Millennium Management Online Assessment

Quant Researcher (intern) — 24-hour Market-Making Project

A 24-hour take-home delivered as a notebook with a provided helper library and step-by-step instructions. Challenge 1: a basic market-making class that quotes bid/ask around a fair price with a fixed spread and trades against a provided hedge-fund order flow (each challenge is broken into many small sub-questions). Challenge 2 (reported as the hardest): inventory-skewed quotes (the example uses 1% and 5% skews), a daily 'risk exposure' metric the prompt deliberately leaves undefined (an accepted approach sums the daily market value of stock + ETF positions), a more complex spread, and a visualization. Challenges 3-4 extend further (candidate-reported). Gotcha: the hedge-fund flow returns 'buy'/'sell' in lowercase and from the counterparty's point of view — opposite to the side you pass in.

• Basic Quoting Engine (Market-Making Take-Home, Challenge 1) Coding · easy
• Inventory-Skewed Quotes and Risk Exposure (Challenge 2) Coding · medium

Quant Researcher (intern) — QR Intern Hiring Test (HackerRank)

An earlier/alternate QR intern OA: 7 questions in 120 minutes. Captured items: Q1 a birthday-holiday scheduling MCQ, Q2 a triangle-interior regression, Q3 a CLT fill-in (limits of the mean and variance of a quadratic-form average), Q5 minimum transactions to settle group debts (the LC 465 'Optimal Account Balancing' pattern), Q6 a pandas order-netting engine (total gross/net shares and orders per netting window), and Q7 cyber-attack breach-points (for each starting level, count points under an energy budget; naive per-start simulation is O(n^2)). Q4 (pandas) was not captured.

• Limiting Mean and Variance of a Quadratic-Form Average Probability · medium

Quant Developer / SDE — QD / SDE Coding OA (HackerRank)

The QD/SDE OA is a ~6-question, 3-hour coding battery. Captured problems: multi-currency PnL calculation (positions + native-currency prices + daily FX, liquidate on the last day), permutation-divisible-by-8, spreadsheet cell-number-to-notation conversion (702 columns/row), 'intelligent substring' (longest substring with at most k normal characters), and multi-label scene classification (predict six binary scene flags from 300-attribute image vectors). An older 4-question software-intern set adds IP-region classification, Subsegment Sort, Product Defect (LC 221 Maximal Square), Shape Classes (OOP), and Queued Seats.

• Multi-Label Scene Classification From Image Vectors Machine learning · medium

Data Engineer — Data Engineer OA (HackerRank)

SQL-heavy data-engineering OA. Captured: max-element frequency in every suffix of an array (right-to-left running max + count), a SQL protocol aggregation per client (GROUP_CONCAT ordered by input+output volume, output ordered by client), and a SQL anti-join task (customers with no invoices and products on no invoice). A DS-adjacent variant adds basic pandas filtering and a Python MCQ.

• Regression with Duplicated Data Regression · easy
• Ridge vs. Lasso Regression Regression · medium · free
• Regression Line Over a Uniform Triangle Regression · medium
• Python List vs. Hash Map: Internals and Performance Coding · easy
• Suffix Maximum Frequency for Range Queries Coding · easy
• Balanced Brackets Coding · easy

Data Scientist — Data Scientist Intern OA (HackerRank)

A mixed DS intern OA: 5 MCQs on probability and ML concepts, 1 SQL question, 1 two-sum-style algorithm, and a coding task to write a gradient-descent function using SymPy (symbolically differentiate the objective and iterate). A separate Quant/ML OA gives a 3-hour guided used-car price-prediction pipeline (max-correlation feature, preprocessing, model selection, results).

• Used-Car Price Prediction Pipeline Machine learning · medium
• Symbolic Gradient Descent With SymPy Coding · medium

Quant Researcher (intern) — Caliper Personality Test

A Caliper personality assessment in the QR intern funnel: a standard trait/work-style questionnaire (forced-choice + abstract-reasoning items). Not a coding or math test.

Practice Millennium OA problems →