Why hot and cold numbers do not work
Every lottery site publishes a hot and cold list. The counts really do differ, sometimes by a lot. The thing almost none of them shows you is the number that makes those differences interpretable.
The gap is real
Take Bonoloto. Across 3,501 draws in our archive, the most-drawn number has come up 92 more times than the least-drawn. Put that in a table, sort it, and the top looks meaningfully different from the bottom.
It is not a rounding error and nobody is making it up. The question is not whether the gap exists. The question is what size of gap you should expect when nothing at all is going on.
The number nobody shows you
Here is the trap. You are not looking at one number and asking whether it is unusual. You are looking at fifty-odd numbers and picking out the highest and the lowest. Take the extremes of any large group and they will be far apart, however identical the group is.
That expected gap can be calculated. For a pool of 49 numbers that all behave identically, chance alone puts the highest and lowest about 4.5 standard deviations apart.
| Game | Draws | Observed gap | Gap chance produces |
|---|---|---|---|
| Bonoloto | 3,501 | 4.7 sd | 4.5 sd |
| US Powerball | 1,404 | 5.0 sd | 4.7 sd |
| La Primitiva | 1,318 | 4.1 sd | 4.5 sd |
| El Gordo | 549 | 6.0 sd | 4.5 sd |
| Mega-Sena | 325 | 4.1 sd | 4.6 sd |
Read that table across, not down. In every one of those games the observed gap sits close to the gap chance produces. The hot and cold lists are exactly as spread out as a fair draw makes them.
Two questions people mix up
How long has a number been missing? and what is the longest it has ever been missing? are different questions, and the second is routinely presented as the first.
A number can hold the record drought for a game and have been drawn last week. "Number 12 has gone 111 draws without appearing" reads as a statement about now. Usually it is a statement about 2019.
On our statistics pages both figures are shown separately and labelled, because the difference is the whole point.
The gambler's fallacy, precisely
A ball that has not come up for a year is not owed an appearance. The machine has no record of what it did last week, no mechanism for keeping score, and no way to act on one if it had. Each draw is independent, which means exactly this: knowing every previous result tells you nothing whatsoever about the next.
The reverse belief — that a hot number is "running" and will continue — is the same error wearing different clothes. Both assume the draw remembers.
Then why do we publish the tables?
Because they are a genuine record of what has happened, and because the alternative is leaving you to find the same list somewhere that attaches an implication to it.
We publish them with the comparison number attached, and we withhold them entirely when an archive is too short for the ordering to mean anything at all. On a young archive a ranked list is noise with a sort applied, which is worse than showing nothing.
Figures computed Sunday 20 September 2026 from our published archive.