Slot RTP is published at 96% — the variance eats 60% of that

Most Australian pokies list an RTP around 96%, and plenty of players treat that number as a promise: feed enough spins through the machine and you'll get most of your money back. That's not what RTP means. The published figure is a long-run average, and the variance around it is wide enough that a real session can hand back almost nothing while the game still technically pays 96%. The gap between the advertised number and what you actually experience is where most of the money quietly disappears.

Start with the arithmetic. A 96% RTP means the house keeps 4 cents of every dollar staked, on average, over a sample large enough to make "average" mean something. The question is how large. For a typical high-variance slot, you need hundreds of thousands of spins before your personal return converges on the theoretical figure. At 500 spins an hour, that's a few hundred hours of continuous play. Almost nobody plays that long. So the number you're quoted describes a player who doesn't exist — an infinite one.

What the 96% actually averages out

The published RTP is calculated by running the game's full paytable and reel weights through a simulation, often over tens of millions of spins. It's an accurate figure for the game as a mathematical object. It says nothing about the next 200 spins you'll play.

Here's the part that trips people up. A 96% RTP game doesn't pay back 96% in a smooth line. It pays back in lumps: long dry spells punctuated by rare big hits. The 96% is the sum of all those lumps divided by all the money staked, and a huge share of the return is concentrated in outcomes so rare you'll probably never see one.

Take a game where 60% of the total RTP comes from a single jackpot or bonus combination that lands once in every 250,000 spins. For most sessions, that 60% simply isn't in the room. You're effectively playing a game that returns 36% until the rare event fires — and if it doesn't fire while you're sitting there, it never fires for you. The 96% was never yours to collect.

The maths behind "the variance eats 60%"

This is the claim in the headline, and it's worth being precise about it. In high-variance slots, the top slice of the paytable — the rarest outcomes — routinely accounts for 50–70% of the theoretical RTP. That means the bulk of the advertised return is parked in events that occur far outside a normal session. The variance hasn't literally removed 60% of the RTP; it has removed 60% of the RTP from the part of the distribution you'll actually experience. The number is real. Your access to it isn't.

A concrete anchor: on a game with 96.2% RTP and a maximum win of 50,000x, the single top prize can contribute several percentage points of that 96.2% on its own. Strip out the top 1% of outcomes and the effective RTP for a 300-spin session can fall into the low 40s. That's not a rigged machine. That's a fair machine whose fairness is defined over a timeframe you'll never play.

Why volatility matters more than the headline RTP

Two games can both list 96% and feel completely different. One pays small amounts often; the other pays almost nothing for ages and then a lot. The RTP is identical. The experience isn't.

Volatility — sometimes called variance — describes how spread out the outcomes are. Low-volatility games cluster results near the average, so your session return will sit close to 96% most of the time. High-volatility games scatter results widely, so your session return can be 20% or 200% with equal ease. The RTP tells you the centre of the distribution. Volatility tells you how far from the centre you're likely to land.

For Australian players, this matters because the popular games — the ones with the big max-win multipliers and the flashy bonus rounds — are almost always high volatility. The 50,000x potential is the marketing. The dry spells are the reality. If you're picking a game by RTP alone and ignoring volatility, you're reading half the label.

A rough way to compare

If a game advertises 96% RTP and a 10,000x max win, expect long sessions with little return and rare spikes. If it advertises 96% and a 500x max win, expect steadier returns and fewer heartbreaks. Neither is "better." They suit different bankrolls and different tolerance for sitting through 400 spins with nothing to show. The 96% doesn't change. Your experience of it does.

The session you actually play

Here's the uncomfortable truth about RTP: it's a property of the game, not of your session. Your session has its own return, and that return is a random variable. It might be 130%. It might be 15%. The published 96% is the average of millions of sessions, most of which you'll never play.

This is why "the machine is due" is a fallacy and why "the machine is hot" is equally wrong. Each spin is independent. The game doesn't remember that you've lost 300 in a row and owe you a bonus. The 96% is baked into the paytable, not into a running tally.

What you can control is how long you expose yourself to the variance. A $200 bankroll on a $1 spin gives you 200 spins to hit a good run. A $200 bankroll on a $5 spin gives you 40. The second player is far more likely to bust before the variance has a chance to swing their way — not because the game is worse, but because they ran out of runway. The RTP didn't change. The sample size did.

This is also where the responsible-gambling angle stops being a footnote and becomes the actual point. If the advertised RTP is mostly inaccessible in a normal session, then treating a session as an investment with an expected return is a mistake. It's entertainment spending, and the sensible question isn't "how much will I get back?" but "how much am I comfortable spending to play?" Set that number before you start, not after the variance has had its way with you. Services like Gambling Help Online (1800 858 858) exist for the sessions where the answer to that question got away from you.

What the 96% is really telling you

The published RTP is honest. It's just answering a question most players aren't asking. It tells you the long-run average return of a game played forever. It doesn't tell you what your next hour looks like, and for high-volatility games it barely constrains it.

So what should you do with the number? Use it to compare games with similar volatility — a 97.3% game is genuinely better than a 94% game over the long run, all else equal. Use it to spot the games that are quietly expensive. But don't use it to predict a session, because it can't. The variance is the part of the machine you actually play against, and it doesn't care what the label says.

The open question is whether the industry should publish something more useful — a session-return range, say, or the share of RTP locked behind the top 1% of outcomes. Some regulators have started pushing for clearer volatility labelling; most haven't. Until they do, the 96% will keep doing what it's always done: describing a player who plays forever, to people who play for an hour.