Why Quants Always Win: The Complete Math Stack.

Every equation a billion-dollar fund runs on. None of it is secret. And honestly, most of it is simpler than it looks.

There is a comforting story people tell themselves about quants. That the people running Renaissance or Citadel are just smarter than the rest of us, or faster, or sitting on information nobody else can see. It makes losing feel less personal.

But that story is mostly wrong, and the real one is far more useful to you.

Quants win because they share one toolbox. A stack of equations that takes something terrifying - an uncertain future - and turns it into a single number you can act on. And here is the part I really want you to sit with: almost none of these equations are secret. They are in textbooks. Most were published decades ago, some more than a century ago. A good number of them are free online right now, tonight, for you.

So the moat was never the math. The moat is knowing which pieces matter, how they fit together, and having the patience to trust them. That last part is rarer than you’d think, and we’ll get to it.

To put numbers on it: Renaissance Technologies’ Medallion Fund earned roughly 66% a year before fees from 1988 to 2018. Thirty years, and not a single losing one. It is the best track record markets have ever seen, and it was not built on a secret. It was built on the stack you’re about to read.

Let me walk you through the whole thing, gently, from a forgotten Frenchman in 1900 to the neural networks of today. For each idea I’ll tell you who wrote it, what it actually does, and how a fund uses it. You do not need a PhD to follow this. You just need to take it one layer at a time.

Layer 1 — The Foundation: Turning Chance Into a Number

Before you can price anything, you need a language for uncertainty. These four ideas are that language, and you already feel most of them in your bones.

**The Law of Large Numbers - Jacob Bernoulli, published 1713. **Bernoulli proved something you half-know already: repeat a bet enough times and your results stop looking random and start looking like the truth underneath them. A 51% edge means nothing across ten trades - you genuinely cannot tell skill from luck. But across ten thousand trades, that tiny edge becomes almost a certainty. This one idea is the whole reason funds make millions of small bets instead of a few big ones. It is also the reason most people quit a perfectly good strategy too early: they stop before the math has had time to work.

**Bayes’ Theorem - Thomas Bayes, published 1763, after he died.**This is just a careful rule for changing your mind when new evidence shows up. You start with a belief, you see fresh data, and you update to a sharper belief. That’s it. A quant model never claims to “know” the future - it holds a probability and quietly revises it as each new price arrives, exactly the way Bayes described two and a half centuries ago. If you’ve ever updated a guess after new information, you’ve already done Bayesian reasoning by hand.

**Markov Chains - Andrey Markov, 1906.**Markov noticed that in a lot of systems, where you go next depends mostly on where you are right now, not on your entire history. Markets get described as regimes - calm, trending, panicking - with probabilities of moving between them. The big regime-switching models at the top funds all grow from this single, humble observation by a Russian mathematician.

**Brownian Motion - Louis Bachelier, 1900.**This is my favorite, because almost nobody is taught it. A French graduate student named Louis Bachelier stood in front of Henri Poincaré at the Sorbonne in 1900 and described stock prices as a random walk - wandering up and down with a precise mathematical shape. He worked out the math of Brownian motion five years before Einstein used it in physics, and the math of option pricing seventy-three years before Black-Scholes. The world ignored him for fifty years. Then modern finance quietly rebuilt itself on exactly his idea. Keep him in mind - there is something hopeful in a forgotten student turning out to be right all along.

Layer 2 — The Engine: Pricing the Unpriceable

Once you accept that prices wander randomly, you need a kind of calculus that works on randomness. This is where things got serious, and lucrative.

**Itô’s Lemma - Kiyosi Itô, 1940s–50s.**Ordinary calculus, the kind from school, falls apart on a jagged random path. A Japanese mathematician named Kiyosi Itô built the version that doesn’t - stochastic calculus. Think of Itô’s lemma as the chain rule for randomness. It is the quiet engine that lets you take Bachelier’s wandering price and actually compute with it. Every derivatives desk on earth leans on it, whether or not the traders there could rederive it from memory.

**Black-Scholes-Merton - 1973.**Fischer Black, Myron Scholes, and Robert Merton took Bachelier’s randomness and Itô’s calculus and combined them into one clean formula for the price of an option. It earned a Nobel Prize in 1997. But the bigger story is what it did to the world: before this formula, options were a niche curiosity. After it, they became a market worth trillions, simply because everyone finally agreed on what an option was worth. One equation, an entire industry.

Layer 3 — The Edge: How Much to Bet

Finding an edge is only half the job. The half that quietly ruins talented people is sizing - deciding how much to put on.

**The Kelly Criterion - John Kelly, Bell Labs, 1956.**Kelly found the exact fraction of your money to bet to grow your capital as fast as possible over the long run. Bet too little and your edge compounds at a crawl. Bet too much and an ordinary losing streak wipes you out, even though your edge was real. Edward Thorp took this single formula, used it to beat blackjack in Las Vegas, then ran a fund for nineteen years without a losing one. When people ask Thorp for his most important tool, he doesn’t name a signal. He names Kelly. So many traders have a genuine edge and still go broke, purely because they never learned how much to bet.

**Modern Portfolio Theory - Harry Markowitz, 1952.**Markowitz turned “don’t put all your eggs in one basket” into actual mathematics. He showed that a portfolio’s risk isn’t just the average of its parts - how the parts move together is what matters, and smart diversification is close to a free lunch you can calculate. Nobel Prize, 1990. Whenever a desk sizes many connected bets at once, it is solving Markowitz’s problem.

**The Capital Asset Pricing Model - William Sharpe, 1964.**Sharpe asked a simple, deep question: how much return should an asset give you for the risk you take holding it? His answer gave markets the words “beta” and the “Sharpe ratio” - return earned per unit of risk. Nobel Prize, 1990. When a quant shrugs that a strategy is “garbage below a Sharpe of one,” they are speaking Sharpe’s language, sixty years on.

Layer 4 — The Signal: Finding Truth in the Noise

Most of what a market does is noise. These tools are how you pull the quiet signal out of it.

**Information Theory - Claude Shannon, 1948.**Shannon, working at Bell Labs, invented the entire idea of measuring information - what a “bit” is, how much a channel can carry, how to tell signal from static. The whole digital world rests on it. And here’s a lovely connection: Kelly’s betting formula came straight out of Shannon’s information theory. A trading signal is, formally, just an information channel - and Shannon wrote the rulebook for channels.

**The Kalman Filter - Rudolf Kalman, 1960.**Kalman solved a hard, beautiful problem: figure out the true state of a system, in real time, when every single measurement you get is noisy. NASA used it to guide Apollo to the moon, blending unreliable sensors into one trustworthy estimate of where the spacecraft really was. Quant desks use the exact same machinery to estimate a hidden relationship between two assets, or the true trend under a noisy price. Same equations - just swap “where is the spacecraft” for “what is the real link between these two prices.”

**Volatility and Cointegration - Robert Engle, 1982 and 1987.**Engle showed that volatility isn’t a fixed number - it comes in waves, and you can model those waves (the ARCH and GARCH models). With Clive Granger, he also gave us cointegration: a careful test for whether two prices are truly bound together and bound to snap back, versus just drifting in step by coincidence. This is the heart of pairs trading. Nobel Prize, 2003.

Layer 5 — The Strategy: Playing Against Other People

A market isn’t a machine you’re solving alone. It’s a room full of other people, all optimizing against you. Game theory is how you think clearly in that room.

**Game Theory and the Nash Equilibrium - John Nash, 1950.**Nash proved that in any game of strategy, there’s at least one balance point where no single player can do better by changing their move alone. Every price you see is exactly that - a temporary truce between buyers and sellers. Every crowded trade is a prisoner’s dilemma waiting to break, because the smart move is to leave first. Every market-making spread sits at the razor’s edge where no one dares move first. Nash wrote this at twenty-one; the Nobel came forty-four years later.

**Convex Optimization - modern; Stephen Boyd and others.**Once you can describe your risk, your return, and your limits, the natural question is: what is the single best portfolio, provably, not just by trial and error? Convex optimization is the branch of math that finds that answer with a guarantee. It’s how a fund turns a messy pile of signals and risk rules into one clean allocation. The standard textbook is free online - another quiet reminder that the math is not the thing being kept from you.

Layer 6 **— **The Modern Stack: Simulation and Machine Learning

This last layer is the one everyone is loud about today. It’s also the newest, and far less mysterious once you see where it comes from.

**Monte Carlo Simulation - Stanislaw Ulam and John von Neumann, 1940s.**Born at Los Alamos to model the atomic bomb, Monte Carlo is a wonderfully honest idea: if a probability is too hard to calculate directly, just simulate the situation thousands of times and count what happens. Funds use it to price strange derivatives and to stress-test a portfolio against futures that haven’t occurred yet. The math built to model a bomb now quietly prices the products in your portfolio.

**Neural Networks - Geoffrey Hinton and others.**A neural network doesn’t predict the future, and it’s worth letting go of that fantasy early. What it actually does is learn the expected outcome given your inputs, across thousands of variables at once - patterns no single indicator could ever hold. Geoffrey Hinton spent decades making this work and won a Nobel Prize for it in 2024. Reinforcement learning (an agent that gets a little better each loop) and random forests (a crowd of small decision trees that vote) are the same spirit: many weak guessers, combined into one strong answer.

The Part Nobody Selling You a Course Wants to Admit

Now go back and look at the dates with me.

Bernoulli, 1713. Bachelier, 1900. Markov, 1906. Kelly and Shannon in the 1950s. Kalman, 1960. Black-Scholes, 1973. Nearly every equation that runs a ten-billion-dollar fund has been public for fifty years or more. You and the quant at Citadel are reaching for the very same mathematics.

So if the math is shared, why do quants keep winning? Three reasons, and none of them is the formula.

Data and speed. The equations are free. Clean data, microsecond execution, and a server sitting next to the exchange are not. A lot of the edge lives in the gap between knowing a formula and being fast enough to act on it before everyone else.

The combination. No single equation is the moat. Kelly without a real edge is just faster ruin. A signal without Kelly is a blown account. The funds that last for decades wire the whole stack together so that each piece quietly covers the weakness of the next.

The discipline. This is the one almost nobody has, and the good news is that it’s the one you can actually build. A 51% edge needs thousands of trades to separate from luck. Most people quit in week three, before the math has had a chance to win, or they override the model on a feeling and hand back a year of progress in an afternoon. The equations are a century old. The patience to truly trust them is the rarest, and the most teachable, edge on Wall Street.

The market is not a chart. It is mathematics, and that mathematics has been sitting in a textbook the whole time, waiting for someone calm enough to use it.

Quants don’t win because they know a secret.

They win because they read the textbook - and then they had the discipline to believe it.

Here’s the question I’d love you to sit with. Every equation here is free, old, and public. If the math was never the moat, then the only thing between you and this stack is which idea you’d choose to learn first, and whether you’d trust it long enough to let it work. So - which one would you start with, and why?