The Optiver Mental-Math Round: Format, Cutoffs, and How to Train

If you apply to Optiver — and to a lesser extent IMC, Flow Traders, or any Amsterdam-lineage market maker — the first real filter is an arithmetic test. Not probability, not brainteasers: raw arithmetic under time pressure. Most rejections in the funnel happen here, and almost all of them are preventable with training.

The format

The classic Optiver screen is 80 questions in 8 minutes. That's six seconds per question, including reading it and typing the answer. The questions themselves are deliberately unremarkable:

• Two-digit addition and subtraction, often with decimals: 47.5 − 18.9
• Two-digit by one-digit and two-digit by two-digit multiplication: 17 × 23
• Division that resolves to clean decimals: 144 ÷ 0.6
• Fractions and percentages: 3/8 as a decimal, 15% of 240

Two properties make it hard. First, wrong answers are penalized — typically at −2 against +1 for a correct answer, so guessing is disastrous and skipping is a legitimate strategy. Second, six seconds does not allow deliberate calculation. Anything you can't do reflexively, you can't do at all in this format.

What score do you need?

Optiver doesn't publish cutoffs, and they vary by office and season. The consistent reports from candidates land in the same band: you need to attempt 60+ with very high accuracy. In penalty-adjusted terms, think of the bar as a net score around 50. Candidates who pass comfortably describe finishing 65–75 questions at 95%+ accuracy.

The implication for training: accuracy first, then speed. A candidate at 100% accuracy and 55 attempts passes; a candidate at 85% accuracy and 75 attempts fails on penalties.

The four-week plan

Week 1 — diagnose and de-rust. Run one timed session daily in the real format. Don't chase the score yet; find your leaks. Almost everyone has one or two specific weaknesses — usually two-digit multiplication or decimal division. Your per-operation accuracy stats will tell you exactly which.

Week 2 — kill the leaks. Drill your weak operations. For two-digit multiplication, learn the three workhorse techniques:

• Distributive split: 17 × 23 = 17 × 20 + 17 × 3 = 340 + 51 = 391.
• Difference of squares: 18 × 22 = 20² − 2² = 396. Any pair symmetric around a round number.
• Squares to 25 memorized cold, plus the identity (a+b)² when just past a known square.

For decimal division, normalize first: 144 ÷ 0.6 is 1440 ÷ 6 = 240. Train the normalization to be automatic — the mistake is always the decimal point, not the division.

Week 3 — speed under penalty. Now run every session with penalty scoring on, and adopt the skip rule: if the path to the answer isn't visible in about two seconds, skip. No pride. The test rewards ruthless triage, which — not incidentally — is also what trading rewards.

Week 4 — simulate. Full-length sessions at the real cadence, daily, same time of day as your scheduled test if you can. The goal this week is variance reduction: your worst day, not your best day, is what shows up under pressure.

Three mistakes that fail prepared candidates

1. Training untimed. Untimed accuracy is a different skill. Every practice rep should have a clock on it from day one.
2. Ignoring the input method. Typing speed and number-pad fluency are worth several questions. Practice with the same input style the test uses.
3. Guessing when behind. Under a −2 penalty, a guess with 50% confidence has negative expected value. If your practice has drilled the skip reflex, you won't tilt into guessing on test day.

The honest summary

This round is the most trainable filter in all of quant recruiting. There is no insight to have, no cleverness to demonstrate — just a skill that responds almost linearly to deliberate, timed practice. Four weeks of daily ten-minute sessions moves most candidates from failing to passing comfortably.

The Quant Ladder mental-math trainer replicates this exact format — timed, penalty-scored, per-operation stats — and it's free.