Just finished walking recently Optiver The entire interview process for 2026 New Grad Data Scientist and finally got the Offer. Optiver's DS interview style is very unique. OA focuses on mathematical speed + probability + brain teasers, while VO focuses on statistical modeling + trading scenarios + in-depth communication. The overall pace is fast and the pressure is high, but the interviewers are professional and friendly.
Optiver Online Assessment
OA for Optiver DS is quite long (~2.5–3 hours) and is divided into two main parts:
Part 1: Coding Challenge (HackerRank, 3 questions)
- Topic 1: Market Data Processing Given a stock's tick data (timestamp, price, volume), calculate VWAP per minute, identify anomalous transactions, and sort the output by time.
- Ideas: Use a dictionary to group by minute, calculate the total value and total transaction volume to get VWAP, and then use moving average + standard deviation to identify anomalies. Pitfalls: I have been debugging the case where volume is 0, timestamp analysis, and edge case processing for a long time.
- Topic 2: Options Portfolio Risk Given a combination of calls/puts with different strike and expiry, calculate the total portfolio delta, P&L for a ±5% change in underlying price, and the position with the highest risk.
- Ideas: Calculate delta using Black-Scholes and then aggregate all positions. For P&L, use delta approximation to estimate the impact of price changes.
- Topic 3: Time Series Forecasting Given time series data, implement a simple ARIMA model, predict 5 time points in the future, and calculate the confidence interval.
- Ideas: First differentially handle non-stationarity, and then use least squares method to estimate AR/MA parameters. This question was the most difficult. It took me nearly 50 minutes, and I spent the last few minutes debugging like crazy.
Part 2: Mathematical Reasoning (20 math questions)
Covering probability, statistics, linear algebra, and calculus, the time is very tight.
A few questions that impressed me deeply:
- Probability question: biased coin P(Head)=p, continue tossing until HHT appears, find the expected number of tossing (use the state transition equation to solve).
- Statistics question: Show that the sample variance S² is an unbiased estimate of σ² (using expected linearity and chi-square distribution properties).
- Linear algebra question: How to quickly calculate matrix Aⁿ? (Eigenvalue decomposition A = PDP⁻¹).
I didn't have time to finish 3 questions in this part, and I felt very stressed. A week later, I unexpectedly received a phone screen invitation, which was such a surprise!
VO - three rounds in total
Round 1: Advanced Coding
- Real-time Market Data Processing: Design a system that processes 1 million tick data per second, requiring calculation of real-time VWAP, support of historical query, and 99.9% latency < 1ms. I proposed the solution of event-driven + Kafka + C++ processing engine + InfluxDB + Redis cache, and discussed latency optimization (pre-allocate memory, SIMD, DPDK, etc.).
- Algorithm Challenge: Given an order book, achieve the optimal execution strategy (considering market impact and slippage). I used the TWAP + dynamic programming method.
Round 2: Statistics & Probability
- How to tell whether a new trading algorithm is statistically significant? (two-sample t-test + Bootstrap + non-parametric alternatives)
- How to estimate the VaR of a portfolio? (Historical, Parametric, Monte Carlo + Kupiec test)
- How to check whether a financial time series is stationary? (ADF test, KPSS test)
Round 3: Machine Learning & Research
- Feature Engineering for intraday returns prediction (price-based, volume-based, market microstructure, alternative data)
- Model validation in finance (time series CV, look-ahead bias, economic significance)
- Deep Learning for price movement prediction (LSTM/Transformer + regularization + interpretability)
Overall feeling
The Optiver DS interview places great emphasis on mathematical skills + statistical thinking + understanding of trading scenarios. OA focuses on speed and foundation, while VO focuses on depth and practical application. Although the whole process was stressful, the interviewers were friendly and proactively guided.
Suggestions for students preparing for Optiver 2026 NG DS:
- Mathematics and probability questions must be practiced a lot (especially expected value and state transition)
- Proficient in pandas + statistical analysis + Black-Scholes Greeks calculations
- Prepare multiple trading-related modeling cases (PnL analysis, strategy evaluation, overfitting judgment)
If you are also preparing for Optiver DS, please feel free to communicate via private message:
- Want to see ARIMA handwritten code or analysis of probability problems?
- Need a detailed framework for a trading strategy case?
- Want to know the difference between DS and SDE interviews?
If you are pressed for time and want to prepare more efficiently, recommended Programhelp Professional assistance services. Provide OA ghostwriting, VO real-time idea guidance and full-process support. Many students have passedProgramhelpI got the Optiver Offer with my help.
If you need help, please send us a private message to discuss in detail.
