Number statistics

The bonus number is left out of the frequency counts. It decides second prize alone and is a different kind of number from the six that decide first prize, so including it would inflate every number's apparent frequency.

The average is the number of draws multiplied by six, divided by 45. If all 45 numbers came up without the slightest bias, every one of them would sit exactly on that line. The dashed line on the chart and the 'vs average' column in the table show how far above or below it each number falls.

For each number we find the last draw it appeared in and count the gap to the latest draw. A number last seen 19 draws ago is shown as absent for 19 draws. If it appeared in the most recent draw, that is stated instead.

For odd and even, each draw is classified by how many of its six numbers are odd, from none to all six. The band distribution sums the per-number frequencies within each band of ten; because 41–45 holds only five numbers, a per-number average is given alongside so the bands can be compared, and the bars are drawn from that. A draw counts as containing a run when any two of its sorted numbers differ by one, and as a run of three or more when three such neighbours follow each other. The sum is simply the six numbers added together. The bonus number is excluded from all four of these as well.

The prize figure is what one first-prize winner received, before tax. Draws where nobody won first prize and the money rolled over carry an amount of zero, so they are excluded from the mean, the median, the lowest-ever record and the distribution by amount. Leaving them in would drag the mean down and hand the 'lowest ever' slot to a zero. In the winner-count statistics, by contrast, '0 (rolled over)' is kept as a band of its own and shown as it is.

The median is given alongside the mean because prizes in the early years were unusually large. When a handful of high values pull the mean upward, the median gives a more honest picture of what is typical. 'What each winner received, by winner count' groups draws by how many winners they had and averages the per-winner amount within each group; it is there to show the relationship between the two, not to predict anything.

A new draw is added to the dataset after each Saturday's draw, and every figure above is recalculated when it is. The calculation runs inside your browser rather than on a server, so simply reading this page records nothing anywhere.

No. Each draw is completely independent of the ones before it. The machine does not remember which balls came out last week, and the chance of any particular number appearing in a draw is fixed at 6 in 45, about 13.3%.

So why do the counts differ? For the same reason that tossing a coin 1,000 times does not give exactly 500 heads. Random draws always leave some natural wobble, and after more than a thousand draws a spread of a few dozen appearances sits comfortably inside it. As draws accumulate the gaps do not disappear — they simply shrink as a proportion of the average.

It is not. This is the classic mistake known as the gambler's fallacy. A number absent for 20 draws has exactly the same 13.3% chance next time as one drawn last week. The balls carry no memory of when they last came out.

The pleasure of seeing the record, and a way to stop dithering when you choose. The odds of first prize are 1 in 8,145,060 for every combination alike, so no statistic here raises your chance of winning.

There is one practical difference, though. A great many players pick only from 1 to 31 because of birthdays, so a line that includes numbers from 32 to 45 tends to be shared by fewer people. The odds are unchanged, but should the line win, there may be fewer winners to split the prize with.