Gross winnings in a Mines round equal the base stake ($B$) multiplied by the active cash-out multiplier ($K_s$) reached after $s$ safe reveals. In a mathematically fair model with zero house edge, the theoretical multiplier is the inverse of the cumulative survival probability of making $s$ consecutive safe picks:
$$\text{Fair Multiplier} = \frac{C(25, s)}{C(25 - M, s)}$$
Here $C(n, k)$ is the binomial coefficient $\binom{n}{k}$, $M$ is the number of hidden bombs, and $s$ is the count of revealed safe tiles. Commercial operators then apply a return-to-player (RTP) factor to that fair figure. In a title calibrated to 97% RTP, the advertised cash-out multiplier at each step is scaled to roughly $0.97 \times \text{Fair Multiplier}$. Individual rounds stay wildly variable; the long-run expected return stays pinned to the house edge.
A worked example makes the scaling explicit. With one mine ($M = 1$) and a single safe pick ($s = 1$), the fair multiplier is $C(25,1) / C(24,1) = 25/24 \approx 1.0417$x. Apply the 97% factor and you get roughly $0.97 \times 1.0417 \approx 1.01$x, which is the familiar 1.01x on screen in most commercial builds. On larger grids the same formula applies with $N$ replacing 25. That is why a 9×9 board holding 80 mines can advertise four-figure multipliers on one successful pick.