Point72 / Cubist Online Assessment
Point72's quant and data hiring runs through Cubist Systematic Strategies. Assessments are HackerRank batteries that differ by track: a 5-question Cubist QR Python OA, a pandas/SQL-heavy Data Scientist/Analyst OA, and a 3-hour Developer task-scheduler test — plus, for data roles, a multi-part take-home (Thinknum review, TreasuryDirect scrape, portfolio P&L). Problem sets are wired later.
Cubist Quant Researcher — QR Python OA (HackerRank)
The well-documented Cubist QR OA: convert a list of date strings to year-quarter form (e.g. '2019-01-21' → '2019Q1', no libraries), sum of all odd divisors of n, square root to a given precision via bisection, longest contiguous subarray with sum ≤ k (sliding window, −1 if none), and classify the relative position of two circles into five cases (concentric, intersecting, tangent, separate, containing) as a one-hot vector. Some cycles add a regression MCQ (high R² but insignificant coefficients → multicollinearity) and a group-by-stated-size partition (LC 1282).
- High R-squared but Insignificant Coefficients Regression · easy
Data Scientist / Data Analyst — pandas + SQL OA (HackerRank)
A pandas/SQL-heavy data OA. Captured: a SQL data-manipulation task on auction lots (buyers/lots/bids → name, starting_price, bid count, current_price = starting + bid_step·count, and current_winner = last bidder), Python mixin classes (DictMixin.to_dict / JSONMixin.to_json, skipping '_'-prefixed attributes, TypeError on non-serializable), validate an IP-address string, longest even-length word in a sentence, splice one linked list into another between positions a and b (LC 1669), and 'max items bought within a budget.' Live rounds add groupwise max and k-th largest per name in both SQL and pandas.
- Product of the Two Regression Slopes Regression · easy
- OLS with Correlated Errors Regression · medium · free
- Ridge vs. Lasso Regression Regression · medium · free
- LASSO on Two Perfectly Collinear Predictors Regression · medium
- Gauss-Markov Assumptions, Linearity, and Regression When k > Regression · medium
- How Swapping or Rescaling Variables Changes a Regression Regression · medium
Quant Developer — Developer Test (task scheduler)
A single 3-hour problem: build a task scheduler. Machines and tasks arrive in time order; you assign tasks to machines, tasks may have dependencies, and you implement addTask, addMachine, completeTask, and a dispatch/checkRunnable method — stating the time complexity of each. Start with naive sort-and-pair, then optimize with heaps and by bucketing tasks by outstanding-dependency count.
- Compose a List of Functions Coding · easy
- Optimal Festival Location to Minimize Total Travel Coding · easy
- Classify the Positional Relationship of Two Circles Coding · easy
- Second Largest Distinct Value Coding · easy
- Convert Dates to Year-Quarter Strings Coding · easy
- Sum of All Odd Divisors of n Coding · easy
Data Scientist / Data Analyst — Take-Home Data Project
A three-part take-home: (1) Thinknum — summarize in < 200 words the features of Thinknum datasets of interest to a systematic trader; (2) Web-scrape (Python) — download 2022 TreasuryDirect competitive-auction results for all Bills and Notes, save raw, parse into a pandas DataFrame, suggest dtypes and key(s), and describe patterns; (3) Portfolio analysis — a 5-position long/short book (COST, KO, NTDOF, TGT, TSLA), compute daily total returns 8/1/2022 → 12/31/2022 (mind corporate actions), show consensus sell-side earnings estimates at both ends, and explain the major performance drivers.
- Product of the Two Regression Slopes Regression · easy
- OLS with Correlated Errors Regression · medium · free
- Ridge vs. Lasso Regression Regression · medium · free
- LASSO on Two Perfectly Collinear Predictors Regression · medium
- Gauss-Markov Assumptions, Linearity, and Regression When k > Regression · medium
- How Swapping or Rescaling Variables Changes a Regression Regression · medium
Point72 / Cubist OA — FAQ
What is the Point72 / Cubist online assessment?
Point72's quant and data hiring runs through Cubist Systematic Strategies. Assessments are HackerRank batteries that differ by track: a 5-question Cubist QR Python OA, a pandas/SQL-heavy Data Scientist/Analyst OA, and a 3-hour Developer task-scheduler test — plus, for data roles, a multi-part take-home (Thinknum review, TreasuryDirect scrape, portfolio P&L). Problem sets are wired later.
What is the format of the Point72 / Cubist OA?
It runs as 4 tracks (QR Python OA (HackerRank), pandas + SQL OA (HackerRank), Developer Test (task scheduler), Take-Home Data Project). Cubist Quant Researcher, Data Scientist / Data Analyst, Quant Developer, Data Scientist / Data Analyst — each timed, auto-graded where applicable.
How do I practice for the Point72 / Cubist online assessment?
Work the sample questions above (each with a full worked solution), then take the interactive Point72 / Cubist OA practice on QuantVault. For the full interview, see the Point72 / Cubist interview-questions set.