Mines: The Anatomy of Risk

Every strategy hits the same fixed edge

A Mines round is simple: a 5×5 grid hides some mines, you reveal tiles one at a time, and each safe tile lifts your multiplier. Cash out whenever you like; touch a mine and you lose everything. The payout for surviving n clicks is set to RTP ÷ P(survive n): that construction is exactly why the survival probability cancels out of the expected value.

The consequence is stark and worth internalising before any chart: across all 300 ways to play (every mine count, every cashout depth) the expected return is identically −1%. There is no clever cashout point, no lucky mine count. Strategy moves you along a curve of identical expected value; it never lifts you off it. So the only meaningful question is what kind of volatility you are buying.

A week of real bets, not a thought experiment

In the seven days from 22–28 June 2026, twenty-five crypto casinos booked 892,777 Mines bets between them, roughly 127,540 a day, holding within a tight band all week. This is one of the most-played originals in the category, and every one of those bets met the exact same −1% edge dissected below.

The volume is also strikingly concentrated: gamdom alone took 27.3% of all bets, and the top five operators accounted for nearly two-thirds of the market. A long tail of smaller books fills out the rest.

How far the reckless player gets

Picture the player who never cashes out and just keeps clicking until a mine ends them. How many safe tiles do they uncover first? The answer follows a clean law: (25 − M) ÷ (M + 1). Our simulation lands right on it. At 10 mines the average run is just 1.36 safe tiles before the board bites back.

The decay is hyperbolic, not linear: going from 1 to 2 mines nearly halves your expected run (12 → 7.7 tiles), while the last few mine levels barely move the needle. Risk piles up fastest at the start.

Most rounds end faster than average

Averages hide the texture. Drag through the mine counts below: at low settings the bust depth spreads out into a long tail, but from 13 mines upward the median collapses to zero: more than half of all rounds end on the very first click. The mean stays positive only because rare deep runs prop it up. What a player actually experiences at high mine counts is repeated instant death.

The risk dial: one click, a 24× swing

Hold the cashout at a single click and walk the mine count up, and you trace a dial from near-certainty to near-lottery. One mine pays 1.03× at a 96% win rate; ten mines pays 1.65× at 60%; twenty-four mines pays a 24.75× on a single reveal you'll win just 4% of the time. Every one of the 300 configurations sits on the same iso-edge frontier: proof, again, that you only ever slide along the curve.

Volatility doesn't care how many mines you set

Here is the finding that surprised us. Substitute the payout rule into the variance and the mine count vanishes entirely: a single bet's volatility is σ = √(RTP · (mult − RTP)): purely a function of the target multiplier. Plot all 300 configs, coloured by mine count, and they fall onto one curve across seven orders of magnitude. A 5× cashout carries the same risk whether you reached it with one mine and patience or five mines and nerve. Mine count is just the dial; the multiplier is the risk.

Every mine count and cashout point, mapped in one grid

Every cell is a way to play: mine count down the side, clicks-before-cashout across the bottom, colour for the multiplier on offer. The bright upper-right corner is where the moonshots live: clearing a heavily-mined board for six- and seven-figure multipliers at probabilities so small that, in 100,000 simulated attempts, the very biggest paid out exactly once. The fair 1% edge only materialises over astronomically many rounds; the lived experience of chasing that corner is near-total loss. Put against the real volume from Fig. 01: even if every one of the 892,777 bets booked across all 25 casinos in a week chased that 5,148,297× ceiling, it would be expected to pay out only about once every six weeks. Hover any cell to read its odds.

Two roads to a big win, both with a ceiling

The week's largest payouts arrived two different ways. One is brute force: Nono9806's $141,944 bet on stake (the single biggest wager of the week) returned $501,872 at a modest 3.54×. The other is a sharper multiplier on a mid-size stake: 500casino's biggest win turned a $34,088 bet into $924,015 at 27.11×. Same fixed edge, two ways to ride the variance.

But notice the ceiling. The largest multiplier anyone actually hit all week was 27.11×, a universe away from the six- and seven-figure multipliers in Fig. 06's bright corner. And here is the quiet validation: every multiplier on these boards (27.11×, 7.96×, 3.54×, 1.74×) lands exactly on a cell of our computed paytable. The live game and the model are the same arithmetic. Of the five biggest bets, four cashed and only one busted; yet 500casino, with under 5% of all bet volume, booked four of the five biggest wins.

All wagers placed in SOL; USD shown at the time of bet. A 0.00× / $0 row hit a mine and lost the stake. Every winning multiplier here matches an exact cell of the paytable in Fig. 06.

The one number the casino actually controls

Everything up to here assumed a 99% RTP. But RTP is the operator's dial, and it varies more than players notice. The identical game of Mines runs at 100% on Duel and Gamdom, 99% on Stake and Shuffle, 98% on BC.Game, and just 96% on Rainbet and Roobet. Because every multiplier equals RTP ÷ P(survive), lowering the RTP simply scales down every payout on the board: a 96% table pays a flat 3% less than Stake on every multiplier, and 4% less than a 100% table. Jeluca's 27.11× hit that paid $924,015 would have returned roughly $27,800 less at a 96% house.

Flip RTP into its mirror image (the house edge, 1 − RTP) and the commercial stakes sharpen. On every $1,000,000 wagered, a 96% table expects to keep $40,000; a 99% table $10,000; Duelbits' 99.44% just $5,600; and the 100% tables, nothing at all. Roobet's edge is four times Stake's.

The market makes this stranger still. The two largest books by volume, Gamdom and Duel, are precisely the two giving the game away at 100%, together nearly 40% of every Mines bet in our sample week. The edge isn't where the volume is: the leaders compete by zeroing it out and monetising elsewhere, while smaller books lean on a 4% take.

Why nobody notices a bad RTP until it's too late

To feel what RTP really does, we sat 1,000,000 players at the same 3-mine / 3-click bet and let them play, flat-staked, under each casino's RTP. The win odds are identical everywhere, only the payout size differs, so this isolates the edge and nothing else. The result splits cleanly in two.

Short-term, the edge is invisible. For roughly the first 40 rounds, every RTP traces the exact same line: a player at 96% Roobet is precisely as likely to be ahead as one at 100% Gamdom. Being in profit takes a whole number of wins, and that bar doesn't budge until the edge slowly piles up. You cannot feel a bad RTP in a short session, which is exactly why it is so easy to run one.

Long-term, it is destiny. Once the lines separate they fan out brutally. At 100% RTP, half the players stay ahead forever: a fair coin. At 96%, the share in profit collapses to 2.9% by 1,000 rounds and 0.4% by 2,000. Same game, same player, same luck: only the dial moved.

The whole paytable, on one card

Rows are mines; columns are the number of diamonds you reveal before cashing out; each cell is the multiplier you'd be paid. These are the exact Stake-standard figures (99% RTP). Toggle to win chance and, unlike the rounded values that circulate on forums, every percentage is computed straight from the combinatorics, so a 1-mine / 1-diamond reveal reads a true 96%, not 96.12%. Hover any cell for both numbers in full.

Colour runs cool (low multiplier, high win chance) to hot (long-odds, high multiplier). Blank cells are impossible: mines plus diamonds can't exceed the 25-tile board. First row/column stay pinned as you scroll.