Standing on the shoulders

This note is about a debt, not a secret. The most quoted firm in quantitative finance is also the most closed, and that combination tempts people to reach for the wrong lesson — to imagine there is a single equation locked in a vault in East Setauket, and that the whole story is the size of the number it prints. That is the sensational reading, and it is the least useful one. The transferable inheritance is not a formula. It is a posture toward the market, and it was assembled by a mathematician who came to trading late and refused to do it the way it had always been done.

Start with the man, because the biography is the argument. Jim Simons was born on April 25, 1938, and died on May 10, 2024, at 86. Before he ran money he did mathematics that had nothing to do with it: he co-developed what is now called the Chern-Simons form, work that became foundational to string theory and topology. He was a codebreaker at the Institute for Defense Analyses from 1964 to 1968, where he learned to pull signal from noise on data that did not want to be read, and he left over his opposition to the Vietnam War. He then chaired the mathematics department at Stony Brook from 1968 to 1978. Only after all of that did he found Monemetrics in 1978, renamed Renaissance Technologies in 1982, and later established the Medallion Fund in 1988. The order matters: the science came first, and the trading was built in its image.

The counterintuitive move was in the hiring, and it is the part worth copying. The firm deliberately staffed itself with mathematicians, physicists, astronomers, statisticians, signal-processing experts and computer scientists — many of them holding PhDs in the sciences — rather than with people who had spent careers on trading desks. The New York Times reported that at the firm "Wall Street experience is frowned on and a flair for science is prized." Read plainly, that is a bet about where an edge comes from. It does not come from market intuition accumulated over years of watching the tape; it comes from people trained to state a hypothesis, measure it honestly, and discard it when the data says no. The domain was finance, but the discipline was research.

Now handle the famous number carefully, because this is where analysis usually collapses into legend. According to Gregory Zuckerman's The Man Who Solved the Market (2019), Medallion returned roughly 66% on average per year gross — before fees — and about 39% net from 1988 to 2018, with the gap explained by a 5% management fee and a performance fee that rose to 44% after 2002. The same account reports that Medallion never had a down year. Those figures are extraordinary, and they should be stated with their caveats attached rather than as folklore. Medallion has been closed to outside capital since 1993; it runs for employees and their families, at a scale most funds could not match. The figures are drawn from a single reported history, not an audited public record you can independently reconstruct. And a track record, however long, is one realization of a process — survivorship and attribution mean the world only ever shows you the funds that lived. The record demonstrates that something real was found. It does not hand you the thing that was found.

So separate what the number proves from what it teaches, because they are not the same. It proves that markets contain structure durable enough to be exploited by a sufficiently rigorous machine, over a long enough horizon, by a team disciplined enough not to fool itself. It teaches something narrower and more useful: that the edge is a research practice, not a possession. The firm's own history makes the distinction visible — the Institutional Equities Fund, opened to outside investors in 2005, has substantially underperformed the closed flagship, a reminder that the process does not transplant cleanly and that capacity, secrecy, and horizon are part of the result. What travels is the method: large data sets, short-term statistical patterns tested for real rather than assumed, extreme discipline about validating a pattern out-of-sample before trusting it, and a culture willing to abandon what stops working. The secrecy — the strict NDAs and non-competes — protects the specific answers. It was never the point.

That is the tradition we place ourselves in, quietly and without claiming its results. We take from it the ordering of priorities, not any of its methods, which are theirs and unknown to us: treat the market as a system to be studied rather than predicted, hold process above any single call, frame everything in terms of risk, and trust nothing until it has survived out-of-sample on data it has never seen. Keep the desk human-in-the-loop, because a model is a hypothesis and someone has to be accountable for when it is wrong. We are not standing where they stood, and we do not pretend to. But the posture — scientific, patient, honest about what a number does and does not prove — is the inheritance worth having, and it is the one we chose.