A game’s return is a fixed total divided between the base game and the feature, and the split is never published. It can be estimated to a useful precision from two figures that are, and the estimate explains almost everything about how the game behaves.
The two figures
The feature purchase price, where one is offered, and the average interval between triggers, which can be measured. The purchase price is close to the expected value of a feature round, since it would be uncommercial otherwise.
Divide the purchase price by the trigger interval and the result is roughly what each base-game round contributes towards reaching the feature. Compare that against the stake and the split becomes visible.
Worked through
Suppose a feature costs a hundred stakes to buy and arrives on average every two hundred and fifty rounds. Each round is then carrying about four tenths of a stake towards the feature, which means a substantial share of the total return is being routed through it rather than paid out along the way.
Measuring the interval
By counting. Running the free play version with a round counter and noting the number at each trigger gives five or six intervals in an hour, which is enough to place the figure within a usable range.
Where no purchase option exists, the price is often documented for other markets and remains a valid estimate of the feature’s expected value even where the option cannot be exercised.
- Find the purchase price or its documented equivalent
- Measure the trigger interval by counting
- Divide one by the other for the per-round contribution
- Compare the result against the stake
- Cross-check against the base-game maximum win
The cross-check
Compare the base game’s own maximum against the game’s overall stated maximum. A large gap confirms the feature carries the ceiling and, by extension, most of the return concentrated at the top end. The two estimates should agree in direction, and where they do not, one measurement is wrong.
What follows from a high split
A base game that is deliberately thin, long intervals, and a session shorter than the interval that samples only the least rewarding part of the design. None of this is a fault; all of it is the arithmetic consequence of the allocation, and it is predictable in advance rather than discoverable only by playing.
Why the estimate is worth the hour
Because it converts a game from something experienced into something described. Two titles with the same return percentage and different splits are different products, the specification does not distinguish them, and an hour of counting does.
