Online Assessments at Quant Firms and How They Are Scored

Quant trading assessments score mental arithmetic, probability, and logic against a moving cutoff that shifts with each cycle's applicant pool.

An online assessment at a quantitative trading firm typically runs sixty to ninety minutes and combines timed mental arithmetic, probability and expected value questions, and logic puzzles, with a coding component added for research and developer seats. Scoring runs against a cutoff set by that cycle's applicant pool rather than a fixed pass mark, which is why two candidates with nearly identical raw scores in different years can see different outcomes.

The arithmetic section usually runs as a series of rapid problems, addition, multiplication, percentage change, and fraction simplification, presented one at a time against a countdown clock measured in seconds rather than minutes per question. A separate section covers probability and expected value, often built around dice, card, or coin-based scenarios that require translating a word problem into arithmetic quickly rather than solving anything conceptually difficult. A market-making or pricing simulation appears at some firms, asking a candidate to quote a two-sided price on a quantity and then trading against a simulated counterparty.

Most platforms present these sections in a fixed order and do not allow returning to a previous section once it closes, so time management across the full ninety minutes matters as much as accuracy inside any single section. A candidate who spends too long perfecting the arithmetic round can arrive at the probability section with less time than the format assumes, which is one of the most common self-inflicted causes of a weak overall score.

Scoring generally rewards both accuracy and throughput, meaning correct answers per minute rather than a fixed count attempted. Guessing carries a smaller penalty than leaving a question unanswered at most firms, since the assessment is measuring speed under a real clock and an unanswered question provides no signal at all. A candidate who works twenty problems at 90 percent accuracy typically scores better than one who works twelve at 100 percent, because the format is built to separate candidates who have automated the arithmetic from those still computing it consciously.

This section is scored more like a conventional test, with partial credit uncommon and full marks awarded only for the exact correct value or the correct multiple-choice selection. The questions tend to reuse a small set of underlying structures: conditional probability, expected value of a wager, and combinatorics problems dressed in different scenarios each cycle, which is why practicing the underlying mechanism matters more than memorizing any single worked example.