Disco Numbers Game: How It Works and How to Score High

Disco Numbers is the assessment game candidates underestimate most. Two numbers appear; you pick the larger one; repeat, fast, for around thirty rounds. Everyone can compare 26 and 23 — so what is being tested?

What Disco Numbers actually measures

Processing speed. The game times every single decision, and your score is built from how quickly and consistently you resolve trivially easy judgements. It is closely related to the reaction time test, but with a cognitive step inserted: perceive two magnitudes, compare, commit. Employers read it as a proxy for how quickly you handle routine decisions when there is no difficulty excuse — which is why hesitation, not ignorance, is the only way to lose points.

There is a wrinkle worth knowing from the research: humans compare numbers by magnitude representation, not digit-by-digit. Pairs that are close together (61 vs 59) take measurably longer to resolve than distant pairs (91 vs 23) — the classic distance effect. The game mixes both; the close pairs are where your average slips.

How scoring works

Typical implementations score median response time across rounds, gated by accuracy — a handful of wrong answers voids far more than the milliseconds they saved. Consistency matters too: a tight cluster of 500 ms decisions beats a mix of 300 ms flashes and 900 ms stalls with the same average. That has a direct strategic consequence: your job is not to produce your fastest possible answer, it is to remove your slowest ones.

Strategy: rhythm over reflexes

• Set a beat and answer on it. Treat the game like a metronome — see, press, see, press. A steady ~600 ms cadence outperforms erratic bursts on every scoring model that weights consistency.
• First digit decides most rounds. 8x vs 5x is over at the first digit; only same-length, same-first-digit pairs (61 vs 59) need the second digit at all. Let your eyes land on the leading digits, not the whole numbers.
• Different lengths = instant answer. A three-digit number beats a two-digit number, always. Spotting length is faster than reading values.
• Keep both hands ready. One finger per side (or thumb per side on a phone) removes travel time and keeps the rhythm physical.
• Do not celebrate or grieve mid-game. Thirty rounds take under a minute; any mental commentary — "I'm doing great", "that one was slow" — inserts exactly the stall the scoring punishes.

Warm up before the real thing

Processing-speed games are the most warm-up-sensitive in the battery: your first ten cold decisions of the day are reliably slower than the rest. Play a practice round or two immediately before the assessment — the free Disco Numbers game runs the same mechanic with percentile scoring, and Numerosity makes a good second warm-up for the numerical block.

For where this fits in the full battery, see the complete guide to game-based assessments; for the cross-game habits that protect fast scores (sleep, caffeine timing, environment), the 10 tips guide covers them all.

Where you'll meet this game

Number-comparison speed tasks like this one turn up across pymetrics-style batteries. If you're preparing for a specific employer, J.P. Morgan's pymetrics assessment is one of the best-known places it appears. For the full picture of the provider's game battery — and the other tasks that sit alongside it — see the complete pymetrics games guide.

What do people actually score?

First attempts here have a median of 76 out of 100 and a top tenth of 86, across 38 players. The median gain from first run to best is 10 percentiles — enough to matter, small enough to show this is a game about rhythm rather than ability.

FAQ

Is this just a maths test? No. The comparison itself is trivial; the scoring punishes hesitation, so it measures how steadily you can make easy decisions at speed.

Is it better to be fast or accurate? Both, in that order of difficulty — accuracy holds up under speed far longer than most people expect, and the score rewards rhythm over caution.

What is a typical score? Our first-attempt data puts the middle at 76 out of 100.