← 返回 optiver 的题目列表Likelihood / Event-Ordering Test
类型:qbank
A newer OA section (appearing in 2026 sittings): read a chart or data table and rank three events from most to least likely. About 15 questions at a fixed ~90 seconds each. It tests data intuition rather than exact calculation.
Requirements
~15 questions, a fixed ~90 seconds per question, introduced in 2026 QR/PhD sittings (replacing some of the classic mini-games for the PhD track).
Each question shows a chart, scatter, network diagram, large data table, boxplot, or fitted regression, plus three events, and asks you to order the events from most to least likely.
Events are typically not mutually exclusive, which makes them hard to rank cleanly — the exercise rewards calibrated data sense, not computation.
Watch the submit mechanics: reordering happens via up/down controls and it's easy to mis-click; leave time to verify the order before the timer expires.
Examples
Reported prompts (paraphrased):
Five exam scores for students A/B/C over five tests; rank: (1) A scores higher than B on the next test, (2) C scores 90+, (3) B and C differ by more than 20.
Scatter of (x, y) with a fitted linear regression; rank: (1) the line over-estimates a new point at a given x, (2) a newly collected point has y > 20, (3) a third event.
A 10×10 grid with two random walkers A and B; rank: (1) A exits the region first, (2) B exits first, (3) A and B meet inside the region.
Notes
Computing each option exactly within 90 seconds is generally impossible; the intended approach is to spot the clearly-unlikely event and gauge the relative likelihood of the other two.
Because this section is recent, the question pool is not yet widely circulated — expect to rely on raw probability/statistics intuition.
Preparation
Practice reading charts and regressions for quick directional inference (over/under-estimation, tail probabilities, correlation strength) rather than point estimates.
Re-check the forum shortly before your interview, since this section is evolving and new examples surface each cycle.