Jim Simons: how a mathematician built the world's best trading machine

Executive overview

Most investors try to understand why markets move. Simons ignored the why entirely and built a system that just needed to be right 51% of the time. After a decade of failed partnerships and mounting losses, his Medallion Fund achieved average annual returns of 66% from 1988 onward — never recording a down year.

The core insight: markets are a pattern-recognition problem, not an economic one — and humans are the noise to be modeled, not the analysts to do the modeling.

Early life and the persistent pull toward wealth

• At 14, Simons declared he would study math at MIT; people laughed; it didn't register.
• From his father he took one lesson: do what you love, not what's expected.
• His first wife noted: "Jim understood at an early age that money is power. He did not want people to have power over him."
• While earning his PhD at Berkeley, he drove to San Francisco each morning to trade commodities at Merrill Lynch before class.
• He described himself as always feeling like an outsider — a self-image he leaned into rather than shed.

The IDA years and the origins of the method

• At 26 he joined the Institute for Defense Analysis to break Soviet codes — earning more than academia paid, with half his time free for his own research.
• There he developed his thinking style: lying still for hours, eyes closed, grinding through problems in silence.
• He and colleagues wrote an early paper asking whether markets could be predicted without economic theory — no interest in why, only in what the patterns suggested.
• He was fired at 29 for publicly opposing the Vietnam War; he later called it "a good thing — you just don't want to make a habit of it."

Building Stony Brook's math department (ages 30–40)

• Hired to build a world-class math department, Simons learned to recruit and manage elite, difficult personalities.
• He defined the people he wanted as killers: "those with a single-minded focus who wouldn't quit."
• He coined a distinction he repeated for decades: "There are guys, and then there are real guys. You want the real ones."
• He spent, by his own description, "a lot of time courting talent" — treating recruitment as a core leadership function.
• His five guiding principles, published in 2020, grew from habits formed here: above all, surround yourself with the smartest people you can find.

Launching the firm and a decade of struggle (ages 40–50)

• Left academia at 40 to trade currencies; his father called it a mistake; his mathematician peers thought he had "sold his soul."
• Guiding principle one: "Do something new. Don't run with the pack. If I am one of N people working on the same problem, there is very little chance I will win."
• He and successive partners built partially automated systems but kept abandoning them when intuition-based trading was making money.
• Partner Leonard Baum racked up $43 million in profits trading by instinct — then lost it all, triggering a forced break-up clause.
• A second partner, James Axe, resisted Simons's push toward short-term algorithmic trading; their partnership also collapsed.
• By 1989, the fund was down $20 million and colleagues thought Simons might shut it all down.

The breakthrough: shorter holds and the casino model

• Elwyn Berlekamp, brought in to help, analyzed which trades had actually won — and found the short-term ones dominated.
• His insight: "Buying and selling infrequently magnifies the consequence of each move. Make a lot of trades and each individual move is less important."
• Target: resemble a casino — handle so many bets that a 51% win rate, compounded at scale, becomes overwhelming.
• Medallion cut average holding time to a day and a half. Results were "almost immediate, startling nearly everyone in the office."
• Medallion scored a 55.9% gain in 1990 after a 4% loss the prior year; daily million-dollar profits became routine.
• Even then, outsiders mocked the methods and called the team "quacks."

Conviction as the differentiating asset

• Berlekamp sold his stake during the 55.9% year, believing 30% annual returns were the ceiling. Simons thought 80% was possible.
• Simons bought him out; Berlekamp left believing he had won a six-times return in 16 months.
• Pattern across all departing partners: none of them believed what they had. Simons had belief and conviction; they had intelligence only.
• Simons to a friend after his last partner left: "To hell with it. I'm just going to run this myself."

Data as structural advantage

• From the beginning, Simons collected more historical price data than any competitor — buying World Bank books, reels of magnetic tape, sending staff to the Federal Reserve by hand.
• Sandor Strauss built and cleaned the database: "more accurate data than anyone else — a massive advantage."
• Some weekly stock data stretched back to the 1800s.
• Later: "We take in terabytes of data every day. We store it, massage it, and get it ready for analysis. You're looking for anomalies."
• A data-entry error once caused Medallion to buy five times its intended wheat futures position, moving the market. The Wall Street Journal attributed the price surge to fears of a poor harvest. Peter Brown (Renaissance CEO): "Anytime you hear financial experts talking about how the market went up because of such and such — remember it is all nonsense."

System design and incentive architecture

• Medallion used a single monolithic trading system; every employee had full access to every line of source code.
• Simons insisted: "You need to make everyone partners. Share all the profits."
• By 2002, fees rose to 30% of profits; later 44%. Investors asked how they could put in more.
• In 2003, he kicked all outside investors out of the fund. Only Simons and employees could hold Medallion shares.
• If you left Renaissance, you lost access to Medallion — eliminating the incentive to leave.
• Result: near-zero employee turnover. Knowledge compounded uninterrupted.
• Operating with ~300 employees, the fund generated $5–7 billion annually in its peak years.

How they thought about markets

• "What you're really modeling is human behavior. Humans are most predictable in times of high stress. They act instinctively and they panic."
• "Our entire premise was that human actors will react the same way humans did in the past. We learned to take advantage."
• They hired from IBM's computational linguistics department: speech recognition and stock prediction share the same structure — digesting uncertain jumbles of information to generate reliable guesses about what comes next.
• They deleted company names from stock data and replaced them with numbers, to prevent human intuition from overriding the model.
• Simons on not understanding why the model made its decisions: "I don't know why planets orbit the sun. That doesn't mean I can't predict them."

Secrecy as competitive moat

• "Visibility invites competition. With all due respect to the principles of free enterprise, the less competition, the better. Our only defense is to keep a low profile."
• Simons quoting Benjamin the donkey from Animal Farm: "God gave me a tail to keep off the flies. But I'd rather have no tail and no flies."
• Renaissance never hired from the financial industry: "We never did because they don't have anything to add."

Jim's five guiding principles

1. Do something new. Don't run with the pack.
2. Surround yourself with the smartest people you can find.
3. Be guided by beauty — in math, in poetry, in an organization running well.
4. Don't give up easily. Some things take much longer than initially expected.
5. Hope for good luck.

Final advice Simons said he would give his 20-year-old self: "It is very important to enjoy your work. Find something you love and then put your heart and soul into it."