Analyses: observed versus expected
Each statistic on this page is shown next to the value you would expect if the draws were purely random. The expected values are calculated exactly (closed-form formulas over all 52,360 combinations), not simulated. The result in advance: everywhere, what we observe stays within the range of chance.
2655 draws · 18.05.2001 – 22.10.2026 · Exact expected values under the hypothesis of uniform 4-from-35 draws
How close together are the four numbers?
The linear span is the distance from the smallest to the largest number (end positions included). The circular span reads the numbers 1 to 35 on a circle where 35 touches 1: the draw 2, 3, 5, 31, 33, 34, 35 is then no longer spread over 34 positions but concentrated on 10. The bars show the distribution of draws by span, observed and expected.
Linear span
Mean: observed 22.8, expected 22.6
Circular span
Mean: observed 18.3, expected 18.3
The tightest draws
A circular span of at most 8 has a probability of 2.34% for a random draw. We therefore expected 62.1 such draws; there were 52.
| Date | Winning numbers | circular | linear |
|---|---|---|---|
| 01.08.2005 | 30313233 | 4 | 4 |
| 02.05.2011 | 19202122 | 4 | 4 |
| 09.04.2007 | 3567 | 5 | 5 |
| 01.02.2016 | 25262729 | 5 | 5 |
| 06.08.2026 | 19202223 | 5 | 5 |
| 05.10.2026 | 10121314 | 5 | 5 |
| 12.02.2002 | 16171821 | 6 | 6 |
| 12.11.2002 | 9111314 | 6 | 6 |
| 25.04.2003 | 57910 | 6 | 6 |
| 01.09.2008 | 12141517 | 6 | 6 |
| 05.01.2009 | 3568 | 6 | 6 |
| 10.07.2017 | 25262730 | 6 | 6 |
What carries over from the previous draw?
For each draw, we count how many of its four numbers were already in the immediately preceding draw. The expected distribution is hypergeometric: 4 drawn from 35, of which 4 are marked.
Exact repeats
Mean: observed 0.468, expected 0.457
Neighbours of the previous draw
Average number of winning numbers that are direct neighbours (±1, circular) of a number in the previous draw, without having been drawn in it themselves.
The last draw contains no information about the next one. Replaying the latest numbers or their neighbours gives exactly the same chances as any other combination.
The slip as a heat map
The 35 numbers in five rows of seven. The darker the cell, the more often the number was drawn. The expected value is the same for all; the differences are the normal spread of a binomial distribution.
All time
2655 draws · Expected: 303.4 / Number · min 266, max 352
| 1314 | 2316 | 3295 | 4352 | 5295 | 6332 | 7280 |
| 8309 | 9271 | 10320 | 11266 | 12314 | 13285 | 14275 |
| 15303 | 16317 | 17316 | 18296 | 19309 | 20295 | 21284 |
| 22307 | 23294 | 24291 | 25315 | 26309 | 27312 | 28289 |
| 29301 | 30344 | 31301 | 32299 | 33318 | 34311 | 35285 |
Last 100 draws
100 draws · Expected: 11.4 / Number · min 4, max 19
| 119 | 212 | 310 | 412 | 515 | 615 | 78 |
| 813 | 95 | 1013 | 1113 | 1215 | 1315 | 1413 |
| 154 | 169 | 1714 | 1811 | 1914 | 2015 | 2110 |
| 2218 | 2313 | 247 | 2511 | 2610 | 2710 | 285 |
| 2918 | 308 | 318 | 3212 | 337 | 3412 | 356 |
The gap between the rarest and the most frequent number looks large, but it is exactly what you expect from 35 independent counts. Over the last 100 draws the relative spread is even larger because the sample is smaller.
How long does a number stay absent?
For each number we measure the gap in draws between two appearances. If the draws are independent, this gap follows a geometric distribution with probability 4/35 ≈ 11.4% per draw: mean 8.8, and long absences are not rare, they are the rule.
10585 gaps measured · Mean: observed 8.73, expected 8.75
The longest absence in the data is 92 draws. The probability that a given number stays absent that long is 0.00%; with 35 numbers it is bound to happen eventually. The longest current absence is 44 draws. A number that has been absent for a long time is therefore not "due": the balls have no memory.
My combination
Enter four numbers: has this combination been drawn before? Which draws came closest?
Everything runs in your browser; nothing is stored or sent. That a combination has never been drawn is normal: out of 52,360 possible combinations only a tiny fraction has ever come up.