A deceptively simple exercise that tests the thing interviewers actually watch: can you compute a fair value faster than the market moves?
Start the Fruit Market drillFree, runs in your browser, nothing to installTwo bags contain apples and oranges. The market's value is (total apples) × (total oranges) across both bags. A quote appears with a bid and an ask. Buy if the ask is below fair value, sell if the bid is above it, pass otherwise.
That is the entire rule. The difficulty is doing it in three seconds while the contents keep changing.
It isolates one skill — cross-multiplication under time pressure — and strips away everything else. There is no probability, no memory, no strategy. You either compute the product fast enough to act, or the moment passes.
Real market making feels like this. The fair value is knowable; the constraint is that everyone else is computing it too.
Add before you multiply, and add the smaller side first — it is easier to hold one number in your head while working on the other.
Round and correct: 7 × 68 is easier as 7 × 70 minus 14. Most fruit totals are near round numbers, so this works constantly.
Watch the supply events. When a vendor adds two apples, you do not need to recompute from scratch — the value increases by 2 × (total oranges). Updating an existing value is far faster than rebuilding it, and that habit is exactly what a trader does when news hits.