What is RTP and volatility in gambling? The honest math

RTP (return to player) is the percentage of all money wagered on a game that it pays back to players, measured over millions of plays. A 96% RTP means the game keeps 4 cents of every dollar staked, on average; that 4% is the house edge. Volatility describes how widely individual results swing around that average: low-volatility games pay small wins often, high-volatility games pay rarely but in larger chunks. Neither number predicts your session tonight.

The detail game screens leave off: RTP applies to the total you wager, not to the money you deposit. Winnings get re-staked, so the same $100 can pass through a game ten or twenty times in an evening, and the edge takes its cut on every pass. Below is what that costs in plain numbers, why volatility makes the edge hard to feel, and why trying to win it back fails on average by design.

• RTP is a long-run average. Regulators check it over huge volumes of play; one evening is statistical noise.
• Expected loss = house edge × total amount wagered, and total wagered is usually several times your deposit.
• Volatility changes how losses arrive, not whether they arrive.
• No pile of negative-expectation bets adds up to a positive one. Chasing adds volume, and volume feeds the edge.

What RTP actually measures

Independent testing agency eCOGRA defines RTP as the percentage of wagered money a game pays back to players over time. The figure in a game's info screen is the theoretical RTP: the return built into the math model. Regulators then track the actual RTP, total player wins divided by total stakes over a period of live play, to confirm the game performs as designed.

Even at scale, the two diverge. eCOGRA's own worked example is a game designed at 91.68% that took £1,200,000 in bets over a month and paid £1,085,000 in wins: an actual return of 90.42%. A full month of every player's action still missed the design figure by more than a point. The UK Gambling Commission, which requires operators to monitor live RTP performance, makes the same point from the other side: measuring a game over a small amount of play is pointless because "the tolerance will be too large for meaningful results". Your two hours on a phone will not land anywhere near 96.00%.

In Great Britain, licensed sites must make each game's RTP or house edge easily available before you commit to a bet, under technical standard RTS 3. Typical figures, from UNLV's casino-math guide and eCOGRA:

The edge itself is just a payout set below true odds. Single-zero roulette has 37 pockets, so a fair payout on one number would be 36 to 1; the table pays 35 to 1. That missing unit is 1/37 of every stake, or 2.7%, and no betting pattern can rearrange it away.

What RTP is not

• Not a session forecast. It constrains the average of millions of spins across all players, and says nothing about your next hour.
• Not the share of your deposit you keep. The edge applies to every wager, so re-staking multiplies what it touches.
• Not a dial that reacts to your losses. British standards state that "adaptive behaviour (that is, a compensated game) is not permitted": a licensed game may not tighten or loosen based on what just happened.

What a 96% slot really costs: a worked example

Say you deposit $100 and play a 96% RTP slot at $1 a spin. Your expected cost is 4% of whatever you put through:

Now add speed. Great Britain set a minimum spin duration of 2.5 seconds and banned autoplay in October 2021 precisely because pace drives losses. Even at that regulated pace, continuous play allows up to 1,440 spins an hour: $1,440 wagered at $1 stakes, with about $58 of it expected to go to the house every hour. From a $100 deposit. With perfectly average luck.

The re-staking spiral is easiest to see in cycles. Run your full balance through the game once and 96% of it comes back on average. Do that ten times and $100 has become about $66 in expectation. Twenty times: $44. Fifty times: $13. There is no unlucky streak in that arithmetic; it is the average player's path, spread over many nights. Bad luck only gets there sooner.

Volatility: same average, very different ride

Volatility (statisticians use variance, or standard deviation) measures spread around the long-run average. Two slots can both be set to 96% and feel like different products: one drips 30-cent and $2 wins constantly, the other can swallow 200 spins in silence and then return 150 times your stake.

Short-run spread is far larger than intuition allows. In the UNLV casino-math guide, Robert Hannum works through 1,000 straight $1 bets on a single number in double-zero roulette: the expected result is $53 down, but the realistic (95%) range runs from roughly $310 ahead to $420 behind. A thousand bets into a 5.3%-edge game and you can still be hundreds of dollars up. No sample one person can play is big enough to reveal the edge, and the edge applies the whole time regardless.

High-volatility games carry a subtler trap: skew. When a big slice of a game's RTP sits in rare events such as jackpots and bonus rounds, the advertised average is held up by a small number of very large wins, and the typical session pays back noticeably less than the headline figure. Most nights run below 96% so that a rare night can run far above it. The early big win that convinces someone a game "likes" them is exactly what volatility is expected to produce now and then.

Why “winning it back” fails

Chasing rests on two beliefs, and both fail on the math. The first is "the machine is due." It cannot be. A certified random number generator has no memory: every spin is drawn fresh, and a losing streak does not raise the probability of the next win by any amount. In Great Britain this independence is a legal requirement under RTS 7, the same standard that bans compensated games. A slot that "owed" players a win after a cold run would be non-compliant by design.

The second belief is "with enough play I can get back to even." Expected results add. Every additional $100 through a 4% edge costs $4 more on average, so extra volume moves you away from even, not toward it. Winning it back tonight is possible (that is volatility) and still a losing plan (that is RTP): the operator's return comes from total turnover, not from any single session.

If the pull to keep going is strongest late in the evening, our guide on getting through a gambling urge at night covers what helps in the moment.

How bonus terms shrink the return

Bonuses change your effective return through wagering requirements: conditions that force you to re-stake bonus funds before anything can be withdrawn. The Gambling Commission's own example is a £10 bonus with a 50x requirement, meaning £500 must be played through first. At a 4% house edge, £500 of forced turnover is expected to cost £20, twice the bonus itself. On average, that "free" offer left players worse off than no offer at all.

Great Britain capped wagering requirements at ten times the bonus from 19 January 2026, after requirements above 50x had become standard industry practice. Even at the cap, the math stays unfriendly: £10 at 10x means £100 through a 96% game, with £4, or 40% of the bonus, expected to be consumed on the way. Check the terms for bet caps and which games count toward the requirement; UK consumer guidance says any restrictions on playing with bonus funds must be made clear to you.

A bonus is deferred turnover, not free money. Whatever the multiple, its practical effect is to guarantee the volume the edge needs.

What you can do in the next ten minutes

1. Find the real numbers for the game you play most. On GB-licensed sites the RTP or house edge must be shown before you stake; elsewhere, check the game's help screen. Then work out your hourly cost: stake × spins per hour × house edge. Your own number, in your own currency, is more persuasive than any article.
2. Set a deposit limit while you are calm. Limits set mid-urge rarely hold. Our guide to setting gambling limits covers deposit, loss, and session caps that survive the moment you want to remove them.
3. Write down what gambling cost you last month. Bank-statement numbers, not remembered wins. A simple gambling budget tracker takes ten minutes to start and replaces impressions with data.

These products are built so that the average stays invisible and the exceptions feel like patterns. If play has stopped feeling like a choice, check the signs of gambling addiction or read what gambling addiction is; the helplines in our country directory are free and confidential.