Strategy guides › Count Rush
How to score big in Count Rush: count in groups, protect the streak
The dots are gone before you’ve finished looking. Four buttons: 21, 23, 24, 26. You think it was “somewhere in the low twenties”, the clock is running, and you tap 21. Wrong, three seconds gone, streak back to zero.
Count Rush looks like a test of eyesight. It’s really a test of three habits, and all three come straight out of the game’s code.
What the game actually does
Each round shows a cloud of dots, then four numbers. Your level goes up by one for every three correct answers, up to level 6 after 18. Misses don’t lower it. At level 0 a cloud has 6 to 14 dots and stays up for 1.3 seconds. Each level adds a few dots and takes 0.09 seconds off the flash, so by level 6 a cloud has 30 to 56 dots and you get 0.76 seconds. The 45-second clock runs the whole time, during the flash and while you decide.
Two details matter. First, the dots vary in size at random and never overlap, so the amount of blue on the screen tells you nothing. Count, don’t judge the area. Second, only the exact count scores. The other three options are all close to the right answer.
See groups, not dots
People see up to about four objects at once without counting. Psychologists call it subitizing, and it’s fast and nearly error-free. Beyond four, we switch to estimating, and estimates get rougher as the number grows.
So don’t try to take in the cloud as one blob. In the flash, let your eye jump from clump to clump and take each clump as a small number you just see: three, four, two, three. Then add. Clouds in this game are scattered at random, so clumps form on their own. You don’t have to sort them neatly, just take each one in at a glance.
At level 0, with at most 14 dots and 1.3 seconds, that’s only three or four clumps. By level 4 and up, with 22 to 56 dots in under a second, you won’t finish the groups, so count one or two clumps exactly and size up the rest against them.
Close is not enough
The four options are packed tight. The code first picks a spacing for the four buttons, spanning at most about a third of the count (twice 18%, and never less than 4), and then drops the true count onto one of the four positions at random. On average the smallest and largest buttons are only 3.9 apart at level 0, 7.9 apart at level 3 and 12.4 apart at level 6. We generated 100,000 rounds per level with the game’s own code: each position holds the answer 25% of the time, and guessing the position from the spacing alone, without looking at the dots, is also right only 25% of the time. (Earlier versions leaked the answer, first through which button it was on and then through the gaps between the buttons; both are fixed.)
So the buttons won’t rescue a rough guess. You need your estimate within a number or two. To see how much precision is worth, we simulated a player whose estimate is typically off by a given share of the true count, picking the button closest to that estimate:
| typical error | level 0 | level 3 | level 6 |
|---|---|---|---|
| 5% | 83% | 70% | 68% |
| 10% | 62% | 52% | 51% |
| 15% | 51% | 44% | 43% |
| 20% | 45% | 40% | 39% |
Halving your error from 10% to 5% lifts the hit rate by 17 to 21 points at every level. No trick with the buttons comes close to that, which is why the grouping habit matters most.
Accuracy beats speed
Points are 10 times your streak, capped at 5, so from your fifth hit in a row every answer is worth 50. A miss costs three seconds and resets the streak. So it doesn’t just cost one answer. The next four hits only earn 10, 20, 30 and 40, which is 100 points less than four hits at full streak.
To see how that plays out, we ran the game’s scoring and timing rules 20,000 times for each combination of how long you take to answer after the flash and how often you’re right:
| answer time | 70% right | 80% right | 90% right | 100% right |
|---|---|---|---|---|
| 0.8 s | 286 | 445 | 708 | 1,150 |
| 1.2 s | 245 | 373 | 570 | 900 |
| 1.6 s | 213 | 319 | 477 | 700 |
| 2.0 s | 193 | 283 | 412 | 600 |
Read along a row and then down a column. Getting 10 points more accurate is worth far more than shaving 0.4 seconds. Answering in 1.6 seconds at 90% beats answering in 0.8 seconds at 80%, even though the fast player sees more rounds. This table holds your accuracy fixed, which real play doesn’t, since the clouds grow as you score. The shape is the point: a long streak is the engine, and a miss is the most expensive thing in the game.
So take the extra half second when you’re unsure. Don’t guess fast. Then play fast when the answer is obvious, which at low levels is most of the time.
Practice
Count Rush has no seeds: every cloud is random. Play a few rounds with one rule only: take each clump in as a small number, then add. Once that feels automatic, start counting one clump exactly and sizing the rest against it, and watch your streaks get longer.
See small groups, count one exactly, and protect the streak.