Mega Millions
Mega Millions methodology
The math in 30 seconds
- 1
Draws are random: every valid line has the same chance of being drawn.
- 2
Players aren’t: many pick birthdays, sequences and old favorites.
- 3
When popular numbers are drawn, more winners share the prize.
- 4
Wisenum estimates how players choose, from official winner counts.
What the results show
Over the last 200 draws (since Dec 24, 2024), when the numbers drawn were among those Wisenum rates as the most played, there were on average 30% more winners for the same number of players than when they were among the least played.
95% bootstrap interval: 24% to 36%, comparing the 50 draws rated most played with the 50 rated least played. It describes past draws, not a forecast.
The estimate is updated every quarter with the new official results: it currently covers draws up to Sep 25, 2026.
Before a model is used, it is checked against official results it was not built from. Readings from a model that has not passed are never presented as validated.
Zero ticket salesWe never sell lottery tickets and don't take a cut.
Official data onlyEvery estimate comes from published winner counts, with an open method.
How the Mega Millions estimate was checked
Every figure below comes from the model files Wisenum serves for Mega Millions. Another game’s test is never shown here.
- Fitted on
- 930 official Mega Millions draws, Oct 31, 2017 to Sep 25, 2026
- What was measured
- Official winner counts in lower prize tiers, against what chance and the number of tickets predict
- How it was tested
- One untouched hold-out of 208 draws (Oct 1, 2024 to Sep 25, 2026), tested once before the model went live
- Result on draws not used for fitting
- Prediction error 45% lower than a model that ignores number patterns
- Jackpot-tier check
- -1.2% on the winner counts closest to the jackpot (five main numbers): weak evidence for exact jackpot sharing
- Popularity calibration
- Exponent 1.667 (95 % range 1.04–2.88), improved the fit in 3 of 5 chronological tests
- Tickets-in-play forecast
- Typical error 7%; 8 in 10 draws within -4% to +15% of the published figures
The history replay of a line covers every draw under the current rules; that count can differ from the draws used to fit the model.
Equal odds
Mega Millions has 290,472,336 possible lines. Every one of them has exactly the same chance of being drawn.
What Wisenum measures
Many players pick their numbers the same way: birthdays, sequences, patterns on the play slip. When those numbers come up, the jackpot and prizes are shared between more winners. Wisenum estimates how closely a line resembles these common choices — its “popularity” — from the official results published after every draw, and suggests lines that resemble them less.
What it cannot measure
Official results do not say how many people played exactly your line. Popularity is an estimate of general tendencies, not a promise of a prize or of being a sole winner.
Sources: Official results · megamillions.com ↗