Before he started investing, Jim Simons broke codes for the American government and chaired a university mathematics department. The fund he founded, Renaissance Technologies, made its trading decisions with statistical models run by computers, without relying on the judgment of traders. The sections below explain how this kind of strategy works and where it usually fails.
Who Jim Simons was
Simons was born in 1938 in Massachusetts. Between 1964 and 1968 he worked on codebreaking at the Institute for Defense Analyses, and from 1968 to 1978 he chaired the mathematics department at Stony Brook University. In geometry, his name is attached to the Chern–Simons forms, used today in physics.
In the late 1970s he began investing, at first partly by intuition, and soon moved to mathematical models. Renaissance hired mathematicians, physicists, astronomers and speech recognition researchers, people who were good at finding faint patterns in noisy data, and almost nobody from Wall Street. Simons died in 2024.
What quantitative finance is
Quantitative finance uses mathematics, statistics and programming to decide what to buy, what to sell and how much to risk. Instead of an opinion about a company, there is a written rule. If this happens in the data, do that. The rule is tested on years of past prices before it is used with real money.
Renaissance's Medallion fund, launched in 1988, is the best known result of this approach. According to Gregory Zuckerman's book, between 1988 and 2018 it averaged about 66% a year before fees and 39% after fees. These results belong to one fund with its own history and say little about what other strategies can earn. Medallion has been closed to outside investors for about two decades.
A small edge, repeated many times
According to the book, Renaissance was right only slightly more than half the time. The fund made money because it repeated that small edge over a very large number of trades and kept its costs low.
The chart shows why this works. Each line is a simulation of 50,000 bets of the same size. With exactly 50% right, the balance wanders around zero. With 50.75%, it climbs steadily. Over a few hundred bets, the two lines look alike.
Where the signals come from
A signal is any regularity in the data that helps predict the next move, even weakly. Some classic families are easy to explain.
- Trend (momentum)
assets that have been rising tend to keep rising for a while, and the same for falling.
- Mean reversion
prices that move too far from their usual level tend to come back.
- Relations between assets
when two assets that usually move together drift apart, the gap tends to close. This is the basis of pairs trading and statistical arbitrage.
- Calendar and flows
some effects repeat at certain hours, days or dates, often because of how other investors trade.
Renaissance spent years collecting and cleaning historical price data before testing signals, because errors in the data produce signals that do not exist in the market.
The biggest danger is seeing patterns that are not there
With enough attempts, any data set will reveal a "pattern" that worked perfectly in the past by pure chance. This is called overfitting, and it is the most common way quantitative strategies fail.
The usual defenses are to test the idea on data the model never saw, keep the rules as simple as possible, ask whether there is a reason the pattern should exist, and assume that real trading will be worse than the backtest once costs and delays are counted.
Words you will come across
- Alpha
- return that comes from the strategy itself and not from the market as a whole.
- Signal
- a measurable regularity that helps predict a price, even weakly.
- Backtest
- testing a strategy on past data as if it had been running at the time.
- Overfitting
- tuning a model so closely to the past that it fails on new data.
- Out of sample
- data that was kept aside and never used to build the model.
- Statistical arbitrage
- many small bets on prices that drift apart from their usual relation.
- Leverage
- investing more than your own money by borrowing, which multiplies both gains and losses.
- Sharpe ratio
- return divided by risk, a common way to compare strategies.
Where to start reading
The first group tells the story of Simons and presents the argument that markets are hard to beat. The second is for practicing with data and code.
The story and the context
- The Man Who Solved the Market Gregory Zuckerman · 2019 · book
The story of Jim Simons and Renaissance Technologies. Published in Brazil by Alta Books as "O homem que decifrou o mercado".
- The mathematician who cracked Wall Street Jim Simons · TED · 2015 · video
Simons himself, in conversation, on mathematics, codebreaking and how he started investing.
- A Random Walk Down Wall Street Burton G. Malkiel · W. W. Norton · book
The classic argument that markets are hard to beat, useful for judging any strategy with skepticism.
To practice
- Topics in Mathematics with Applications in Finance MIT OpenCourseWare · 18.S096 · free course
Free MIT lectures on the probability, statistics and linear algebra used in finance, with guest talks from the industry.
- QuantConnect Learning Center QuantConnect · free tutorials
Hands-on tutorials to write and test your first strategies in Python with historical data.
- Quantitative trading Investopedia · article
A short reference on the vocabulary of the field.
At ImpactUFSCar
Our Research and Markets group studies and builds quantitative strategies, and our teams have placed in the top 1%, 5% and 10% of the Itaú Asset Quant AI Challenge. See our achievements or get in touch to study with us.
Renaissance figures come from the appendix of The Man Who Solved the Market (Zuckerman, 2019). The chart is an illustrative simulation.