Cold numbers and the gambler's fallacy: why 'overdue' is a word, not a strategy

"Number 34 hasn't come up in twenty draws — it's overdue." This sentence feels so reasonable that it has its own name in psychology: the gambler's fallacy. It is also, checkably, false. Let's do the checking.

The one probability that matters

In a 6/49 game, the probability that any particular number appears in a given draw is:

P(appears) = 6/49 ≈ 0.1224

Why? Six of the 49 balls come out; by symmetry your number is equally likely to be any of them. Equivalently: the number of 6-ball draws containing your number is C(48,5) = 1,712,304, out of C(49,6) = 13,983,816 total, and 1,712,304 / 13,983,816 = 6/49. Same answer from both directions — always a good sign.

The crucial word in that formula is what's not in it: history. The machine mixes a fresh set of balls every draw. No component of the apparatus stores what happened last week. P(appears) = 6/49 whether the number appeared five draws running or has been missing since the machine was purchased.

How long do droughts actually last?

The probability a specific number misses one draw is 43/49 ≈ 0.8776. Draws are independent, so:

Now the punchline. There are 49 numbers, so the expected number of numbers currently sitting on a 20-draw drought is roughly 49 × 0.073 ≈ 3.6. A twice-weekly lottery will essentially always have a few numbers that look shockingly "cold". A cold number isn't a signal; it's furniture. The average gap between appearances of any given number is 49/6 ≈ 8.2 draws, and gaps of double that are routine.

(These are per-number figures; the exact count of simultaneous droughts shifts a little because numbers share draws, but the order of magnitude is the point — and you can watch it live on our statistics pages.)

The demonstration: conditional probability, then Monte Carlo

Here is the claim the "overdue" theory makes, written as maths: P(34 appears next draw | 34 missed the last 20 draws) > 6/49.

And here is why it fails. Independence means P(A | B) = P(A) when B carries no information about A. The last 20 draws used balls that have since been returned, remixed, and in most lotteries swapped for a different certified ball set entirely. There is no physical channel through which the drought could push on the next draw. For the drought to make 34 more likely, the machine would need memory — and if machines had memory, the chi-square tests that statisticians run on decades of draw data would light up. They don't: John Haigh's peer-reviewed analysis of the UK National Lottery (JRSS Series A, 1997) specifically tested the waiting times between appearances of each number — the exact quantity the "overdue" theory says should be distorted — and found them fully consistent with independent random draws.

If algebra doesn't move you, simulation will. The Monte Carlo logic is simple enough to describe in four lines, and you can run it yourself with our draw simulator:

1. Simulate millions of independent 6/49 draws.
2. Every time a number completes a 20-draw drought, mark the next draw.
3. Count how often the drought-ridden number appears in those marked draws.
4. Compare to 6/49 ≈ 12.24%.

Run it with ten million draws and the drought-conditional hit rate lands within a whisker of 12.24% — same as the no-drought hit rate, same as the after-a-hot-streak hit rate, same as everything. The condition you sliced on is invisible to the outcome. That is what "independent" means, made physical.

Why our brains insist otherwise

The gambler's fallacy comes from a true fact applied at the wrong scale: in the long run, every number's frequency drifts toward 12.24% of draws. Our intuition converts that into "the universe must correct the shortfall soon." But the law of large numbers works by dilution, not correction: a 20-draw shortfall doesn't get repaid, it just gets swamped by the thousands of ordinary draws that follow. The absolute gap between a number's count and its expectation can keep growing forever, even as the percentage converges. Nothing is ever "due."

The cost of playing cold numbers

Here's the consolation: unlike most lottery mistakes, this one is nearly free. Cold numbers win at exactly the same rate as any others, so a cold-number strategy doesn't lower your odds — it just wastes the mental energy. The only real cost is if a seller charges you for a "due numbers" list, at which point you have paid actual money for the statistical equivalent of a horoscope. If a number is widely publicised as "due" and lots of players pile onto it, you've also bought into a sharing problem — more co-winners if it ever does hit.

Balls have no memory. The good news is that this cuts both ways: the lottery can't hold a grudge against your numbers either. Every draw, everyone starts at exactly 6/49 — which is either comforting or damning, depending on how you feel about 6/49.