Checking the RTP of a Low-Volatility Slot by Playing It: After 100,000 Spins the Standard Error Is Still 1.8 Points
Players measure a game by their own sessions and call the result its real RTP. The exact arithmetic of how many spins a measurement needs on our low-volatility model: 313,788 for a standard error of one point.

Because the standard error shrinks only with the square root of spins, you need 313,788 spins to nail a 1% error - that’s 22 days of nonstop play at 600 spins per hour. A 500-spin evening still leaves a 25-point error, so any single-session return tells you virtually nothing about the slot’s true RTP.
- After 1,000 spins the standard error is 18 percentage points; after 100,000 spins it’s 1.8 points.
- To achieve a 1% standard error you must play 313,788 spins, which at 600 spins per hour equals 22 days nonstop.
- The two RTP settings (96.5% vs 94.0%) are 2.5 points apart, yet a sample of 50,207 spins still has a standard error that large.
- A 500-spin session carries a 25-point standard error, so its observed return is essentially noise.
One spin against its own average
One spin of the low-volatility model at 96.5% RTP loses 0.035 stakes on average, while the standard deviation of a single spin is 5.6 stakes. The result of one spin lies 160 times further from zero than its average does.
The standard error of an observed return is the standard deviation divided by the square root of the number of spins. After 1,000 spins the standard error is 18 percentage points; after 100,000 spins it shrinks to 1.8 points; after 1,000,000 spins it is 0.56 points. Because the divisor grows as the square root, precision grows with the square root of the sample, not with the sample itself.
To bring the standard error down to 1% takes 313,788 spins, which at 600 spins an hour is 22 days of play around the clock. Halving the standard error requires four times as many spins, so a 0.5% error needs 1,255,151 spins and 87 days of continuous play. Achieving a tenth of a point of error demands 31,378,754 spins, or 6.0 years at the same pace.
A session of 500 spins has a standard error of 25 percentage points on this model, meaning the return seen in one evening tells next to nothing about the underlying setting. Two settings of the same game at 96.5% and 94.0% RTP are 2.5 points apart; a sample needs 50,207 spins for its standard error to be as large as that gap, and 200,825 spins for half that gap, which equals 14 days of play around the clock at 600 spins an hour.
The standard error of the observed return after 100,000 spins is 1.8 percentage points, which takes 7 days of play around the clock at 600 spins an hour.
What a sample of spins can pin down
The dispersion of a measured return shrinks only with the square root of the spin count, so each extra zero in the sample adds a modest bite of precision. For the low-volatility model the exact arithmetic from the published payout bands gives a standard error of 18 percentage points after 1,000 spins, 1.8 points after 100,000 spins, and 0.56 points after 1,000,000 spins. A single evening of 500 spins still carries a 25-point error, meaning the observed return is essentially noise.
Because the error falls as σ/√N, cutting it in half demands four times as many spins.
Translating the multiplier into time is straightforward. At a steady 600 spins per hour, the 12,552-spin run consumes about 21 hours of continuous play. Doubling the spin rate would halve the calendar time, but the spin count requirement remains unchanged.
If your goal is to confirm an RTP claim within a one-point margin, the model tells you that 313,788 spins are required.
For a modest 5-point band, schedule roughly 21 hours of play at 600 spins per hour.
The spins a given precision needs
To hit a standard error of one percentage point you must survive 313,788 spins; half a point demands 1,255,151 spins; a tenth of a point explodes to 31,378,754 spins. Those are the exact counts produced by the model’s arithmetic.
At the assumed pace of 600 spins an hour - one spin every six seconds - 313,788 spins stretch to 22 days of nonstop play.
Because the standard error shrinks with the square root of the sample, halving the error requires four times as many spins.
A slot with a higher top payout or a rarer bonus round raises the standard deviation, so the spin counts needed for the same precision would be larger than the numbers listed here.
In practice, the model shows that achieving a standard error of 0.1 percentage points requires 31,378,754 spins, equivalent to 6.0 years of play at 600 spins per hour.
Telling one RTP setting from another
The two RTP settings differ by 2.5 points. Because the standard error shrinks with the square root of the sample, a sample of 50,207 spins produces a standard error equal to that whole gap.
The calculation rests on the published payout bands of the low-volatility model calibrated to 96.5% RTP. It tells how far samples scatter, but it does not claim the scatter follows a bell curve; with the 393x top payout and a bonus that appears on average once per 120 spins, short samples are far from normal.
If a real slot raises the top payout or makes the bonus rarer, the single-spin standard deviation grows. A larger deviation pushes the required spin count higher than the 200,825 spins needed for the model’s half-gap precision.
Thus, to distinguish a 2.5-point RTP difference on a game with the model’s parameters, you must accumulate at least 200,825 spins, which translates to 14 days of continuous play at 600 spins per hour.
The RTP of a low-volatility slot cannot be determined from a single session or even a few hundred spins. Getting a 1% error requires 313,788 spins - about 22 days of nonstop play at 600 spins per hour - so any short-term observation is essentially meaningless.
The data behind this piece
Every figure in the text comes from the tables below. They are counts over our own catalogue or our copy of a public register and exact calculations, with the method stated for each. Nothing here was measured on a live casino.
How many spins it takes to check the RTP of the low-volatility model
How we ran it: Exact arithmetic from the payout bands of the low-volatility model calibrated to 96.5% RTP: any-win frequency 32.6%, bonus round on average once per 120 spins, 25% of the return comes from the bonus round, top payout 393x the stake: the standard deviation of one spin, and from it the standard error of the observed return for samples of different sizes. Hours of play assume 600 spins an hour. No simulation is involved.
| Spins in the sample | Standard error, percentage points | Time at 600 spins an hour |
|---|---|---|
| 1,000 | 18 | 1.7 hours |
| 10,000 | 5.6 | 17 hours |
| 100,000 | 1.8 | 7 days of play around the clock |
| 1,000,000 | 0.56 | 69 days of play around the clock |
| 10,000,000 | 0.18 | 1.9 years of play around the clock |
| Standard error wanted, percentage points | Spins needed | Time at 600 spins an hour |
|---|---|---|
| 5.0 | 12,552 | 21 hours |
| 2.5 | 50,207 | 3 days of play around the clock |
| 1.0 | 313,788 | 22 days of play around the clock |
| 0.5 | 1,255,151 | 87 days of play around the clock |
| 0.1 | 31,378,754 | 6.0 years of play around the clock |
Questions players ask
How many spins do I need to see the true RTP of a slot?
Because the standard error falls with the square root of the spin count, you need 313,788 spins to bring the error down to 1 percentage point. At a typical pace of 600 spins per hour that means about 22 days of nonstop play. Anything less leaves the measured RTP swamped by noise.
Is a 500-spin session enough to know if a slot is fair?
No. On the low-volatility model a 500-spin session has a standard error of 25 percentage points. That error dwarfs the 2.5-point gap between a 96.5% and a 94.0% RTP setting, so the observed return from a single evening tells you virtually nothing about the game’s true payout rate.
What does a 1% standard error mean for my bankroll?
A 1% standard error means the measured return could be off by plus or minus one percentage point from the true RTP. Reaching that precision requires 313,788 spins, far beyond a typical casino session.
How long would it take to differentiate a 96.5% RTP slot from a 94.0% RTP slot?
To make the standard error as small as the 2.5-point gap between the two settings you need 50,207 spins. Anything shorter leaves the gap buried in statistical noise.