How we analyse draws
Guides on this site are published by LottoScan Research, the analysis desk at Lunana Global Inc.. Everything below describes how the figures in those guides are produced, so you can decide how much weight to give them.
Where the numbers come from
Draw results, advertised jackpots, cash values, winner counts and prize-tier breakdowns are compiled from official public sources: the Massachusetts State Lottery public draw-results APIthe Mega Millions official draw-data endpoint at megamillions.comthe published draw-result and prize-table pages at powerball.comthe New York State Open Data (Socrata) lottery datasets. Nothing is copied from another statistics site. Every count in a guide is computed from that record at the time the guide is written, and the guide says which draws it covers.
Results here are unofficial. Operators correct published figures occasionally, and a compilation can lag. Verify anything that matters with the official operator or the retailer before acting on it.
Why counts start at a matrix change
This is the single thing most frequency tables get wrong. When a game changes its ball pool, every count before that change describes a different game. Powerball moved to 5 from 69 plus 1 from 26 on 7 October 2015. Mega Millions moved to 5 from 70 on 24 October 2017, and changed its Mega Ball pool again in 2025.
So each pool is counted only from the date it last changed, and the white balls and the bonus ball are windowed separately, because they do not change together. The effect is that every ball in a pool has had exactly the same number of chances, and no count needs an asterisk. A table that counts from the beginning of a game instead will always show the newest numbers as ice cold, purely because they did not exist for most of the period.
How the statistical test works
Where a guide says a spread of counts is or is not meaningful, that comes from a chi-square goodness-of-fit test against a uniform distribution across every ball in the pool, not from an impression. We report the statistic, the degrees of freedom and the p-value so the claim can be checked rather than taken on trust.
For Powerball’s white balls across 1,388 draws, that is a statistic of 80.48 on 68 degrees of freedom, p = 0.14. A perfectly fair machine produces a spread at least that uneven about one time in seven, so the gap between the most and least drawn ball is exactly what randomness looks like.
What we do not claim
We analyse draws that have already happened. We never predict ones that have not. No number is due, no combination is likelier than another, and nothing on this site improves anyone’s odds, because nothing can. The odds of any single combination are fixed by the game’s matrix.
The one honest effect of number choice is on how a prize is shared rather than whether it is won. Combinations that many players pick get split more often when they hit, which is the argument in how to pick lottery numbers.
Corrections
If a figure here is wrong, we want to know and we will fix it. Write to hello@lottoscan.co. The terms set out the limits of what this service can tell you.
For entertainment. Lottery play carries a near-certain loss of the amount staked. If gambling is causing you harm, the National Problem Gambling Helpline is on 1-800-522-4700.