Mines strategy: the honest maths
How Mines pays
The round draws m tile positions without replacement. After k safe reveals the payout is 0.99 × Π (25 − i) ÷ (25 − m − i) for i from 0 to k − 1, floored to four decimals. The product is one over the probability of surviving k picks, so payout times survival is 0.99: the ladder is the odds inverted, with 1% taken off.
What the mine count does
| Mines | First pick safe | Pays after 1 | Picks to reach 2× | Survive that far | Pays | Clear the board, one in | Board pays |
|---|---|---|---|---|---|---|---|
| 1 | 96.0% | 1.0312× | 13 | 48.0% | 2.0624× | 1 in 25 | 25× |
| 2 | 92.0% | 1.0760× | 8 | 45.3% | 2.1838× | 1 in 300 | 297× |
| 3 | 88.0% | 1.1250× | 6 | 42.1% | 2.3498× | 1 in 2,300 | 2,277× |
| 5 | 80.0% | 1.2375× | 4 | 38.3% | 2.5848× | 1 in 53,130 | 52,599× |
| 10 | 60.0% | 1.6500× | 2 | 35.0% | 2.8285× | 1 in 3.3 × 10^6 | 3,236,072× |
| 15 | 40.0% | 2.4750× | 1 | 40.0% | 2.4750× | 1 in 3.3 × 10^6 | 3,236,072× |
| 20 | 20.0% | 4.9500× | 1 | 20.0% | 4.9500× | 1 in 53,130 | 52,599× |
| 24 | 4.0% | 24.7500× | 1 | 4.0% | 24.7500× | 1 in 25 | 25× |
More mines makes the ladder steeper: fewer picks to 2×, but each pick riskier. With 3 mines you need 6 safe picks for 2× and survive that 42% of the time; with 10 mines it is 2 picks at 35%. The two 2× bets are worth the same. The mine count is a variance dial with no expectation attached.
What staking plans do
“One tile and out” at 3 mines is a 88% bet paying 1.1250×: a Dice roll at 88%, to the fourth decimal. People pair it with martingale because it “almost always wins”. It does; it also pays 1.125×, so a loss cannot be recovered by doubling. Getting back to even after a loss at that payout needs L ÷ (x − 1), which is 9 times the previous stake.
| Losses in a row | Lost so far | Stake needed next | Fits a 1,000 bankroll? |
|---|---|---|---|
| 1 | 10 | 80 | yes |
| 2 | 90 | 720 | yes |
| 3 | 810 | 6,480 | no |
| 4 | 7,290 | 58,320 | no |
The simulator's martingale doubles rather than multiplying by 9, so it under-recovers at this payout and drifts instead of cliff-diving; either way the return on turnover reads −1%. Run one-tile-and-out with martingale, then flat-bet five picks at 3 mines (49.6% for 1.9973×), then martingale three picks at 10 mines (19.8% for 5.0043×). Three shapes, one expectation.
Corner tiles, patterns and “the mines are never next to each other” are the other folklore. Positions are drawn uniformly without replacement from the seed; every tile is equally likely to hold a mine on every round, and the layout is fixed before you see the grid.
Understood the maths? Craze runs the real thing.
Same published odds, real stakes. PlinkoLab is sponsored by Craze; whether to play for money is entirely your call.
Questions
- How many mines should I play in Mines?
- Any number returns 99% of the stake on average, so the honest answer is: the number whose swings you can stand. With 1 mine the first pick is safe 96% of the time and pays 1.0312×; with 24 mines it is safe 4% of the time and pays 24.75×. Survival times payout is 0.99 in both cases.
- Is picking one tile and cashing out a good strategy?
- It is a Dice roll. One pick with 3 mines is a 88% chance paying 1.125×, which is exactly what Dice pays at 88%. It has the same 99% expectation as picking ten tiles; it just loses less often and wins less each time. Adding martingale to it hits a wall fast, because recovering a loss at 1.125× needs 9 times the previous stake.
- Do the mines move after I start clicking?
- No. All mine positions are drawn without replacement from the HMAC-SHA256 floats when the round starts, before the grid is shown. Clicking only reveals. Reveal the server seed by rotating and the Verify link recomputes the whole layout, which is how you can check that nothing moved.
Keep going
Last reviewed 2026-09-27