Next draw: …
Play this
Our lines
The lines we play, checked against the latest draw. Pick an older draw or paste extra lines to check those too.
The math behind it
Skip or play: the numbers behind the call
What one $2 line returns after the cash option, taxes and the chance of sharing. It never pays for itself, so Play means 40¢ or more back per dollar, which is as good as these games get.
How this is calculated
Expected value is every prize times its probability, summed. The jackpot term is the cash option, after federal and state tax, times your expected share if you win, divided by 292,201,338.
EV = Σ prize × P(prize)
share = E[1 / (1 + K)], K ~ Poisson(tickets / 292,201,338) = (1 − e−λ) / λ
Other tickets are independent draws from the same ~290M combinations, so the number of other people holding your winning line is Poisson. With 19.5M tickets in play, λ is 0.067 and there is a 93.5% chance nobody else has your jackpot line.
Lower tiers are fixed prizes, so they are just prize over odds. Prizes of $5,000 or more are taxed; the rest are not. Powerball's Power Play is scored separately as its own $1 bet: the jackpot is never multiplied, the $1M tier becomes a fixed $2M, and everything else is multiplied by a ball drawn from 42 (24 twos, 13 threes, 3 fours, 2 fives), or 43 with a 10x ball when the jackpot is $150M or under. Mega Millions bakes a multiplier into every $5 ticket (2x to 10x, averaging 3x), so its lower tiers are scaled by 3 inside the ticket's own value.
The tickets-in-play guess is a straight line fitted by eye to Lotto Report's 2026 figures, with Monday draws selling more because they carry the weekend. Override it with the real number when Lotto Report posts it.
Kelly criterion: the fraction of your bankroll to stake is (bp − q) / b. For any negative-EV bet it is zero. The footnote on every verdict says so.
The odds
Computed from the 5-of-69 and 1-of-26 matrix, not copied from a PDF. Any prize: 1 in .
Where the numbers come from
Contrarian picker
Every line has the same chance. These are lines fewer people hold, so a win is shared with fewer hands. The crowd score is how many other holders to expect compared with a Quick Pick: 0.5 means half as many.
The crowd model
Every published study finds the same shapes. Numbers 1 to 31 are over-picked because they are dates; 1 to 12 doubly so. Seven is the most chosen number in every draw the Israeli lottery studied, and 3, 11, 13, 17, 21, 23, 33 and 42 trail it. People scan the slip top to bottom, so the high numbers near the bottom get picked less. People avoid neighbouring numbers even though about half of real draws contain a pair. Evenly spaced lines and famous sequences are held by thousands.
Each number gets a weight for those biases, scaled so the average over the pool is 1. A line's crowd score is the product of its six weights, with multipliers for all-under-32, evenly spaced, neighbours (which lowers it) and famous lines. The picker samples against those weights and rejects anything above your max.
The weights are rough and documented in docs/popular-numbers.md. The shape is right; the exact values are ours to tune. Most Powerball tickets are Quick Picks, which dilutes the effect, so the edge is real but smaller than the UK numbers suggest.
History and the randomness check
Only draws on the current matrix are counted. The charts are for fun. The chi-square test is the point: if the drums ever develop a bias, this is where it would show.
White balls
Powerball
Why "hot" and "overdue" mean nothing
Each draw is independent; the drum does not know what came out last week. With about 1,500 draws and 69 balls, every ball has been drawn roughly 110 times and the natural spread is about ±10 either side. The leaders and laggards you see are exactly what fair randomness looks like in a sample this size.
The chi-square test adds up (observed − expected)² / expected over every ball and asks how often a fair drum would produce a spread at least that large. That is the p-value. Below 0.01 would be worth a hard look. Expect something boring, and if it ever isn't, we will be the first to know.
χ² = Σ (O − E)² / E, df = 68 (whites) or 25 (Powerball), p = Q(df / 2, χ² / 2)