Stake Mines Strategy: The Math Behind Every Setting

Stake Mines looks like a game of decisions: how many mines, how many tiles, when to cash out. The math says those decisions change only one thing, the shape of your results. None of them changes what the game costs you. Below are the exact numbers behind that, and what they mean when you pick a setting.

How Mines pays

The board has 25 tiles and you choose between 1 and 24 mines. The chance of revealing k safe tiles in a row with m mines on the board is the number of safe combinations divided by all combinations:

P(win) = C(25 − m, k) / C(25, k)

Stake pays the inverse of that probability, minus its 1% house edge:

multiplier = 0.99 × C(25, k) / C(25 − m, k)

Two quick checks against the game: one mine and one tile gives 0.99 × 25 / 24 = 1.03x, and twenty-four mines with one tile gives 0.99 × 25 = 24.75x, the largest single-tile payout on the board.

Multiplier and win chance by setting

The game rounds the multiplier it displays to two decimals, so what you see on screen can differ from this table by 0.01.

Look at the diagonal: 3 mines with 5 tiles pays exactly what 5 mines with 3 tiles pays, 2.00x at 49.6%. The formula is symmetric in mines and tiles, so "more mines, fewer clicks" and "fewer mines, more clicks" are the same bet. The only difference is how long the round takes.

Every setting costs the same

Multiply any multiplier in the table by its win chance and you get 0.99. Whatever the setting, a round returns 99% of the stake on average: the expected loss is 1% of each bet.

The less obvious consequence is that the 1% is charged once per round, not once per tile. After k safe tiles, the next tile survives with probability P(k+1) / P(k) and would pay 0.99 / P(k+1). The product is 0.99 / P(k), exactly what cashing out pays right now. So once a round has started, continuing or stopping never changes your expected result. What changes it is the number of rounds you play and the stake per round.

The "safe" setting is the one that loses most reliably

To see what this does to real sessions, we simulated 200,000 sessions of 1,000 rounds at 1 unit per round, for three settings with the same 1% cost:

All three lose about 10 units per 1,000 rounds on average. The one-mine setting feels safe because it wins 96 rounds out of 100, but a single loss erases 32 wins, and over a session the edge grinds through almost every time: only one session in seventeen ends ahead. The riskier settings lose the same amount on average, spread differently: deeper bad sessions, and many more good ones.

Neither is better. They are two distributions of the same cost, and the right one is the one whose bad case you can afford.

How many wins does one loss cost?

This is why progressions look so tempting on low-risk settings, and why they break there first. Recovering a loss in one round at 1.03x means betting 32 times the amount you just lost. Two losses in a row and the bet is a thousand times the base.

Mines predictors do not exist

The mine positions are fixed before your first click. They come from HMAC-SHA256 of Stake's server seed, your client seed and the bet's nonce, and until you rotate your seeds you only ever see a SHA-256 hash of that server seed. Predicting a board would mean recovering the seed from its hash, which is exactly what SHA-256 is built to prevent.

A "predictor" that highlights tiles is showing random tiles. One that asks for your Stake login or API token is after your account. The full mechanism, with a script that recomputes any past board from your own seeds, is in our provably fair guide.

Choosing a setting

• Start from the bad session you can accept, for example the 5th-percentile column above, and pick the setting whose bad case fits.
• Keep the stake per round small against your bankroll. With a fixed 1% per round, the number of rounds is what drives the expected loss.
• Treat mine count and tile count as one choice: 3 mines × 5 tiles and 5 mines × 3 tiles are the same bet.
• Cash-out timing inside a round changes variance, never the edge.

Automating Mines

In SSPilot's Mines mode you set the mine count (1 to 24), how many tiles to reveal before cashing out, what happens to the bet after a win or a loss, and stop conditions on profit, loss or number of rounds. Everything above applies unchanged. Automation plays faster, and more rounds means your results converge on the 1% cost sooner, so set the loss limit before you press start. Progressions and bankroll rules that apply to every game are covered in the Stake strategy guide.

How these numbers were produced: multipliers and win chances are exact combinatorics with Stake's 1% edge. The session table is a Monte Carlo simulation of 200,000 sessions of 1,000 rounds per setting, with a fixed random seed.