Jim Simons
Operationalized by
Monte Carlo Lab — 109 tabs of distribution
Integration rate
45% — computed, with the gaps below
The record
What is publicly documented.
Verified against multiple public sources. Where sources disagree, the disagreement is stated rather than resolved silently.
- Born
- United States, 1938 · died 10 May 2024
- Education
- MIT, BS mathematics 1958 · Berkeley, PhD 1961
- Codebreaking
- NSA-affiliated cryptanalysis during the Vietnam era
- Academic work
- Chern–Simons theory, with S. S. Chern — foundational in differential geometry
- Academic honour
- Oswald Veblen Prize in Geometry
- Academic post
- Chair of the mathematics department, Stony Brook University
- First firms
- Monemetrics and Limroy, 1978 — merged and renamed 1982
- Renaissance Technologies
- Formed 1982 from those entities
- Medallion Fund
- Launched 1988 · closed to outside capital, employee-only
- Reported returns
- ≈ 62–66% gross annually over roughly three decades; ≈ 37–39% net. Sources vary by window — treat as reported, not audited.
- Firm today
- Renaissance manages multiple funds; 13F equity book reported around $64bn in 2026
Simons had a complete career before he had a trading career. He took his mathematics degree at MIT and his doctorate at Berkeley by twenty-three, did cryptanalytic work during the Vietnam era, chaired the mathematics department at Stony Brook, and co-developed Chern–Simons theory — deep, abstract differential geometry that later found application in quantum field theory and won him the Veblen Prize. He could have stopped there and been remembered as one of the best geometers of his generation.
He left in 1978, at forty, to trade currencies. The early firms — Monemetrics and Limroy — merged into Renaissance Technologies in 1982, and the Medallion Fund launched in 1988. What made Renaissance unlike its competitors was staffing: physicists, mathematicians, statisticians and computer scientists rather than traders or MBAs, on the theory that finance is learnable and scientific talent is not. Researchers tested obscure signals across enormous datasets; almost none survived, and the few that did were scaled ruthlessly.
Medallion's reported performance has no parallel in the record — figures around 62–66% gross annually across roughly three decades, and something near 37–39% after the fund's unusual fee structure. Those numbers come from journalism and secondary analysis rather than public audit, sources disagree on the exact window, and the fund has long been closed to outside money and run for employees. Simons died on 10 May 2024. He is included in this library not because his method is reproducible — it emphatically is not — but because one principle underneath it is.
The framework
The idea underneath the method.
Not the trades — the reasoning that decides which trades are permitted, at what size, and when they stop.
No single trade matters
The transferable idea, and the reason this dossier is open rather than sealed. Renaissance's edge per trade was small and its confidence in any individual outcome was near zero. What it had was an enormous number of weakly favourable bets and the infrastructure to take them all, so that the distribution rather than any instance produced the result. The consequence for behaviour is total: a losing trade carries no information, a winning trade carries no vindication, and the only unit of analysis worth anything is the sample.
The distribution is the object of study
If outcomes are what matter and instances are not, then the thing to understand is the shape of the outcome distribution — its tails, its variance, its persistence, what a bad run inside it looks like. This is a fundamentally different mental model from the one most traders operate, in which the last few trades are the evidence and the distribution is an abstraction. Reversing that priority is available to anyone, at any account size, for free.
Diversify by correlation, not by count
Renaissance's portfolio construction sought many strategies with low correlation to each other, on the understanding that ten positions that move together are one position. Counting instruments tells you nothing; the joint behaviour is the whole question. This is the part of the framework MARS reproduces least well, and the dossier says so explicitly rather than gesturing at it.
Secrecy as a structural condition
Worth stating plainly because it bounds everything else in this profile. Renaissance's specific signals have never been published and almost certainly never will be. Anyone describing what Medallion actually trades is inferring. The framework above is drawn from what Simons and others said publicly about method and staffing, not from knowledge of the models, and this dossier claims nothing further.
Mechanics
How it actually runs.
The operating detail, stated the way a reference describes a technique.
Signal source
Statistical patterns discovered in very large datasets, tested exhaustively, most discarded. Narrative explanation not required for admission.
Edge per trade
Small. The reported success comes from the number of favourable bets and the reliability of the process, not from conviction on any of them.
Holding period
Predominantly short, with statistical arbitrage exploiting brief price discrepancies across many instruments.
Portfolio
Many strategies selected for low mutual correlation, sized against the joint distribution.
Staffing
Scientists rather than traders — the explicit bet that research ability, not market experience, was the scarce input.
Capacity limit
Medallion was closed and capped. The strategy's capacity was finite and the firm chose returns over assets — itself a statement about what was being optimised.
Divergence
Where MARS does something else.
Listed first, and at length, because a mapping that only claims similarity is a poster. Some of these are scale limits; at least one on every dossier is a deliberate refusal.
Where the edge comes from
Jim Simons
Discovered from data by a research organisation, at a scale and cost no individual can approach.
MARS
The operator brings their own edge. MARS has no signal discovery of any kind and does not claim to. It governs the deployment of whatever the operator has.
Correlation
Jim Simons
The central variable. Portfolio construction is a correlation problem before it is anything else.
MARS
Not modelled. The exposure pool sums risk grossly, which is exact for perfectly related positions and conservative for independent ones — it errs toward forgone deployment, never toward hidden leverage. This limit is stated publicly on the risk pages and is the largest single gap in the library.
Sample size
Jim Simons
Enormous. The distribution resolves quickly because the number of bets per unit time is very large.
MARS
Small. A retail operator generates a handful of trades a week, which is why the insufficient-sample state is a first-class output rather than an inconvenience, and why the Monte Carlo layer resamples rather than waits.
Infrastructure
Jim Simons
Execution systems, data acquisition, cost modelling and a research staff — a substantial part of the edge lives in the plumbing.
MARS
A spreadsheet framework and a human. The plumbing does not transfer and pretending otherwise would be dishonest.
Replication
What the rail carries over.
Each mapping names the module that performs the function, so the claim can be checked against the product rather than taken on trust.
Distributional thinking, enforced
Monte Carlo Lab · the sample gate
The one principle that transfers completely, and MARS builds structure around it. Results are read as samples from a distribution rather than as events: the sample gate refuses to grade a read below a trust threshold, the insufficient-sample state is an output the system will print rather than hide, and a single red week is explicitly not an instruction to change anything. That is Simons's discipline, implemented for someone with four trades a week instead of four thousand.
Simulate before you deploy
Monte Carlo Lab · Dynamic 7-Tier Benchmark
Renaissance tested exhaustively and discarded almost everything. MARS resamples a trader's own outcome distribution to produce the drawdown and terminal-equity envelopes a deployment profile implies, and states per tier what the model expects. The purpose is identical — know the distribution's shape before committing capital to it — even though the scale is not comparable.
Live against modelled, not against memory
Live vs Benchmark comparison
The monthly comparison asks whether live expectancy, drawdown, conversion efficiency and behaviour are tracking what the model said they would, and classifies any gap as alpha, variance, drag or failure only after it persists. Judging performance against a modelled distribution rather than against a feeling is the practical form of the no-single-trade-matters principle.
Process over instance
The weekly scorecard's boundary
The scorecard grades expectancy, not profit, and a profitable week can grade red. That inversion — the process is the subject, the outcome is a sample — is the same reordering Simons's framework requires, arriving at the level of a single operator's Saturday review.
Integration rate
Scored, with the shortfall shown.
Each dimension is judged separately and the headline is their mean — recomputed at render, so it cannot be hand-set. Every dossier in this library carries at least one dimension below 35%. Four uniformly high scores would be marketing.
Integration rate
What MARS actually reproduces
Monte Carlo Lab — 109 tabs of distribution
45%
mean of 5 dimensions
No single trade matters
90%
Sample gates, insufficient-sample as an output, one red week is not a signal
Simulate before deploying
85%
Monte Carlo resampling and the seven-tier benchmark envelope
Low-correlation portfolio construction
30%
The pool sums gross and does not model correlation — a stated honest limit
Execution infrastructure and cost modelling
15%
Fee and swap drag are measured post-trade; nothing resembling execution engineering
Signal discovery from data
5%
MARS has none. Entries are the operator's, always
Why it is not higher
45 is the honest number and it is the most useful figure on this page. MARS reproduces Simons's epistemics — the distribution is the subject, the instance is noise — and reproduces none of his machine. There is no signal research, no correlation modelling, no execution engineering and no data operation. A dossier claiming otherwise would be describing a different product.
Jim Simons built the governance function privately, because nothing off the shelf existed to buy. So did every other trader in this library. That private apparatus — the constraint stack, the sizing authority, the rule about when to stop — is the part that never gets published, and it is the part that separates a documented edge from a surviving account.
The claim on this page is not that MARS outperforms anyone. It is narrower and considerably more useful: their method is public, their apparatus was not, and this is the apparatus — at a scale one person can actually run.
Sealed · nearest neighbours
Three dossiers sit next to Simons's.
These mappings are not in the bundle and not rendered anywhere on this site. They are the closest methodological neighbours to the dossier you have just read — which is precisely why they are the ones held back.
Eleven dossiers remain sealed, each at this depth — the framework, the record, the divergences, and the integration rate with its shortfall shown. Waitlist registration unlocks all fifteen.
Join the waitlist — unlock all 15Also declassified
The other three open dossiers.
Sources and standing notice
Jim Simons has no affiliation with Aura Logic Systems or the Montex AlphaRail System, and nothing on this page constitutes an endorsement. This dossier summarises publicly documented method and publicly reported career facts, in the way a reference work describes a technique. It contains no quotations. Performance figures are as reported by the sources listed below, are historical, and are not audited by us; past performance of any trader, fund or method does not indicate future results. Nothing here is investment advice.
The doctrine, made executable.
Eleven governed modules that turn documented method into arithmetic you can actually run.