The most hit numbers in roulette, and why the list keeps changing
Every live roulette table has a panel showing which numbers have come up most in the last hundred or few hundred spins. The panel is accurate. It is also the most misread thing in the casino, because it shows a sample, and samples cluster. Here is what a real record looks like, what counts as unusual, and why the most-hit list is a different list every time you look.
Put it into practice
This is an affiliate link to a roulette site we partner with. You must be 18 or over.
What 1,000 spins look like
On a 37-pocket wheel each number expects 1,000 ÷ 37 ≈ 27 hits. The actual counts spread around that: the standard deviation is about 5, so most numbers land between 22 and 32, a few land at 17 or 37, and in most samples one number somewhere is up near 40 and another is down near 15. That is pure chance on a fair wheel. The number at 40 is the "hottest number" on the panel and has exactly a 1-in-37 chance on the next spin.
What 100 spins look like
Worse. Expected hits per number: 2.7. Typical range 0 to 6. Around 2 or 3 numbers will not have appeared at all and one or two will have hit five or six times. The panel will show those as hot and cold with confidence. A sample of 100 is far too small to say anything about the wheel, and it is the sample size most panels default to.
Play roulette for real money
We send you to one of the roulette sites we work with. Affiliate link. 18+.
Clustering versus bias
A biased wheel is one where a pocket or sector really does come up more often, because of a physical fault. It was a real thing in casinos decades ago and it is the one historical case where tracking numbers paid. Detecting it needs thousands of spins on the same physical wheel and a result far outside the normal spread, sustained over time. Online: RNG tables have no physical wheel to be biased, and live wheels are maintained and checked precisely to prevent it. The clustering you see on a panel is sampling noise, not a fault.
Why the list changes
Because the sample moves. Each new spin adds one result and drops one from the window. A number that was hot at 9pm is average by 11pm. If hot numbers were a property of the wheel they would stay hot; they do not, which is the simplest evidence that they are a property of the sample.
A test you can run
Pick any live table. Write down its five hottest numbers. Come back in 200 spins and check how many of them hit more than the expected 5.4 times in that stretch. Do it a few times. The hot numbers will perform exactly like five numbers picked at random, because that is what they are.
What to do with the panel
Enjoy it. It is a record of what happened, and watching results cluster and un-cluster is a nice illustration of how randomness looks. Just do not change a stake because of it. Every number on the panel, hot, cold or absent, is 1 in 37 on the next spin.
Ready to play?
Affiliate link: we may earn a commission. Play responsibly. 18+ only.
Quick answers
Which roulette number hits the most?
Over a long enough record, none. Over any short record, whichever happened to cluster. A live table's "hot numbers" panel shows the past 100 to 500 spins and the list is different at every table and every hour.
How many times should a number hit in 1,000 spins?
About 27 on a single-zero wheel, with a normal spread from roughly 17 to 37. A number at 40 or at 14 is unusual but happens somewhere on the wheel in most 1,000-spin samples.