"150 Most Frequently Asked Questions on Quant Interviews" by Stefanica, Radoicic, and Wang is a key preparation resource for quantitative finance roles, covering topics like mathematics, programming, and brainteasers. The third edition (2024) expands on previous versions by adding over 200 questions, including new content on machine learning, option pricing, and stochastic calculus. For more details, visit FE Press .
: Give a real modeling example where it matters. Which models have high bias vs. high variance?
Binomial distribution – properties, applications Q55 - Q56: Poisson distribution – limiting case of binomial, arrival processes Q57 - Q58: Geometric and negative binomial distributions 150 Most Frequently Asked Questions On Quant Interviews
A box contains 3 red marbles and 7 blue marbles. You pick two without replacement. What is the probability they match?
: Explain how you would identify cointegrated pairs and implement a mean-reversion strategy. "150 Most Frequently Asked Questions on Quant Interviews"
Mental models & interview strategy (10)
: Define a stationary distribution of a Markov Chain. How do you calculate it given a transition matrix Additional High-Frequency Probability Prompts Expected number of aces in a 5-card poker hand. Expected value of rolling an -sided die where you can choose to reroll once. : Give a real modeling example where it matters
Pay special attention to eigenvalues and matrix decompositions – they come up repeatedly in factor models and risk decomposition.
Supervised vs. unsupervised vs. reinforcement learning – key distinctions. Q174 - Q176: Linear regression assumptions – linearity, independence of errors, constant variance, no multicollinearity. Q177 - Q178: Regularisation – L1 (Lasso) vs. L2 (Ridge) – effect on coefficients, use cases. Q179 - Q180: Logistic regression – how is it used for classification in trading signals? Q181 - Q183: Overfitting – how to detect, prevent (cross‑validation, regularisation, early stopping). Q184 - Q185: Explain the bias‑variance tradeoff with a concrete modelling example. Q186 - Q187: Decision trees, random forests – advantages, interpretability, overfitting. Q188 - Q189: Gradient boosting – XGBoost, LightGBM – why it often works well for structured data. Q190 - Q191: Neural networks – backpropagation, activation functions, vanishing/exploding gradients. Q192 - Q193: What is wrong with constant (e.g., 0 or 1) initialisation of weights in a neural network? Q194 - Q195: Time series forecasting with ML – LSTMs, GRUs, handling non‑stationarity. Q196 - Q197: Feature engineering, feature selection, data leakage – how to avoid leakage in a trading pipeline. Q198 - Q199: Model evaluation for imbalanced data – precision, recall, F1, AUC‑ROC. Q200: How would you choose a machine learning model for a real‑time trading task, balancing latency, interpretability, and data volume?
