20 · The Arsenal

MATHEMATICS: stochastic calculus, optimisation, Itô calculus, multivariable calculus, martingales, eigenvalue decomposition, convexity proofs, Black-Scholes-Merton models, Monte Carlo simulations, Schrödinger's cat, PDEs. STATISTICS AND DATA MODELLING: signal extraction from noisy data, maximum likelihood estimation, Bayesian inference, PCA, Kalman filters, regime detection, multivariable regression. CODING AND PROGRAMMING: Python, C++, SQL — without libraries, IDEs, or internet access — algorithmic complexity, memory layout, cache behaviour, numerical stability, transcending foundational syntax; ideally mastery of six to seven languages. LOGIC AND INTERVIEWS: rote intelligence, IQ, intellectual brain teasers, logic interviews — the Green Book and the rest.

Pre-hire preparation: Putnam, IMC-style mathematics, olympiad probability, the International Olympiad in Informatics, contest rankings, published research starting from undergrad, open-source systems. Post-hire training: shadow research, tool building, recreation of existing models, simulation, backtesting, small-account live trading, demo trading, model validation with consistent CAGR, consistent paper trading and stress testing. PnL, Sharpe, drawdowns, hit rates, risk contribution.

Chat With Traders podcasts. Becoming a Hedge Fund Analyst: Inside Point72's Trading Academy. Flirting with Models. The Jim Simons extended interview.

Jamie, if your goal is Citadel or Jane Street for a year before your masters, your stack should look like: one, mathematics; two, probability theory; three, programming; four, statistical modelling; five, market microstructure. Trading comes last, not first — monitoring markets, live sessions, building models, backtesting GitHub projects. And I can do research, I can do hard maths, I can do probability, I can code. It's something I am going to dive into, and I guarantee it will be harder, more stressful and more painful than expected — it's up to you whether you can push to the point of resistance and beyond tolerance.

Aman Manazir's interviews about quants — seeking alpha: statistics and probability; linear algebra and multivariable calculus; early internship or trainee experience at SME prop shops; Berkeley's statistics and machine learning courses; DSA and why each algorithm works; Python and C++; dynamic programming; thinking probabilistically, coding efficiently, optimising algorithms; solving hard problems fast; Markov chains; the Green Book — wrestle it down, training mental muscles; Option Volatility and Pricing; finance puzzles; calculus; 100+ Leetcode problems across arrays, graphs, trees, heaps, string manipulation, recursion, dynamic programming. Don't let perfect be the enemy of good — one is better than zero. Solve quant brain teasers while you commute, work out, eat. Leverage all downtime; low ROI is still something. Memorise a shitload of Leetcode questions if you want return offers from big tech or quant firms as training ground. Buy Leetcode Premium — it's self-reinvestment. Practice system design questions, pub-sub systems, third-party AI packages. Eighty percent-plus of Leetcode mediums. Tag Facebook on Leetcode; there are a hundred unique questions — understand them.

At the fundamental level: how does a computer operate? Open up your old M1 MacBook and see what each system is — you have the flexibility to do so. Operating systems, compilers. High-level systems design; watch the systems experts' classes; take distributed systems. Know the keywords recruiters look for. Code quality matters a lot. OA — online assessments. Category theory. Computer engineering and hardware classes are more hands-on: embedded systems, networking, pipelining.

The first foundation for the next years is knowledge and skillset ingestion — trading is just exposure for the first half decade. I think I can work through Roman Paolucci's roadmap on how to be a quant before I get into any UK university, then post-freshers leverage all that experience into something hands-on.