Home › Game Guides › Mines

Mines Gambling Explained: The Math Behind Every Payout

Mines is a 5×5 grid of 25 tiles, a number of hidden bombs you set yourself, and a multiplier that climbs with every safe tile — and it contains no decisions. The bomb layout is fixed before your first click, one combinatorial formula forces every multiplier on the board, and every configuration — 1 mine or 24, one pick or twenty — returns the identical fraction of your stake. The mine count buys variance, not value. Only two things can move your edge, and neither of them is on the grid: the operator’s RTP, and a max-win cap where one applies.

What is Mines, and are the multipliers fair?

Mines is a crypto-casino grid game: 25 tiles, a player-chosen number of hidden bombs, a multiplier that climbs with each safe tile, and a cash-out button you can press at any point — and its multipliers are mathematically fair in the sense that each one is the exact inverse of its own probability, scaled by the operator RTP. That is the whole design. Nothing about the payout table is arbitrary; you can reconstruct every value in it from scratch with a calculator.

What the game feels like is a test of nerve and pattern-reading. Players talk about hot corners, avoiding the centre, clicking the same four tiles every round, quitting after two safe picks because the board "feels loaded". None of it touches the math. The bomb positions are fixed by a shuffle that runs before your first click, so the board is already written when you start reading it. Everything below is the arithmetic that follows from that, and it comes to an unusual conclusion for a casino game guide: there is nothing here to optimise.

What is the one formula behind every Mines multiplier?

The chance of surviving k picks on a 25-tile board holding m mines is P = C(25−m, k) ÷ C(25, k), and the multiplier the game offers you is simply RTP ÷ P. Both parts are published openly by the calculator tools built around the game — 100RTP states the fair multiplier as C(25, s) ÷ C(25 − N, s) and MinesCalc states survival as C(25 − M, s) ÷ C(25, s), which are the same identity written from the two ends.

The combinatorial form is compact but hides the mechanism. Written as a running product it becomes obvious what the game is doing to you:

Here is the core table, computed for the standard 25-tile board at a stated 99% RTP. Read the last two columns first: the expected return never moves.

Table: 16Best analysis, computed from P = C(25−m, k) ÷ C(25, k) on a 25-tile board, with the offered multiplier at an assumed 99% RTP (the figure Stake and BC.Game publish for their in-house Mines). The fair-multiplier column matches the published 100% RTP reference chart at 100RTP.games to three decimals.

Readers of our crash gambling explainer will recognise this immediately: it is the same structural result in different clothing. On crash, P = RTP ÷ m and the payout is m, so P × m = RTP and every cash-out target returns 0.970 on a 97% game. On Mines the probability is combinatorial rather than a simple reciprocal, but the cancellation is identical. This is not a new discovery about Mines; it is the defining property of the whole instant-game family, and Plinko behaves the same way across its risk levels. What differs between the three games is only how violently the outcome distribution swings.

The six-item Mines checklist, ordered by how much each item moves the math

Six parameters describe a Mines round completely, and only two of them can change your expected loss. The other four change how the loss arrives. They are listed here in the order the math cares about, which is close to the reverse of the order players usually think about them.

1. The operator RTP — the only lever that changes the edge, and the one you cannot influence at the grid. 96% costs four times what 99% costs.
2. The max-multiplier cap — invisible in the interface, and capable of gutting the RTP of the highest settings without changing the advertised number.
3. The mine count — changes volatility by four orders of magnitude and the house edge by nothing.
4. The number of tiles you intend to clear — same story, and the parameter most often decided mid-round when it should be decided before.
5. Auto cash-out availability — enforces the plan you already made; alters no probability.
6. Provably-fair verification — proves the board was not moved under you; proves nothing about profitability.

1. Does the mine count change the house edge?

No. Setting 1 mine or 24 mines produces exactly the same expected return — 0.990 of stake at a 99% RTP — because the multiplier is derived from the probability the mine count creates. Raising the mine count raises the payout by precisely the amount it lowers your chance of collecting it. What it does change, enormously, is variance: the width of the distribution of outcomes around that fixed average.

Variance is the honest way to describe what "riskier" means here, and it has a closed form. Since a round either returns RTP ÷ P (with probability P) or zero, the standard deviation of the return per unit staked is SD = RTP × √((1−P) ÷ P). Every row below has an expected return of 0.990. Only the spread changes.

Table: 16Best analysis, standard deviation computed as 0.99 × √((1−P) ÷ P) from the survival probabilities above. Every row shares an identical expected return of 0.990 of stake.

Set against our Plinko numbers, where high risk carries roughly 20× the standard deviation of low risk at the same 99% RTP, Mines offers a far wider dial: the 10-mine, 10-pick setting alone is 162× the volatility of the single-mine single-pick setting. The word "riskier" in Mines discussion nearly always means this and only this. It never means a worse deal, and it never means a better one.

Why do 3 mines and 5 mines pay the same 1.9974x?

Because the formula is symmetric in the two numbers you set: swapping the mine count and the pick count leaves the probability, and therefore the payout, completely unchanged. Three mines with five picks and five mines with three picks are the same bet. So are one mine with twenty-four picks and twenty-four mines with one pick — both are a 4% shot paying 24.75x.

This matters more than it looks. It means the "risk level" of a Mines round is not the mine count at all — it is the pair, and the pair is commutative. A player who believes 10 mines is aggressive and 3 mines is cautious is describing a habit, not a setting: 3 mines with 12 picks (12.43% survival) is a far wilder bet than 10 mines with 2 picks (35.00%). The number on the mine selector carries no information on its own.

2. How many tiles do you intend to clear, and why decide before you start?

Because the multiplier ladder is designed so that each additional pick looks cheap while the cumulative survival probability collapses — and the interface only ever shows you the per-click view. This is the single most common error in Mines coverage: quoting a per-tile safety figure as if it described the round.

Table: 16Best analysis. Per-click chance at pick i is (25−m−i+1) ÷ (25−i+1); cumulative survival is the running product. Both columns are correct; only the cumulative one describes the bet you are actually making.

The end of the ladder is worth showing, because it is where the headline numbers come from. Clearing an entire board — every safe tile, no cash-out — has probability exactly 1 ÷ C(25, m):

Table: 16Best analysis, full-clear odds computed as 1 ÷ C(25, m) with payout C(25, m) × 0.99. Reported payouts from FreeTips (March 2024) and CrashGamblingSites (August 2025), both describing wins on Stake in-house Mines.

3. Is the operator RTP published, and what does 97% instead of 99% cost?

RTP is the only parameter on this page that changes your expected loss, and the figures on offer run from 96% to 99% depending on who built the game — a spread that makes the worst version four times as expensive as the best. Only one of those numbers comes from an operator rather than a tracker. Stake’s own game page states 99% RTP and a 1% house edge for its in-house Mines, and Casino.band’s tracker (updated 1 May 2026) puts BC.Game at 99%, Roobet, Spribe and Turbo Games at 97%, and SmartSoft at 96%. Two of those figures are contested by other trackers — Win.gg puts BC.Game at 98%, and MinesCalc and Roobet’s own listings put Roobet at 96%. The table below uses the Casino.band figures throughout so the comparison is like-for-like; the conflicts are unpicked in the methodology section.

Table: RTP and house-edge columns from Casino.band’s Mines RTP tracker (updated 1 May 2026); Stake’s 99% is also stated on Stake’s own Mines game page. Roobet’s row is disputed — MinesCalc and Roobet’s own listings show 96%, which would put it at $4.00 an hour on the same assumptions. The three right-hand columns are 16Best analysis, computed as stake × rounds × house edge at an assumed 100 rounds per hour — our assumption, not a published rate.

4. Does an auto cash-out change anything?

No — an auto cash-out that stops the round at a preset number of safe tiles removes execution error and nothing else. It cannot change P, cannot change the multiplier, and cannot change the 0.990 expected return. Its only function is to enforce the value of k you already committed to, which is the whole reason it is worth using if you are playing at all: the failure mode in Mines is not a bad decision, it is the absence of a decision.

The equivalent feature on crash is exactly as neutral, and for exactly the same reason. Where the two games differ is that a crash round gives you one continuous exit window under time pressure, while a Mines round gives you a discrete exit after every click, with no clock. Mines therefore has more decision points and a slower rhythm — which, as the hourly-cost section shows, makes it cheaper per hour and no cheaper per dollar wagered.

5. What does a max-multiplier cap quietly do to your RTP?

Where an operator applies a maximum-win cap, it truncates the top of the payout distribution and can drop the effective RTP of the highest settings from 99% to under 1% — while the advertised RTP figure stays at 99%. This is the one item on the checklist that can make the house edge vary by mine count, and it does so through the terms and conditions rather than the math.

Stake’s in-house Mines has demonstrably paid above 5,000,000x, so no meaningful cap binds there. Spribe’s widely distributed version is commonly listed with a 10,000x maximum win. If a cap of that size applies, the arithmetic is brutal, and it is the same mechanism as a max-cashout cap on a casino bonus — the kind we quantify in our wagering requirement guide.

Table: 16Best analysis, computed as EV = P × min(0.99 ÷ P, 10,000). This is a conditional calculation: it applies only where a 10,000x cap is in force. Stake has paid 5,148,297x on Mines, so no such cap binds there; Spribe-supplied versions are commonly listed at a 10,000x maximum win. Always read the specific game’s max-win term.

The practical reading: a cap is harmless at ordinary settings and catastrophic at the settings the game markets hardest. Any configuration whose payout would exceed the cap is being sold at a headline RTP it does not deliver. On a 99% board, the cap starts biting at 4 mines with 21 picks (12,524x) and bites at every mine count from 4 to 21 somewhere on the ladder.

6. How do you verify a Mines board was provably fair?

The mine layout is generated before your first click from three inputs — a hashed server seed the casino commits to in advance, a client seed you control, and a nonce — and after you rotate seeds you can reproduce the exact board yourself. Stake’s provably-fair implementation documents the byte-generation step — HMAC-SHA256 over the server seed, client seed and nonce, converted to a stream of floats between 0 and 1 — and the verifier write-ups built on it describe the Mines step the same way: those floats drive a Fisher-Yates shuffle of the tile indices 0 to 24, and the first m positions in the shuffled order are the bombs.

Two consequences follow, and only one of them is comforting.

• What it guarantees: the board was fixed before you touched it and could not be altered mid-round. The same three inputs always regenerate the same permutation, so any player can re-derive their own board after the server seed is revealed. This is the strongest integrity guarantee available in online gambling, and we walk through the verification step by step in our provably fair guide.
• What it does not guarantee: anything about your expected return. The RTP multiplier is applied after the shuffle, at the payout table. A perfectly verifiable board still pays RTP ÷ P. Provable fairness proves honesty, not profitability — the same conclusion we reach on crash and Plinko.

There is one genuine, checkable point here that touches the spine. Because the full permutation is fixed before the first click, the tiles are exchangeable: tile 1 and tile 17 have identical probabilities of being a bomb, and they stay identical no matter what pattern the previous rounds showed. Every "hot tile", "avoid the corners" and "the centre is safer" heuristic is describing a uniformly random permutation. Third-party builds are a different case, and the sources disagree on how different: Casino.band lists the Spribe, Turbo Games and SmartSoft versions as not provably fair in the seed-verification sense, while Spribe-facing listings advertise a SHA-256 hash-chain verification of their own. Either way it is a real difference in what you personally can check, and no difference at all in the underlying odds.

Does Martingale work on Mines?

No, and on Mines it fails in an extra way that does not apply to crash: the closest thing the game has to a 2.00x bet actually pays 1.9974x, so the doubling sequence stops breaking even before the streak gets long enough to matter. The natural Martingale setting is 3 mines with 5 picks (or its symmetric twin, 5 mines with 3 picks), which survives 49.57% of the time — a 50.43% loss rate, close enough to a coin flip to tempt anyone.

Table: 16Best analysis, doubling from a $1 base at the 3-mine, 5-pick setting (1.9974x, loss probability 0.50435). Cumulative staked is 2^N−1; net profit on recovery is 1.9974 × 2^N−1 − (2^N−1).

How much does Mines cost per hour?

Expected loss per hour = stake × rounds per hour × house edge — roughly $1 an hour at $1 a round on a 99% version, and $4 an hour on a 96% one. No published figure exists for how many Mines rounds players complete per hour, so 100 is our stated assumption: a round is several clicks plus a cash-out, which puts it between crash’s pace and a slot’s. Treat the hourly numbers as scaling, not measurement.

Table: 16Best analysis, computed as stake × rounds × house edge. Rounds-per-hour rates are assumptions consistent with our crash, Plinko and RTP pages, not measured play rates. Comparisons hold only at equal stake per round.

Make it physical: a $1 stake, 100 rounds an hour, three hours an evening, three evenings a week on a 97% version is $27 a week — about a streaming subscription and a coffee — and that is the average, arriving as a long series of small wins interrupted by losses whose size depends entirely on where on the volatility table you were sitting.

Why do published Mines numbers disagree?

Because five different things get quoted under the same words, and only one of them is defined by the formula. Every discrepancy we found in the sources for this page falls into one of these five buckets, and knowing them lets you check any Mines page — including this one — in about a minute.

• RTP published by the operator, or merely tracked. Only one figure on this page comes from an operator: Stake states 99% RTP and a 1% house edge on its own Mines game page. The rest are third-party tracker numbers, and they do not agree. Casino.band (updated 1 May 2026) lists BC.Game at 99%, Roobet, Spribe and Turbo Games at 97%, SmartSoft at 96%. Win.gg puts BC.Game at 98% and gives a general 96–98% range with a 99% maximum. MinesCalc and Roobet’s own listings put Roobet at 96%, not 97%. Two live conflicts, and we cannot settle either from outside: the authoritative figure is the one in the specific game’s own info panel on the day you look, and where a tracker and an operator disagree, believe the operator. Every multiplier on this page assumes 99%. At 97%, divide each of our offered multipliers by 1.0206; the probabilities are unaffected.
• Multipliers quoted gross or net of the stake. A 1.9974x payout returns 1.9974 units on a 1-unit stake — a profit of 0.9974 units. Charts that quote "profit multiplier" and charts that quote "total return" differ by exactly 1.00x and are frequently mixed within a single article. When a chart’s 3-mine, 1-pick value reads 0.13x rather than 1.13x, it is a profit chart.
• Win chance quoted per tile instead of cumulative. The most common error by a wide margin. With 3 mines the first click is 88% safe and the fifth click is 85.71% safe, but the five-pick round survives 49.57% of the time. A page quoting "88% win chance" for a five-tile plan is out by a factor of 1.78. Both figures are true; only the cumulative one is the bet.
• Grids other than 5×5. Every number on this page assumes 25 tiles. Win.gg notes variants configurable from 3×3 up to 9×9 — Hacksaw’s Dare2Win across that whole range, Turbo Mines at 3×3, 5×5, 7×7 and 9×9 — and the same nominal setting means completely different things across them. A 17-fold spread on one setting:

Table: 16Best analysis, computed as C(n−m, k) ÷ C(n, k) for board size n. Grid sizes are the ones Win.gg names as configurable; the probabilities and payouts are ours, not a published paytable for any specific product. The house edge is 1.00% in every row — grid size changes the payout and the odds together, exactly as mine count does.

• House edge quoted per round or per tile. The 1% edge is a property of the round, applied once at the payout. It is not charged per click. A page implying that each additional tile costs another 1% is describing a game that does not exist — and the survival table above shows why the confusion is easy: each click does cost you probability, just not edge.

One further conflict worth naming, because it produced the most-repeated wrong number in Mines coverage: several write-ups attribute the 5,148,297x maximum multiplier to a 24-mine board. C(25, 24) = 25, so 24 mines pays 24.75x at 99% RTP. The maximum is at 12 or 13 mines, where C(25, m) peaks at 5,200,300, and the March 2024 Stake win that produced the figure is reported as a 13-mine, 12-gem full clear. Where a source’s number and the formula disagree, the formula is checkable in one line and the source usually is not.

Key takeaways

• Mines contains no decisions. The bomb permutation is fixed before your first click, so which tiles you pick and in what order changes nothing. Two pre-round numbers — mine count and intended picks — determine the entire game.
• One formula forces every multiplier: P = C(25−m, k) ÷ C(25, k), payout = RTP ÷ P. The RTP cancels out of the expected value, so every setting returns 0.990 of stake at 99% RTP. This is the same structural result as crash’s flat 0.970, not a new one.
• Mines and picks are interchangeable — P(m, k) = P(k, m) across all 300 valid pairs. Three mines with five picks is the same bet as five mines with three picks, at 1.9974x.
• What the mine count buys is variance, not value: an 11,172× range in standard deviation across settings that share one expected value.
• Only two things can change your edge, and neither is on the grid: the operator’s RTP (96% costs 4× what 99% costs) and a max-win cap, which can cut effective RTP to 0.19% on the settings the game advertises hardest.
• Martingale is worse here than on crash. The 2x-equivalent setting pays 1.9974x, so a successful recovery at the tenth step leaves you 35 cents down after staking $1,023.
• Provably fair proves the board was not moved, not that the game is beatable. The RTP multiplier is applied after the shuffle, at the payout table.
• The strategy content written about this game is describing variance, not edge. Patterns, hot tiles and cash-out timing rearrange the same negative expectation. That is not a technique to be improved on; it is the reason no technique exists.

Frequently asked questions

What is the Mines gambling game?

Mines is a crypto-casino game played on a 5x5 grid of 25 tiles containing a number of hidden mines that the player chooses, usually between 1 and 24. Each safe tile revealed raises a multiplier, and the player can cash out at any point. Hitting a mine ends the round and the stake is lost.

How are Mines multipliers calculated?

The chance of surviving k safe picks with m mines on a 25-tile board is C(25 minus m, k) divided by C(25, k). The fair multiplier is 1 divided by that probability, and the multiplier the game offers is the RTP divided by that probability. On a 99% RTP game, 3 mines with 5 safe picks survives 49.57% of the time and pays 1.9974x.

Does changing the number of mines change the house edge in Mines?

No. Every mine count and every cash-out point returns the same fraction of stake, because the multiplier is derived directly from the probability that the mine count creates. At a stated 99% RTP the expected return is 0.990 of stake whether you set 1 mine or 24. What changes is volatility, which ranges over four orders of magnitude across settings at identical expected value.

What is the maximum multiplier in Mines?

On a 5x5 board at 99% RTP the maximum is 5,148,297x, achieved by clearing an entire board at 12 or 13 mines, where C(25, m) peaks at 5,200,300. The odds are 1 in 5,200,300. A real payout at exactly that multiplier was reported on Stake in March 2024. Contrary to a widely repeated claim, 24 mines does not produce the maximum: it pays 24.75x.

What RTP does the Mines game have?

It depends on who built it. Stake states 99% RTP and a 1% house edge on its own Mines game page. Casino.band's tracker, updated 1 May 2026, puts BC.Game at 99%, Roobet, Spribe and Turbo Games at 97%, and SmartSoft at 96%. Trackers disagree on two of those: Win.gg gives BC.Game 98% and a general 96 to 98% range, and MinesCalc lists Roobet at 96%. The authoritative figure is the one shown in the specific game's own info panel.

Does a max win cap affect the Mines house edge?

Yes, and it is the only thing on the board that can. Where an operator applies a maximum-win cap, any configuration whose payout would exceed it is paid the cap instead, so the probability no longer cancels. Under a 10,000x cap, the 12-mine full-clear setting has an effective RTP of about 0.19% rather than 99%. Stake has paid above 5,000,000x on Mines, so no such cap binds there.

Is there a winning strategy for the Mines game?

No. The mine layout is fixed by a cryptographic shuffle before the first click, so tile choice and click order carry no information. Every configuration has the same expected return, and doubling systems only concentrate losses. At the 3-mine, 5-pick setting a doubling sequence has staked $1,023 by the tenth loss, and because that setting pays 1.9974x rather than 2.00x, a recovery win there still leaves the sequence 35 cents down.

Sources

• Stake.com — Provably Fair Implementation (HMAC-SHA256, server seed, client seed, nonce)
• Casino.band — Best Mines Game Casinos 2026: RTP Data, Probability Tables & Provably Fair (updated 1 May 2026; accessed 23 July 2026)
• Win.gg — Best Mines Game Casinos 2026: Mines Gambling with High RTP
• 100RTP.games — Mines Multiplier & Payout Chart: RTP Audit Calculator
• Stake.com — Mines (Stake Originals) game page — stated 99% RTP, 1% house edge
• MinesCalc — Mines Multiplier Chart & Formula
• MinesCalc — Roobet Mines Calculator — Odds, Payout & 96% RTP Chart (the conflicting Roobet figure)
• MinesCasinoGames — Mines (Spribe) review: 97% RTP, 5x5 grid, max 10,000x
• FreeTips — Stake player hits 5,148,297x Mines multiplier (6 March 2024)
• CrashGamblingSites — Stake Mines player uncovers 1,070,759.20x multiplier (5 August 2025)
• Roshtein — How Stake’s Provably Fair System Works: Fisher-Yates shuffle of the 25 tile indices