Published · 2026-10-25 · Last reviewed · 2026-10-25 · 6 min read

English · 한국어

Hot and cold lottery numbers: does frequency mean anything?

Open any lottery statistics site and you will find a "most frequent numbers" list and an "overdue numbers" list. The first group is called hot, the second cold. The tables are tempting: frequent numbers seem to have something going for them, and numbers that have not appeared for a long time feel due. Few people check whether such a table says anything about the next draw. This page does, using every Korean Lotto 6/45 draw from round 1 to round 1247 (six main numbers each, bonus ball excluded).

The counts really do differ

Across 1,247 draws each number should appear about 166.3 times (1,247 × 6 ÷ 45). What happened:

Number 34 and number 9 are 50 appearances apart, which looks large. Whether it means "34 has something going for it" is a separate question. Even a fair draw cannot give all 45 numbers exactly 166.3; some come up more, some less. The real question is whether the spread exceeds what chance produces.

Compared with 300 fake histories

An experiment answers it. We simulated 1,247 Lotto 6/45 draws on a computer, built the same table, and repeated this 300 times. In the fake histories every number has exactly the same chance, so "hot" and "cold" mean nothing by construction. We summarise how spread out the counts are with one figure (the chi-square statistic) and compare.

• Chi-square of the real history: 28.8
• Fake histories: median 38.8, 5th percentile 27.7, 95th percentile 51.8

The real value is inside the normal range and near its lower end, at roughly the 5th to 10th percentile. The real counts are slightly less spread out than a fair draw typically produces. The gap between frequent and infrequent numbers therefore comes from nothing special about a few numbers; it is what random selection produces over a limited number of draws.

The standard deviation of the counts is 10.3 in reality; a fair draw would give about 12.0 (√(1,247 × 6/45 × 39/45)). The real draw is a little more even than theory, which, if anything, counts against the idea that particular numbers are favoured.

Does recent frequency predict the next draw?

Hot-number believers say to follow the recent trend. We can test that too. For every round from 11 to 1247 and every number, we counted how often it appeared in the previous 10 rounds, then recorded whether it appeared in the current round. Any number has a 6/45 = 13.33% chance in each round.

Between numbers that missed all of the last 10 rounds and numbers that appeared three or more times, the difference is 0.3 percentage points. The 13.62% for 7,789 cases equals the expected 13.33% within the sampling error of about ±0.8 points. Recent frequency carries no information about the next draw.

Is a long-absent number "due"?

The cold-number logic runs the other way: a number that has not appeared must come soon to restore balance. Using the same method, we grouped cases by how many rounds had passed since the number last appeared.

Numbers absent for 20 or more rounds appeared in the next round 12.67% of the time, slightly below the expected 13.33%. With a sampling error of about ±1.2 points that means no difference. A long-absent number is not "owed" an appearance. The draw machine has no memory.

The longest absence is also unremarkable. One number once missed 73 draws in a row. The fake histories say how surprising that is: their longest gaps had a median of 64 rounds, with a 5th–95th percentile range of 55–82. Missing a draw has probability 39/45 = 86.7%, so 73 in a row is about a 0.003% event for one given number, but 45 numbers each get 1,247 chances, so somewhere a gap like that must occur.

Why hot numbers still look convincing

Three reasons. First, sorting after the fact always yields a top and a bottom; every simulated history had a "most frequent" and a "least frequent" number as well. Second, people are good at finding patterns and read chance clusters as signals. Third, statistics sites show the past; they do not show whether anyone predicted the next draw from their tables. The two tables above are exactly that test.

Try it below. The tool gives every Lotto 6/45 number the same probability and simulates draws. Even with 1,000 draws it produces a most-frequent and a least-frequent number, and the chi-square p-value is an ordinary random value.

The shape of the numbers is within chance too

Beyond hot numbers, claims circulate such as "the sum should be between 100 and 175", "an odd-even split of 3:3 is most common" or "there is usually one consecutive pair". The same method checks these. The theoretical values come from exact combinatorial counting.

The observed values track theory almost exactly. "3:3 is the most common split" is true (414 rounds, 33.2%), but not because the machine prefers it: more combinations have that split. Do not read it backwards. One ticket with three odd and three even numbers is no likelier than one with six odd numbers; each is 1 in 8,145,060. There are simply more combinations in that category. Extreme sums such as 1-2-3-4-5-6 are rare for the same reason: there are few such combinations, and no individual one is disadvantaged.

Questions that come up

Do statistics sites give wrong numbers? Mostly the numbers are right. The interpretation attached to them is what goes wrong. A frequency table is an accurate record of the past; only its use to explain the next draw has no basis.

Would the last 100 rounds look different? The smaller the sample, the larger the gaps between numbers appear. Over 100 rounds each number is expected 13.3 times with a standard deviation of about 3.4, so a number appearing 6 times and another 20 times is unremarkable. The gaps look large because the sample is small, not because they carry meaning.

Still, can I pick frequent numbers? You can. Every choice has the same chance of winning, so it costs nothing and gains nothing. It is worth remembering that if many others share the idea, the risk of splitting a prize may be larger.

Limits of this check

• This does not prove the machine is perfect. It shows that number frequency and the previous results have no visible influence. We did not search for every more subtle effect (for example correlations between pairs). These two assumptions are, however, what the hot-and-cold belief depends on directly.
• The bonus ball is excluded. Including it changes the figures slightly and not the conclusion.
• A count's size does not change the next draw's probability. Each number's chance stays 6/45 in every round, and one ticket's jackpot chance stays 1 in 8,145,060.

One practical point remains. Every choice of numbers has the same chance of winning, but if many people choose the same numbers, a win is shared. If many follow hot numbers, that feeds split risk, which is the subject of the winner-count analysis.

LottoFlick is independent and not an official lottery operator. Numbers are random and cannot be predicted.

Sources and verification

• Donghaeng Lottery: Lotto 6/45 round results, rounds 1–1247 (collected by hand 2026-10-24) — verified
• LottoFlick aggregate output used on this page (includes 300 simulated histories, seed 20261025) — computed by LottoFlick

Related reading

• The Gambler's Fallacy: Why It Is Never 'Due' Myths, corrected
• Is There a "Lucky Store"? Sales Volume and Jackpot Counts Myths, corrected
• The Birthday-Number Trap: Same Odds, Different Split Risk Myths, corrected
• Statistics: how to read draw history
• Lotto 6/45 game guide