The published percentage is a total, and it says nothing about where within the game that total is generated. The split between base rounds and feature rounds is not disclosed and is the single most useful thing to know about how a session will go.
Why the split varies so much
It is a design choice. A studio can build a game where features carry most of the return and ordinary rounds are thin, or one where value is spread across all rounds, and both will quote the same headline figure.
The two produce completely different sessions. In the first, everything depends on reaching the feature; in the second, the base game is the game and the feature is a bonus.
Estimating it
Measure the base game directly. Count rounds, exclude every feature entry, total what those rounds staked and returned, and divide. The gap between that figure and the published one is roughly what the features carry.
Where to run the count
Free play, over several hundred rounds. Using the Playtech demo for the measurement costs nothing and produces the same numbers as a paid session would, since the outcome generation is identical.
The sample will be noisy and the answer approximate. Approximate is sufficient — the question is whether ordinary rounds return most of the stake or a fraction of it, not the second decimal place.
- Count several hundred rounds excluding features
- Total the stake and the return separately
- Divide to get the base game percentage
- Subtract from the published figure for the feature share
- Treat the result as a band, not a point
What a moderate split indicates
A game playable in short sessions. If the base return is close to the published figure, a session that never reaches a feature has still performed roughly as expected, which removes the dependence on session length entirely.
What a wide split indicates
A game that requires commitment. Ordinary rounds are funding something that arrives rarely, and a short session is paying for a payoff it will not see. That is a legitimate design and a poor match for a small budget.
Why the measurement is worth twenty minutes
Because it answers a question nothing published addresses, it holds permanently for that build, and it determines whether the game fits the way a particular person actually plays — which no volatility label or return percentage can tell you on its own.