on what actually repeats.

In the early 1970s, Simons collaborated with mathematician Shiing-Shen Chern on what became Chern-Simons theory. They were trying to find a combinatorial formula for something called the first Pontryagin class. They failed at their original goal.

But the work they did instead—creating secondary characteristic classes called Chern-Simons forms—turned out to be fundamental to modern theoretical physics. When Edward Witten showed in 1988 that this mathematics described topological quantum field theory, it became central to string theory, knot theory, and quantum computing.

Simons said: “We didn’t know any physics. It never occurred to me that it would be applied to physics. But that’s the thing about mathematics—you never know where it’s going to go.”

He carried this philosophy into investing. Renaissance’s approach came from code-breaking, speech recognition, and signal processing—fields completely outside traditional finance. The idea that markets could be treated like encrypted messages or noisy audio data wasn’t obvious in 1978.

Here’s what I take from this: elegant solutions often come from asking different questions than everyone else.

When Simons looked at markets, he didn’t ask “What’s this company worth?” or “Where’s the economy headed?” He asked “What patterns exist in price data that repeat with statistical significance?”

That’s a fundamentally different question than what traditional investors ask. It led to fundamentally different answers.

I’m not suggesting everyone should become a quant. But I do think there’s value in questioning assumptions. Why do we analyze companies the way we do? Because it works? Or because that’s how everyone’s always done it?

The investors who’ve generated exceptional long-term returns—Simons with quant, Buffett with value, Dalio with macro—all asked different questions than their peers. They found approaches that matched their skills and temperament.

Simons left academia at 40 after winning the Veblen Prize. The math community thought he’d sold out. One colleague said it was like “selling his soul to the devil.”

But Simons had scratched his itch in mathematics. He’d done work that would influence physics for decades. He wanted to try something else. So he did.

Then after retiring as CEO of Renaissance in 2009, he returned to active mathematical research. At 71. He collaborated with Dennis Sullivan on differential K-theory and published multiple papers through 2024.

This is rare. Most people who leave a field don’t come back. Simons did, because he genuinely loved mathematics—not for career advancement, but for its own sake.

I think about this with investing. How many people are doing it because they love it versus because they think they should? How many are following strategies that don’t fit their personality because some expert said it’s “best practice”?

Simons succeeded in investing partly because he approached it like mathematics: pattern recognition, rigorous testing, collaboration with smart people, willingness to be wrong. Those were his strengths.

If you’re naturally good at fundamental analysis and enjoy reading 10-Ks, lean into that. If you’re quantitatively minded and prefer data to narratives, build systematic approaches. If you’re patient and disciplined, maybe passive indexing fits better than active trading.

The worst investment strategy is one you can’t stick with. Simons found an approach that matched his skills and temperament. That’s why it worked for 34 years.

Simons and his wife Marilyn gave away approximately $6 billion during their lifetimes. The Simons Foundation holds over $5 billion in assets and distributes roughly $450 million annually in grants.

The June 2023 gift of $500 million to Stony Brook University was the largest unrestricted gift to any American university in history. Math for America supports roughly 1,000 STEM teachers annually in NYC public schools. The Simons Foundation Autism Research Initiative has identified approximately 100 genes related to autism versus one when they started.

Here’s what gets me: Simons could’ve kept it all. He’d earned it. 66% annual returns over 34 years isn’t luck—it’s skill applied systematically over decades.

Instead, he redirected his fortune toward basic scientific research. Not applied research with obvious commercial applications. Basic research—the kind that asks fundamental questions about the universe without knowing where it’ll lead.

Just like his mathematical work on Chern-Simons theory had no obvious application until decades later, Simons funded research trusting that good science takes you places you don’t expect.

His own summary shortly before death: “I did a lot of math, I made a lot of money, and I gave almost all of it away.”

I’m not here to tell anyone what to do with their money. But I do think Simons’ approach to philanthropy reflected the same philosophy that made him successful in mathematics and investing: be guided by beauty, surround yourself with smart people, trust that you don’t know where good work will lead.

He wasn’t trying to optimize tax deductions or build a legacy. He funded what he found beautiful and important.

Jim Simons proved that markets aren’t perfectly efficient. They’re discoverable. Patterns exist. Data matters more than narrative.

But I think the deeper lesson is this: success comes from finding approaches that match your strengths, building teams smarter than you, and having the conviction to ignore conventional wisdom when you’ve got a better answer.

Simons left academia when colleagues thought he was crazy. He hired scientists instead of traders when Wall Street thought he was naive. He charged 5-and-44 when the standard was 2-and-20. He gave away billions to basic research when others were buying yachts and sports teams.

Every decision was contrarian. Every decision was right.

I won’t achieve 66% annual returns. Nobody will. Medallion’s performance was a unique combination of brilliant people, proprietary data, decades of model refinement, and probably some luck in finding anomalies before they got arbitraged away.

But I can ask better questions. I can focus on process over prediction. I can work with people smarter than me. I can stick with approaches that fit my temperament.

That’s what Jim Simons taught me about investing. Not the math—I’ll never understand the Baum-Welch algorithm or hidden Markov models. The philosophy: be rigorous, be systematic, be humble enough to learn from data, and be bold enough to ignore everyone when you know you’re right.

Sixty-six percent per year for 34 years. That’s the exclamation point on one hell of an argument.