Interview report
How it went
Coding round (Python)
Market making (graded exactly)
Fair value and spread come from enumerating every outcome of each game. Each card shows your quotes against the exact fair value at the time, your updates against the exact change, the trades and the settlement. Explanations are quoted, not scored. Counts toward the Market making area; outside the interview label.
Full case
Calculations are checked exactly against the case's numbers; a common mistake is named when your number matches it. Structure, insight, ideas, the recommendation and the twist are model-judged, not exact: a model quoted your words for each point and code kept only quotes found word for word.
Conversational (model-judged, not exact)
A model quoted your words for each key point, the twist and any common misconception; code kept only quotes found word for word. A capacity number is checked exactly. Not part of your rating or the interview label.
Behavioural (model-judged, not exact)
A model quoted your words for each step; code kept only quotes found word for word and checked the numbers and "I" versus "we". Not part of your rating or the interview label.
Technical questions (graded exactly)
Session finding
The interview
Overall performance
Evidence you can check
One card per question: your words, your number beside the exact value, and the receipt.
Detailed breakdown
Model estimate (assumed likelihoods)
| Explanation | Model estimate |
|---|
These are model estimates from assumed likelihoods, not a score. The label above is what counts.
Drill a mistake
Up to three questions aimed at the mistake, then one check with new numbers and no help.