One invented game returns 1.20 units on 80% of rounds. Another returns 48 units on 2%. Both can have a 96% return to player. The average agrees; the chance of a round returning nothing is very different. Slot volatility concerns the spread of payouts that an RTP percentage leaves undescribed.
These are deliberately simplified models, not two games tested in a casino. They make a useful point about reading a slot: an average return, the frequency of paying rounds and the largest available prize are separate pieces of information. None provides a schedule for the next payout.
Two payout models, one average
Each round in our example costs 1 unit. Model A returns 1.20 units with probability 80%, and zero otherwise. Model B returns 48 units with probability 2%, and zero otherwise. Payout means the total amount returned after the stake has been paid, not profit on top of a refunded stake. There are no bonuses, jackpots or changing game states, and successive draws are independent.
| Fictional model | Returned payout per 1-unit round | Expected return per round |
|---|---|---|
| A | 1.20 units with 80% probability; zero with 20% | 0.80 × 1.20 = 0.96 units |
| B | 48 units with 2% probability; zero with 98% | 0.02 × 48 = 0.96 units |
Dividing either expected return by the 1-unit stake gives 96% RTP. Subtracting the stake gives an expected net result of minus 0.04 units per round in either model. Frequent payouts have not removed the mathematical loss in A, and B's much larger prize has not improved its average.
The difference is the dispersion around that 0.96-unit mean. A's only returned amounts are 0 and 1.20; B's are 0 and 48. Calculating standard deviation from the complete two-outcome distributions, including the zero returns, gives 0.48 units for A and 6.72 for B. That is 14 times as much payout spread measured this way, despite identical expected returns. These calculated values belong to our illustration, not to a supplier's star rating.
The distinction is a game-design principle rather than a new claim about a particular casino. Robert Muir's 2013 Elements of Slot Design, chapter 2, illustrates how equal RTPs can accompany very different payout patterns. It is a technical introduction from the maker of Slot Designer software, not an independent audit of the games in today's lobbies.
Our RTP explainer examines what the average percentage measures. The additional question here is how widely the individual returned amounts can differ while keeping that average unchanged.
A hit is not necessarily a profit
In the two models above, every positive payout exceeds the 1-unit stake. Real product information needs closer reading. If a supplier defines a hit as any round awarding money, that definition can include a return smaller than the amount staked.
A hypothetical 0.40-unit return after a 1-unit stake leaves a net loss of 0.60 units. Counting that event as a paying round would not make it profitable. Our article on slot losses presented like wins explores the separate design issue of how such results can be displayed or celebrated.
A frequency number therefore needs its counted event: a paying round, a net-positive round or a particular feature trigger. Without that definition, it cannot safely be translated into the chance of increasing an account balance. Nor does an average frequency tell a customer which round will produce the event.
In our independent models, the 80% and 2% payout probabilities do not mean exactly 80 or two payouts in every batch of 100 rounds. Actual sequences can depart from those proportions. The size of that departure and the sizes of the payouts both matter to a session's result; the RTP average alone supplies neither sequence.
A product profile is not a probability table
NetEnt's public Twister profile lists 96.04% RTP, 20.00% hit frequency and a maximum payout of 4,300 times the bet. Its Wild Bazaar profile lists 96.15%, 28.00% and 800 times the bet, respectively. The pages present RTP, volatility, hit frequency and maximum payout as separate fields. Neither the frequency nor the ceiling supplies a mathematical volatility rating.
The named profiles are examples of published product information, not recommendations or returns independently verified by Betting Tribune. Their relatively close RTP figures coexist with different frequency and maximum fields. Those fields still do not reveal the complete payout distributions, or the probability of achieving either stated maximum.
Both pages also show a demo in a Playin Coins context. A public product or demo page does not establish which cash-game configuration a casino has loaded. The current game's own rules and help information remain necessary. Our game-version guide explains why a familiar title and artwork do not identify a complete mathematical configuration.
A volatility label summarises variation; it does not replace the outcomes and their probabilities. A ceiling states the largest advertised payout under the relevant rules, not how regularly it occurs. Turning either item into an estimate of a customer's next session would require information those headlines do not contain.
No countdown to the next return
Model B has a 98% chance of returning zero on each independent draw. Ten previous zero returns leave that next-draw probability at 98%, because the model's probabilities and state have not changed. Its large payout does not become due when the preceding run looks unusually long.
That assumption needs to stay visible when discussing real games. A bonus or special feature can introduce a different rule set; a fixed-distribution illustration is not a description of every feature state. Great Britain's RTS 7 prohibits adaptive behaviour that compensates for prior outcomes, while distinguishing bonus or special features based on random events. That is a requirement for the remote products within that British standard, not a claim about every physical machine or every jurisdiction.
The technical rule also does not establish that a particular site or game complies with it. A few visible outcomes cannot reconstruct the full probability model or verify an implementation. Volatility describes the spread built into a design; it is not a diagnosis of a malfunction from one losing session.
Read the average beside the payout pattern, then keep the exact game rules in view. An impressive maximum leaves its probability unanswered. A frequent return may still be smaller than the stake. Those are missing facts to establish, not reasons to expect that the next round will repair the last one.
