Arrowstreet Capital Online Assessment
Arrowstreet's quant-research screen, as captured by a candidate in the 2027 cycle, runs on a HackerRank-style platform (bookmark flag per question, 'Clear Selection', 'Save & Proceed', a visible countdown, questions revisitable). It has a numbered section of 11 questions, mostly 'complete the blanks in the following question with the appropriate answer' numeric entry with explicit rounding instructions, plus a block of 'Pick ONE option' / 'Pick ONE or MORE options' econometrics MCQs. The numeric questions are finance-flavoured probability and optimization: sums of normals, a two-deck likelihood-ratio threshold, a geometric-weight divisibility probability, a factor-model portfolio Sharpe ratio, a tangency portfolio (with a hint about maximizing the inverse standard deviation at unit expected return), beta of a stock to an equal-weighted portfolio, a cone volume under a radius-plus-height constraint and the KKT linear system of an order-execution problem. The MCQs test regression judgment: a dummy-variable specification for grade inflation and SAT-era shifts, adding quadratic terms when a supervisor expects a U-shape, why a January test set underperforms a February validation set, the most parsimonious seasonal-momentum-trend model for quarterly revenues, and what a t-stat of 4 does and does not imply. The countdown read about 39 minutes at question 3 and under 14 minutes at question 10 of the numbered section, so budget roughly 4 minutes per question. Everything here is candidate-reported: 13 distinct questions were captured (questions 1 and 5 of the numbered section were not), every numeric answer was recomputed independently by QuantVault, and the few option lists that were out of focus are reconstructed and flagged in the solutions. The second section below covers what candidates report after the screen: a live HackerRank CodePair session (Python/pandas concept questions plus two LeetCode easy/medium problems) and a superday that opens with a paper-based written exam of statistics and asset-pricing derivations.
Practice every firm's OA: timed replicas, free game replicas and guides in one place →
Quantitative Researcher — Quant research screen (numeric entry + econometrics MCQs)
One sitting, two flavours of question. The numbered section is 11 'complete the blanks' numeric problems (rounding rules spelled out: 'rounded down to the nearest integer', 'in %') on sums of normals, a modified-deck likelihood ratio, geometric weights, factor-model and tangency-portfolio Sharpe ratios, beta to an equal-weighted portfolio, a constrained cone and a Lagrange-multiplier system. The MCQ block is single- and multi-select econometrics judgment: dummy variables, quadratic terms, calendar-month train/test splits, seasonal time-series models and t-stat interpretation. Our timed replica below runs all 13 captured questions at the real pace with verified answers and full worked solutions; the related bank problems linked here drill the same shapes.
Divisibility Events Under Geometric Weights →
Maximum Sharpe Ratio Portfolio with Uncorrelated Assets →
Identifying the Biased Coin From Limited Flips →
Handling Non-Linearity in Data →
Cross-Validation Leakage in Financial Time Series →
Omitted Variable: Does the Simple Slope Match the Multiple Slope →
- Predicting GPAs: Dummy-Variable Specification Regression · easy
- Sum of Random Variables Probability · easy
- Drawing From Two Decks Probability · medium
- Random Variable Over Positive Integers Probability · easy
- Analyzing a Portfolio of Two Funds Finance · medium
- Optimizing a Portfolio of Two Funds Optimization · medium
- Computing Beta to a Portfolio Finance · medium
- Maximizing the Volume of a Cone Optimization · easy
- Optimal Execution With Lagrange Multipliers Optimization · medium
- Predicting Electricity Usage Regression · easy
- Forecasting Security Returns Machine learning · medium
- Predicting Firm Quarterly Revenues Time series · medium
- Statistically Significant Predictor Statistics · medium
Quantitative Researcher — CodePair coding screen + superday written exam
What comes after the online screen, as reported by candidates in the 2025-26 cycle. The CodePair session mixes Python semantics (slicing, views vs copies, @staticmethod vs @classmethod under inheritance) and pandas operations (merge with duplicated keys, groupby agg vs transform, rolling with min_periods / centered windows, resample closed / label) with two LeetCode easy/medium problems -- 'count the words with exactly k distinct letters' was one. The superday then opens with a paper exam of statistics and asset-pricing derivations: mean-variance / tangency, risk parity and the equal-risk-contribution condition, regressing X on Y versus Y on X, effective sample size under autocorrelation, and log-utility certainty equivalents. The thirteen problems below are original worked versions of every reported topic; every pandas and Python output was produced by actually running the snippet, and the two coding problems run on the judge.
- Python Slicing Semantics: Bounds, Steps, Aliases and Views Python · easy
- @staticmethod Versus @classmethod Under Inheritance Python · medium
- Merging Trades and Quotes With Duplicated Keys pandas · easy
- groupby: transform Versus agg pandas · easy
- Rolling Windows: min_periods and center pandas · easy
- Resampling Irregular Ticks: closed and label pandas · medium
- Count Words With Exactly k Distinct Letters Coding · easy
- Longest Anagram-Free Run of Words Coding · medium
- Regressing Y on X Versus X on Y Regression · medium
- Effective Sample Size Under AR(1) Autocorrelation Time series · medium
- Log Utility: Certainty Equivalent, Maximum Price and Growth-Optimal Sizing Asset pricing · medium
- Risk Parity: Two-Asset Closed Form and the Equal-Risk-Contribution Condition Asset pricing · hard
- Tangency Portfolio for Two Assets: Numbers and Intuition Asset pricing · medium
Arrowstreet Capital OA — FAQ
What is the Arrowstreet Capital online assessment?
Arrowstreet's quant-research screen, as captured by a candidate in the 2027 cycle, runs on a HackerRank-style platform (bookmark flag per question, 'Clear Selection', 'Save & Proceed', a visible countdown, questions revisitable). It has a numbered section of 11 questions, mostly 'complete the blanks in the following question with the appropriate answer' numeric entry with explicit rounding instructions, plus a block of 'Pick ONE option' / 'Pick ONE or MORE options' econometrics MCQs. The numeric questions are finance-flavoured probability and optimization: sums of normals, a two-deck likelihood-ratio threshold, a geometric-weight divisibility probability, a factor-model portfolio Sharpe ratio, a tangency portfolio (with a hint about maximizing the inverse standard deviation at unit expected return), beta of a stock to an equal-weighted portfolio, a cone volume under a radius-plus-height constraint and the KKT linear system of an order-execution problem. The MCQs test regression judgment: a dummy-variable specification for grade inflation and SAT-era shifts, adding quadratic terms when a supervisor expects a U-shape, why a January test set underperforms a February validation set, the most parsimonious seasonal-momentum-trend model for quarterly revenues, and what a t-stat of 4 does and does not imply. The countdown read about 39 minutes at question 3 and under 14 minutes at question 10 of the numbered section, so budget roughly 4 minutes per question. Everything here is candidate-reported: 13 distinct questions were captured (questions 1 and 5 of the numbered section were not), every numeric answer was recomputed independently by QuantVault, and the few option lists that were out of focus are reconstructed and flagged in the solutions. The second section below covers what candidates report after the screen: a live HackerRank CodePair session (Python/pandas concept questions plus two LeetCode easy/medium problems) and a superday that opens with a paper-based written exam of statistics and asset-pricing derivations.
What is the format of the Arrowstreet Capital OA?
It runs as 2 tracks (Quant research screen (numeric entry + econometrics MCQs), CodePair coding screen + superday written exam). Quantitative Researcher, Quantitative Researcher, each timed and auto-graded where applicable.
How do I practice for the Arrowstreet Capital online assessment?
Work the sample questions above (each with a full worked solution), then take the interactive Arrowstreet Capital OA practice on QuantVault.