How long does a group of three Powerball numbers wait to come back?
Powerball has been drawn as 5 numbers from 69 plus a Powerball from 26 since October 2015. Everything on this page counts only the 1,423 draws since that change.
Across 1,423 Powerball draws the same group of three numbers has come back 1,793 times, where chance predicts 1,772.1.
Each of those returns closed a wait, and the waits ran 469.55 draws on average where chance predicts 453.54 — a difference of 16.01 of a draw, with the record above the prediction. The shortest wait there is, a single draw, has been closed twice against a predicted 2.71, and the longest ran 1,391 draws.
What was measured one wait per return
A group of three is simply three of the pool's numbers, in no order. Draw 1423 was 0121445057, and any three of those numbers make one, which is ten groups from that draw alone. Once a group has been drawn, every later appearance of it closes a wait: the number of draws from the previous appearance to this one. A wait of one means the same three numbers arrived in the very next draw. The longest the record has closed is 165869, drawn in draw 15 and then not again until draw 1406, 1,391 draws later.
A wait is counted at the draw that closes it, so a group still standing is not counted at all and nothing has to be assumed about waits that are still open. That also settles the one trap that would otherwise sink this page: a wait longer than the record cannot be seen, so the figures here are about the waits the record was able to close, and every prediction is summed over exactly the same chances. Counted that way the record holds 1,793 closed waits out of 14,230 groups drawn.
One named group sits inside one draw with probability 10 in 52,394, or about 1.9 chances in ten thousand at the pool of 69. Left to run for ever that rate would put the average wait at 5,239 draws, which is longer than the whole record, and that is exactly why the average above comes out smaller: a run of 1,423 draws can only close the waits that fit inside it. This game has drawn from one pool of 69 numbers throughout, so that same rate stands behind every draw read. Each prediction below is still worked out at its own draw's pool, because that stays right if the pool ever changes.
The prediction is not that single rate either. Going into a draw, every group on record has been standing for some number of draws, and those standing waits are what the draw picks from. How many groups stand at any one length is made by what the earlier draws happened to do, so it moves from draw to draw even when the pool never changes, and every prediction here is summed draw by draw from where the groups actually stood. Because a draw's prediction is read off the draws before it and never off itself, the differences do not lean on one another: they add up, and so do the variances, which is what makes the ranges on this page exact rather than approximate.
The record 1,793 waits, 8 rows
Red marks a figure outside the range ordinary variation produces
| Length of wait | Chance | Waits |
|---|---|---|
| 1 to 85 draws | 222.312.54% | 20511.43% |
| 86 to 177 draws | 221.712.51% | 19710.99% |
| 178 to 278 draws | 221.212.48% | 24613.72% |
| 279 to 391 draws | 221.112.47% | 20111.21% |
| 392 to 521 draws | 220.712.45% | 23513.11% |
| 522 to 679 draws | 222.112.53% | 23212.94% |
| 680 to 889 draws | 221.312.49% | 23513.11% |
| 890 to 1,422 draws | 221.812.51% | 24213.50% |
Each row is a length of wait and how many of the record's waits ran to it, beside the number chance predicts and the share of all the waits read it comes to. The bands were cut from the prediction alone, so that chance puts a similar number of waits in each: between 220.7 and 222.3 across the eight rows. That matters, because a row covering a wide range of lengths would be the biggest row on the table for that reason alone, and cutting the bands this way leaves every row on the same footing. The biggest row in the record is 178 to 278 draws, while chance puts 1 to 85 draws first. The predictions add up to 1,772.1 against 1,793 waits read, so the table compares the shape of the record against the shape chance predicts rather than its size.
Nothing in the table is printed in red, and that is deliberate rather than an omission. A row here gathers several standing classes at once, and the groups inside them are not independent of one another: one draw can close up to ten waits at once, and the groups involved share numbers. Expectation stays exactly additive, which is why every prediction above is exact, but the spread of a row's count would need the covariance between groups that share numbers and is not honestly available. A range would have to be invented, so none is given. The figures that do carry ranges are the two counts read draw by draw: how many waits a draw closes, and how long they were in total.
That second one is the range carrying form of the average. Across the record the waits closed came to 841,898 draws of accumulated waiting against a predicted 803,735, which is 1.63 standard errors above it. The average itself is one of those totals divided by the other, and both of them move, so no range is put on the ratio. The count of waits closed sits 0.50 standard errors above it.
At the far end, the longest wait the record has closed ran 1,391 draws. That was 165869, back in draw 1406 on 2026-09-12 after being drawn in draw 15 on 2015-11-25.
Waits that long or longer were predicted 0.76 times across the record and happened once. Nothing in this paragraph is printed in red, and that too is deliberate: the longest wait was picked out for being the longest, so marking it as unusual would only repeat how it was chosen. The longest of 1,793 waits is above the average by definition, and the honest reading of it is the pair of figures just given.
Recent trend last 100 draws, then blocks of 20
Counted as waits closed, the last 100 draws cannot be compared with the first 100: the record had almost nothing on it to begin with, so those early draws closed seven waits against the later 229 waits. Divided by the groups standing at each draw the two become comparable, and that is the second and third figures above. The rate is what chance fixes; the count is what the length of the record decides. The average wait closed is the other half of the same point. Over the last 100 draws it runs 642.64 draws against a predicted 656.99, which is longer than the 469.55 across the whole record, and chance predicts that: a longer record has older groups standing in it, so the waits it closes have more room to be long.
Returns per ten thousand standing groups, one point per block of 20 draws
result for the blockrange ordinary variation allowsoutside that range
Oldest block on the left, newest on the right. Each point is the share of the groups standing at those draws that came back in them, and the band is what chance predicts for those same draws, worked out from each draw's own pool. Reading the share rather than the count is what makes the chart flat instead of rising: the count climbs as the record fills up, while the share has nowhere to go. No point sits outside the band here. Across ten blocks, chance alone puts 0.5 of them outside, so a point outside the band would not be explained by the number of blocks read.
Applied to this draw draw 1423 · 2026-10-21
Draw 1423 was 0121445057 from a pool of 69, and 12,431 groups of three were standing on the record going into it. Of the ten groups it holds, four closed a wait, where chance predicted 2.37, which is 1.12 standard errors above it. That leaves six groups here that the pool had never produced until now.
The longest of the waits it closed, and where those groups had been:
Those four waits average 618.75 draws, where chance predicted 679.39 for this draw from where the groups stood going into it. The list is sorted by length, so its first line is where a sorted list has to begin rather than a group that was owed anything, and it carries no red mark for the same reason.
Related statistics
Share this page
If this report does not answer your question, tell us what you need and we will look at building it. Contact