How to become a quant, in the order that works

Before you start

Be honest about where you are. If calculus and probability feel rusty, start there; everything later depends on them. If you already program for a living, you can move faster through the coding steps and spend the time on statistics and markets instead.

The steps below are a sequence. Each one assumes the ones before it.

The route

1. Learn how markets work. What stocks, bonds, futures and options are, how an order book matches buyers and sellers, and where trading costs come from. Without this, you will test strategies that cannot be traded.
2. Build the maths. Calculus, optimisation and linear algebra, aimed at portfolios and risk rather than textbook exercises.
3. Learn probability and statistics properly. Distributions, fat tails, estimation, hypothesis tests and their limits. This is the single most important step, and the most often skipped.
4. Program on real data. Python for analysis and SQL for getting data out of a database. Aim to load years of prices, adjust them correctly and compute returns you can trust.
5. Learn to test honestly. Backtesting with costs, out-of-sample checks, and the many ways a test can flatter a strategy. Only now does it make sense to try trading ideas.
6. Pick a specialty. Statistical arbitrage, derivatives pricing, risk, portfolio construction, machine learning or high-frequency trading. Each has its own tools, and none of them make sense without steps one to five.
7. Build one complete project and write it up. This is what you will talk about in interviews.

Step one: markets

Many technically strong people skip this step and pay for it later. A strategy that buys at the closing price it used to make its decision, or ignores the cost of crossing the spread, can look profitable on paper and lose money every day in practice.

Steps two and three: the maths and the statistics

Work through calculus and linear algebra with a finance purpose in mind: the portfolio that minimises risk for a given return is a calculus problem, and the covariance matrix of a set of stocks is linear algebra. Then give probability and statistics more time than feels necessary.

Step four: code on data

The test of this step is practical. Can you take a raw file of daily prices for a few hundred stocks, handle the missing days and the corporate actions, and produce returns you would bet on? When you can, you are ready to test ideas.

Step five: testing without fooling yourself

This is where amateur and professional work separate. Learn the main traps: using information that was not available at the time, leaving out companies that later disappeared, ignoring costs, and trying many variations while reporting only the best one.

Step six: choose a specialty

Choose the specialty that matches the role you want. Some common pairings:

• Research at a fund: statistical arbitrage, factor models, time series.
• Trading: position sizing, execution, order books.
• Banks and pricing: stochastic calculus, option models, numerical methods.
• Risk and validation: risk measures, stress testing, model risk.

Step seven: the project

Pick one idea and take it all the way: the rule written down before testing, clean data, realistic costs, a held-back test period, and a short write-up that says what failed as well as what worked. Keep it small enough that you can explain every line. In an interview, that explanation is worth more than any return figure.

If you want structure

The Institute’s three study paths set out this order with marked coursework and proctored exams at the credential levels: the practitioner path , the researcher path and the trader path , which needs no programming. If you are not sure yet, start with the free course and read about the roles you could aim at.