A jackpot screenshot is a house edge in disguise

Big wins get treated as marketing. A number flashes across a lobby ticker, the figure gets screenshotted into a Telegram channel, and everyone argues about whether it was real. Almost nobody does the one thing the screenshot invites, which is arithmetic.

For a certain class of casino game, a single winning multiplier is enough to recover the operator's house edge to three decimal places. No session logs, no million-spin sample, no trust in a certificate. Just division.

Which games leak their edge

Slots do not. A slot's return comes from a weighted reel configuration nobody publishes, so the only honest way to measure it is to play it a few million times or to take the auditor's word.

Games built on combinations behave differently. Their payouts are not chosen by a designer, they are computed. If a game pays out for surviving a run of independent choices, the fair multiplier for any given depth is the reciprocal of the probability of getting there. The operator then multiplies that fair figure by one minus its edge and prints the result on the paytable. Which means the edge is sitting inside every multiplier in the game, waiting to be divided back out.

Mines is the cleanest example on the market. Twenty-five tiles, you pick how many mines are buried, then you reveal tiles hunting gems. Clear the entire board and the fair multiplier is exactly C(25, m), the number of distinct ways m mines can be placed on 25 tiles. That is it. There is no other number it could be.

Working one example through

The largest win logged on Duel's Mines is a multiplier of 2,040,932.03x, which paid out $20,421.83. Two divisions take you from that to the house edge.

Start with the stake, because it sanity-checks everything else. Divide the payout by the multiplier: $20,421.83 / 2,040,932.03 gives a shade over one cent. A minimum bet, in other words, which is exactly the profile of somebody farming full clears rather than playing for a return. Nothing about the figure is inconsistent so far.

Now the multiplier itself. C(25, 9) is 2,042,975. Divide the paid multiplier by that coefficient and you get 0.999. Run it the other way for a cleaner look: 2,040,932.03 divided by 0.999 lands on 2,042,975 exactly, with nothing left over.

Three facts fall out of a single line of arithmetic.

The house edge is 0.1%. Not approximately, not on average over a sample. That is the multiplier the operator chose to pay against a known fair value.

The round was a complete clear on the nine-mine setting. All sixteen gems, none left on the board, because a partial cash-out would have produced a different coefficient entirely.

The odds were 1 in 2,042,975. That is the probability of picking sixteen tiles in a row and missing every mine.

There is a pleasing footnote here. C(25, 9) and C(25, 16) are the same number, so the identical multiplier is reachable at sixteen mines by clearing nine gems instead. The paytable cannot distinguish those two rounds.

What the method is actually good for

Once you have the edge, you no longer need to measure anything. Every multiplier in the game follows from grid arithmetic, at every mine count and every depth, and you can generate the whole schedule on a laptop. The independent analysis at duel-5.com does exactly that and publishes the full table, including the maximum payout the game can produce, which turns out to be 5,195,099.70x at twelve or thirteen mines rather than anything larger.

That completeness has a practical use: it lets you catch bad numbers.

Search for Mines multipliers and you will meet 5,148,297x repeatedly, usually credited to the twenty-four mine setting with a single gem found, and described as a 1 in 25 shot that pays seven figures. Both halves of that claim are wrong, and the arithmetic shows why in seconds. At twenty-four mines there is exactly one gem among twenty-five tiles, so finding it is indeed a 1 in 25 event, and it pays about 24.98x. It is the joint-cheapest outcome in the entire game. Meanwhile C(25, 12) is 5,200,300, and shaving 1% off that lands on 5,148,297 precisely. So the figure is a twelve-mine full clear priced by an operator charging a 1% edge. Right family of numbers, wrong setting, wrong operator, wrong edge.

That figure has been copied between affiliate pages for years because nobody divided it by anything.

The limits of the trick

Recovering an edge from a payout is not the same as verifying a game is honest, and it is worth being precise about the difference.

Provable fairness is a separate property. It concerns whether the mine layout was committed to before your first click, which you check by hashing a revealed seed against a published commitment. That test says nothing about price. A layout you have personally verified, generated from a seed you contributed to, still carries whatever edge the paytable carries. The edge lives in the payout schedule, not in the shuffle.

The reverse is also true. A game can have an admirably small edge and still be implemented by an operator you would not lend twenty dollars to. Arithmetic tells you the price of the bet. It does not tell you whether you will be paid.

What the method does give you is a defence against being told a number and having no way to test it. When a game's payouts are combinatorial, the edge is not a claim. It is a quotient, and you can compute it yourself before you stake anything.