← Optiver Interview Insights

Optiver·Software Engineer·Technical Phone Screen·Intermediate

Intermediate
Jun 2026

Summary

Optiver quant engineer interview with a pretty gnarly probability and statistics question centered on linear regression residuals. One question, but it had enough layers to keep me second-guessing myself the whole time.

Questions Asked (1)

Q1

You have a scatter of (x, y) points with a fitted linear regression line. Rank these three events from most to least probable: (1) the regression overestimates y for a new x drawn uniformly from the training x-range, (2) a new (x, y) pair has y greater than 20, (3) the absolute residual for a new point exceeds 1.5 times the standard deviation of the training residuals. Justify your ranking using what you know about residual distributions, zero-mean residuals, normality, and homoscedasticity.

A/B Testing & ExperimentationTechnical Trade-offsProduct Analytics & Metrics
Author's notes

This one is deceptively layered.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

First, clarify the assumptions of linear regression: residuals are zero-mean, normally distributed, and homoscedastic. Then, analyze each event's probability using these assumptions, comparing the likelihood of overestimation, exceeding a fixed threshold, and exceeding a multiple of residual standard deviation. Finally, rank the events based on the derived probabilities.

Pro tip: Emphasize that the ranking depends on the specific values of the standard deviation and the threshold 20, so state your assumptions clearly and consider edge cases.

1. State assumptions

Assume residuals are independent, zero-mean, normally distributed with constant variance (homoscedasticity).

2. Analyze event 1

The regression overestimates y for a new x if the residual is negative. Since residuals are zero-mean, this probability is 0.5.

3. Analyze event 2

The probability that y > 20 depends on the distribution of y, which is centered around the regression line. Without knowing the mean or variance, this probability is unknown but likely less than 0.5 if 20 is far from the mean.

4. Analyze event 3

The probability that |residual| > 1.5σ is approximately 0.134 (since residuals are normal, P(|Z|>1.5) ≈ 0.134).

5. Rank events

Compare probabilities: event 1 (0.5) > event 3 (0.134) > event 2 (unknown, but likely smallest if 20 is extreme).

Key Points to Mention

  • Residuals are zero-mean, so overestimation probability is 0.5.
  • Normality of residuals implies P(|residual| > 1.5σ) ≈ 0.134.
  • Homoscedasticity ensures constant variance for all x.
  • Event 2's probability depends on the distribution of y and the value 20.
  • Without additional info, event 2 is likely the least probable if 20 is far from the mean.
  • The ranking assumes the model is correctly specified and residuals follow the stated assumptions.

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.