Jim Simons: When Mathematics Went to Wall Street

Jim Simons built the most profitable trading operation in history without hiring anyone from Wall Street. His Medallion Fund averaged about 66% a year before fees from 1988 to 2018, according to Gregory Zuckerman's account of the firm. Simons, who died in 2024, got there by treating markets as a data problem.

A mathematician first

Simons was a codebreaker for the US government during the Cold War and then chaired the mathematics department at Stony Brook University. His work with Shiing-Shen Chern produced Chern-Simons theory, which physicists still use. He was already eminent before he placed a trade.

He left academia in 1978 to trade currencies and founded Renaissance Technologies in 1982. His early trading mixed models with intuition, and he found the emotional swings unbearable. He decided the models should make every decision.

Hiring scientists, not traders

Simons recruited mathematicians, physicists, astronomers and speech-recognition experts. Leonard Baum, James Ax and Elwyn Berlekamp shaped the early systems. Peter Brown and Robert Mercer, who arrived from IBM's speech research group, later ran the firm.

His reasoning was simple. He could teach a scientist about markets, but he could not teach a trader to do science.

How Medallion works, as far as anyone knows

Renaissance guards its methods closely, but the outline is public:

• Vast, clean data. The firm collected and corrected price data going back centuries, long before rivals bothered.
• Small, repeatable edges. The models find faint patterns that are right only slightly more than half the time.
• Enormous trade counts. A tiny edge repeated across thousands of trades becomes a reliable return.
• Short holding periods. Most positions last days or less.
• One model. All signals feed a single system, so every researcher's work improves everyone else's.

The models do not need to explain why a pattern exists. If it is statistically sound and persists, it trades.

Knowing the limits

Medallion stopped accepting outside money in 1993 and has been owned by employees since 2005. Its strategies only work up to a certain size, so the fund pays out profits each year instead of growing.

Renaissance's larger public funds use slower strategies and have produced far more ordinary results. The magic did not scale, and Simons never claimed it would.

The human parts

Simons's own contribution was management. He chose the people, kept them collaborating, paid them well and shared ownership widely. He also overrode the models on rare occasions, such as the quant crisis of August 2007, when he cut risk.

What to take from it

• A small edge is enough if it is real and repeated often.
• Data quality comes before model cleverness.
• Remove discretion where emotion does damage.
• Every strategy has a capacity. Respect it.

Part 8 of 15 in the series Lessons From the Masters. Next: Peter Lynch: Finding Investments Hiding in Plain Sight.

Independent educational commentary, not investment advice. References to people and firms do not imply affiliation or endorsement. The picture is an illustration, not a photograph. Past results say nothing about future ones. Trading leveraged products carries a high risk of loss.