The four preparation stages
1. Numerical screening
Build reliable arithmetic habits: decomposition, complements, percentage anchors, fraction–decimal conversions, squares and sequence recognition. Accuracy comes before maximum speed.
2. Probability and expected value
Practice counting, complements, conditioning, Bayes, linearity of expectation, stopping times and fair-game reasoning. Explain each step as though an interviewer can interrupt.
3. Trading and strategy rounds
Move from a fair value to a quote. Consider uncertainty, adverse selection, inventory and how new information changes the market. Add game theory, betting and calibration.
4. Role-specific depth
Researchers need statistics, stochastic processes and coding. Options desks need derivatives and hedging intuition. Developers need a separate algorithms and systems plan.
A focused four-week plan
A repeatable daily session
- Five minutes of easy arithmetic to become fluent.
- Ten to twenty targeted questions from one weak category.
- One strict mixed session with no feedback until the end.
- Review misses and classify the error: method, arithmetic, assumption or time management.
- Re-solve the missed family with different parameters.
Practice the interview conversation
Do not let silent calculation become your only training mode. State what is known, identify the random variables, say which method you are considering and flag uncertainty. When the interviewer changes a condition, update the structure rather than restarting blindly.
A useful answer structure
Clarify → model → compute → sanity-check → communicate sensitivity. This structure works for probability puzzles, estimates, market-making games and research questions.
Common preparation mistakes
- Memorizing final answers without learning why the method works.
- Practicing only favorite topics and avoiding weak categories.
- Using a calculator during speed training.
- Ignoring coding for research or development roles.
- Believing a firm guide predicts the exact interview.
- Taking full mock interviews too early, before basic methods are stable.
How to measure progress
Track accuracy by category, average response time and the type of mistake. A higher score can hide a persistent weakness if the session contains mostly familiar topics. The useful signal is whether performance improves on fresh variants of the same family.
Turn the guide into timed practice
Select the matching category or firm profile, then run a short session and review every miss.
Start a fresh session