5 market lessons from the greatest trader ever

Every hedge fund, investment bank and kid with a Robinhood account now thinks artificial intelligence is going to make them rich. They’re running machine-learning models on petabytes of data, back-testing strategies on supercomputers, and using language that makes them sound like they’re launching the Mars rover instead of buying shares of Nvidia NVDA.

They’re about to learn what hedge-fund investor Jim Simons proved in 1988: Having the same tools as a genius doesn’t make you a genius.

When he died in May 2024, Simons was worth $31.4 billion, having extracted more money from markets than any human in history. Gregory Zuckerman’s 2019 bestseller, “The Man Who Solved the Market: How Jim Simons Launched the Quant Revolution,” revealed how: Simons’s Medallion Fund averaged 66% annual returns for three decades using mathematical algorithms that would make ChatGPT weep.

Warren Buffett, who built Berkshire Hathaway BRK.A BRK.B into a $900 billion empire by understanding businesses better than the people running them did, has averaged 20%. Both are considered geniuses. Simons just found patterns no human could see.

Simons’s Renaissance Technologies was doing AI trading before anyone called it AI. He started in 1988, back when “machine learning” meant teaching your VCR to record “Cheers.” He hired astrophysicists, cryptographers and speech-recognition experts — no Wall Street experience allowed. They built models so sophisticated their own creators couldn’t explain what they were doing. Japanese typhoons triggered Mexican peso USDMXN trades. Sunspot activity correlated with pork-belly futures. It made no sense. It made billions.

Here’s the part that should terrify everyone buying “AI-powered trading systems” in 2025: Even now, with every firm on Wall Street running the same technology, Medallion still beats them. This is what happens when a mathematician discovers that Wall Street is just Las Vegas with Bloomberg terminals, and then spends 30 years figuring out how to be the house.

The tools are available. The talent is available. The data are available. And nobody else can do it.

In 2008, when Lehman Brothers learned that lending money to dead people had consequences and the S&P 500 SPX cratered 37%, Medallion made 82%. Everyone else was googling “bankruptcy lawyer.” Simons was taking delivery of “Archimedes,” a $100 million, 222-foot superyacht named after the Greek mathematician who discovered why things float.

While Wall Street drowned, Simons commissioned a floating monument to buoyancy. You can’t make this up.

How did Simons afford it? By charging fees that would make a loan shark blush. Simons charged investors 5% management plus 44% of profits — and billionaires who negotiate prenups that would make Machiavelli weep begged to get in.

Here’s the real joke: Simons was an antiwar mathematician who thought capitalism was morally dubious. He built history’s greatest money machine by accident. Simons was trying to solve a puzzle — the billions were just what fell out when he cracked it.

Which brings us to why a six-year-old book matters more in 2025 than it did in 2019. Back then, algorithmic trading was a curiosity. Now, it’s a religion. And like most religions, it’s going to disappoint a lot of believers.

James Harris Simons was born in 1938 to the manager of a shoe factory in Brookline, Mass. His parents weren’t wealthy, but they were ambitious for their son. Young Jim showed an early gift for mathematics that bordered on savant territory. He was doing complex calculations before most kids master long division.

By 23, he had his Ph.D. from the University of California, Berkeley. By 26, he was breaking Russian codes for the NSA during the Cold War. This wasn’t your grandfather’s intelligence work.

Simons was part of an elite unit at the Institute for Defense Analyses, using mathematical models to find patterns in encrypted Soviet communications. Think of it as the 1960s equivalent of what Edward Snowden was doing, except legal and aimed at the bad guys.

Here’s where it gets beautiful: In 1968, the government fired Simons for writing a letter to the New York Times saying the Vietnam War was a bad idea. Which it was, but the government doesn’t like being told it’s wrong.

Most people would have crawled to academia, grateful for tenure and dental insurance. Simons went to Stony Brook University in New York, built one of the world’s best math departments, won the Oswald Veblen Prize — one of the highest honors in mathematics — and then walked away from it all to become, essentially, a professional gambler.

While breaking Russian codes, Simons had realized something. The stock market was just another encrypted transmission — meaningless statics hiding meaningful patterns. When you cracked this code, instead of discovering troop movements, you discovered money movements. And you could intercept them.

In 1978, at age 40, Simons founded Renaissance Technologies in a strip mall on Long Island. Not Manhattan. Not Greenwich. A strip mall. His first office was above a women’s clothing store. This tells you everything about his approach. He wanted to be as far from Wall Street’s groupthink as possible.

His hiring strategy was deliberate sabotage against conventional wisdom. One rule: No Wall Street experience. “I wanted people who would look at data without preconceived notions,” he said.

His first trades were funded by a $5,000 wedding gift, equivalent to about $25,000 today. Not a trust fund, not an inheritance — the same seed money anyone might scrape together. Simons turned it into more money than most countries have.

The early years were brutal. He made a fortune in soybeans S00, then lost it all. Currency trading worked until it didn’t. Bond futures nearly bankrupted them. By 1984, Renaissance was barely surviving.

Most people would have returned to teaching calculus to hungover sophomores, but Simons doubled down. Except the chips were algorithms, and the dealer was mathematics.

The breakthrough came when Simons started hiring codebreakers instead of traders.

Leonard Baum, a mathematician who helped develop hidden Markov models — algorithms that find patterns in seemingly random sequences — started collecting everything. Not just stock prices but weather patterns, shipping rates, political polls, satellite imagery. Renaissance was building a data advantage before “big data” existed.

Elwyn Berlekamp, who’d worked with Claude Shannon (the father of information theory), brought one crucial insight: Stop trying to be Warren Buffett. They weren’t value investors finding underpriced companies. They were running a casino. Casinos don’t need to know why the roulette ball lands on red. They just need to know that the house wins 52.7% of the time.

The real breakthrough came from former NSA cryptographer Nick Patterson and researcher Peter Laufer. They started analyzing market data in five-minute intervals instead of daily closes. This revealed short-term correlations invisible to traditional investors.

Some patterns made intuitive sense — commodities affecting related currencies, for instance. But many were purely statistical relationships with no obvious causation. Renaissance didn’t need to understand why correlations existed. They needed to know three things: that the correlation was statistically significant, that it persisted over time, and that it was profitable more often than not.

The edge wasn’t predicting correctly 100% of the time. It was being right 50.75% of the time, on enough trades that the edge compounded into billions of dollars.

The fund was capped at $10 billion because beyond that size, the strategies would move markets too much — destroying the very inefficiencies they exploited.

In 1993, Simons made perhaps his smartest hires: Peter Brown and Robert Mercer from IBM IBM. These weren’t finance guys. They were language-recognition experts who’d spent their careers teaching computers to understand drunk people ordering pizza over the phone.

Simons realized finding profitable signals in market noise was exactly like finding words in static. Same problem, different stakes.

Brown and Mercer built algorithms that evolved and adapted. These systems recognized “market states” the way their IBM computers recognized speech patterns.

By 2000, the system had become so sophisticated that its creators couldn’t explain what it was doing. Half of the trading signals made no sense. (Cue Japanese typhoons triggering Mexican peso trades.)

By 2008, every hedge fund on earth had decided to become a Renaissance clone. Two Sigma, D.E. Shaw, Citadel — all of them hired physicists, bought supercomputers and started hunting for meaningless correlations that somehow made money.

And it worked — sort of. These quant funds now dominate trading volume. They’re finding exactly the kind of inexplicable patterns Simons pioneered. Butterfly migrations affecting copper futures HG00. Solar flares predicting semiconductor stocks. The kind of signals that make fundamental analysts reach for whiskey.

Here’s the problem: When everyone’s hunting the same prey with the same weapons, the prey gets scarce.

The genius of Renaissance wasn’t discovering that Japanese rainfall predicted the value of Mexican pesos. It was discovering it first and keeping it secret long enough to extract a haul before anyone else noticed. In today’s market, with every firm running similar models on similar data, inefficiencies vanish like free beer at a fraternity party.

The modern quant funds post good numbers — 15%, 20% a year, sometimes better. Respectable. But Medallion averaged 21% annually over the past decade. Not the 66% of its golden age, but still obscene by normal standards. It’s like watching Tiger Woods shoot his age at 50. Not peak Tiger, but still Tiger.

But here’s where it gets weird: Even Renaissance can’t replicate Renaissance at scale.

The firm’s two funds availble to outside investors — Renaissance Institutional Equities Fund (RIEF) and Renaissance Institutional Diversified Alpha (RIDA) — have both struggled. In 2020, while Medallion returned 76%, RIEF lost 22.6%. Same firm, same management team, different results.

The difference: Medallion trades $10 billion with perfect employee alignment. The institutional funds trade larger amounts with outside capital and can’t execute the same strategies.

Renaissance continues under Peter Brown’s leadership — still closed to outsiders, still profitable, but acknowledging what everyone in quant knows: The market got smarter. AI democratized signal discovery. What used to be Renaissance’s private monopoly is now an industrywide capability.

The gap narrowed, but it never closed. Medallion remains the standard nobody has matched. The question isn’t whether AI can find Renaissance-style patterns — it can. The question is whether anyone can sustain that edge when everyone’s using the same tools.

So far, the answer is no.

Let’s be clear: You cannot replicate Renaissance’s strategy in your E-Trade account. They’re executing millions of trades on microsecond timescales using signals derived from petabytes of data. Their infrastructure costs run into the hundreds of millions of dollars annually.

But here are the lessons you can learn from Simons:

1. Ignore the narrative. Every market move gets explained after the fact. “Stocks fell on inflation fears.” “Gold rose on geopolitical tension.” It’s backward storytelling. Simons proved the narrative is nonsense. Trade probabilities, not stories.

2. Your edge is being different. Renaissance succeeded because they thought differently. They hired poets and physicists, not MBAs. What’s your equivalent? Maybe it’s investing in what you understand while everyone else chases hot tips. Maybe it’s buying bitcoin BTCUSD while Wall Street calls it a bubble — something Simons himself did, by the way.

3. Size matters (inversely). Medallion capped at $10 billion. Beyond that, the strategies stopped working. Individual investors have an advantage here: You can invest in opportunities too small for big funds to touch.

4. Compound quietly. Simons founded Renaissance in 1978 and spent the first decade losing money on soybeans and nearly going bankrupt. When Medallion launched in 1988 and started compounding at 66% annually, Simons didn’t celebrate by going on television, mostly because he was too busy hiring cryptographers to work in a Long Island strip mall. By the time Wall Street figured out who he was, Medallion was already closed to outside investors.

5. Respect the unknown. Renaissance never assumed their models were “true.” They assumed they were temporarily useful. This humility kept them from overleveraged disasters. Don’t confuse a bull market for brains.

More from Charlie Garcia:

The robots have won. The smart trade now is investing in companies that make them.

Think your bond funds and ETFs are safe investments? The credit market is lying to you.