The Mental Arithmetic Standard at Optiver and How to Reach It

Optiver's process tests arithmetic across multiple rounds, and reaching the standard candidates describe takes weeks of structured practice, not raw speed.

Optiver is widely described by candidates as weighting mental arithmetic more heavily than most trading firms, and reaching the standard its process expects means treating speed and accuracy as a trained skill built over weeks rather than a talent a candidate either has or does not. People who have been through the process describe multiple numerical rounds spread across a phone screen and an on-site day, each testing addition, subtraction, multiplication, division, percentages and quick estimation under a visible clock. The standard is less about one fast trick and more about holding error-free speed for the full length of an interview instead of one strong round followed by a tired one.

Candidates typically encounter a timed numerical screen early, often before any conversation about markets, followed by arithmetic embedded directly inside trading games later in the process rather than tested on its own. A market-making round that asks you to keep a running position and a running profit and loss across ten or more trades is, underneath the trading judgment, also an arithmetic test, since every price you quote depends on adding or subtracting correctly from the last one.

This is why an error early in a round costs more at Optiver-style interviews than a single wrong answer on a standalone drill. A running position that is wrong by one unit stays wrong for the rest of the round, and an interviewer watching you trade will notice the second and third decisions built on the mistake even if the first one passes unnoticed.

Candidates who have gone through the process describe a combined score in the high 40s to 60s correct answers on a two-minute mixed-operation drill as competitive, a range reported by people who have sat the interviews rather than a published company figure. At the midpoint of that range, roughly 55 correct answers in 120 seconds works out to a little over 2 seconds per answer on average, which only holds up if no single operation is taking noticeably longer than the other three.

A market-making round moves at a similar pace. If a trading game runs for two minutes and asks for ten price updates, each update needs to be produced in about 12 seconds including the arithmetic, the judgment about where the price should move, and the decision to trade or pass, so the raw calculation inside that update needs to take a small fraction of the total time available.