Most stake-sizing advice for online slots treats volatility as a label: low, medium, high. That is a category error. Volatility is a distribution, and a distribution has a spread you can measure. Once you measure it, stake sizing becomes arithmetic rather than vibes. This deep dive sets out a working formula, tests it against Pragmatic Play releases, and dismantles a few myths that cost players money.
Why the low, medium, high label fails as a staking tool
The label describes the shape of the payout distribution, not its scale. Two slots both tagged «high volatility» can differ by an order of magnitude in how violently results swing around the mean. A standard deviation of 15x stake per spin on one game and 25x on another means the second game demands a materially smaller stake for the same bankroll and the same tolerance for drawdown. Treating them as equivalent is how players bust on a session they had budgeted for a fortnight.
Pragmatic Play’s catalogue makes the point sharply. The 37 titles documented on this site between May and October 2026 carry a default return to player range of 96.42% to 98.00%, so the house edge is broadly comparable across the set. What differs is variance. A 98.00% RTP slot with a 25,000x ceiling behaves nothing like a 96.42% slot with frequent small pays, even though both sit under the same supplier badge.
The fix is to stop asking «is this high volatility?» and start asking «what is the standard deviation per spin, expressed in multiples of stake?» That single number converts directly into a stake figure. Everything else is commentary.
The formula: bankroll divided by (100 x volatility factor)
The working rule is simple: optimal stake equals bankroll divided by (100 x volatility factor). The volatility factor is the per-spin standard deviation expressed in stake units. A game with a 15x standard deviation has a factor of 15, so a 1,000 bankroll yields a stake of 1,000 / (100 x 15) = 0.67 per spin. A gentler game at 8x gives 1,000 / 800 = 1.25 per spin. The divisor of 100 is not arbitrary: it buys roughly 100 standard deviations of headroom, which is a defensible buffer for a session where you want a high probability of surviving normal bad runs.
Why standard deviation and not max win? Max win is a tail event that most sessions never approach. Gates of Olympus 2500 has high volatility and a 25,000x max win, but the 25,000x figure tells you almost nothing about the next 200 spins. The standard deviation tells you what a typical swing looks like, which is what actually determines whether your balance survives to the bonus round.
Why not RTP? Because RTP is a long-run average and variance is what happens on the way there. A 97% game still has a spread, and the spread is what bankrupts underfunded sessions. RTP sets the drift; volatility sets the noise. Stake sizing is a noise-management problem.
Worked examples across the Pragmatic Play range
Take Coven Rising as the high end. Its swings can reach roughly 15x stake per spin in standard deviation terms, so a 500 bankroll supports about 0.33 per spin. Forever Split Megaways, with up to 72,030 Megaways, concentrates payout into rare, deep combinations, which pushes its factor higher still; the formula naturally shrinks the stake to compensate. At the calmer end, a 96.42% RTP release with frequent scatter pays might sit near 8x, allowing a stake of 1.25 on the same 500 bankroll.
| Game | Volatility factor (per-spin SD, x stake) | Stake on a 500 bankroll | What the factor reflects |
|---|---|---|---|
| Deep Sea Frenzy | 8 | 0.63 | Frequent small pays, shallow swings |
| Cheetah Wild | 10 | 0.50 | Mid-range variance, cluster pays rhythm |
| Wonders of Aztec | 12 | 0.42 | Bonus-weighted, moderate tail |
| Coven Rising | 15 | 0.33 | Heavy swings, long dry spells |
| Forever Split Megaways | 20+ | 0.25 or less | Extreme Megaways tail, rare deep hits |
The table is illustrative, not gospel. Your own session logs will produce better factors than any published figure, because Pragmatic Play does not publish per-game standard deviations and the numbers shift with bet configuration. Log 500 spins at a fixed stake, compute the standard deviation of net result per spin, divide by stake, and you have your factor. It takes an evening and beats every volatility badge on the internet.
Myths that survive because nobody runs the numbers
Myth one: a Pragmatic Play APK or predictor app improves your staking. It cannot. Pragmatic Play acts as a B2B software supplier and publishes no native mobile applications in Google Play or the App Store, so any APK claiming to be the official Pragmatic Play Android app is unofficial at best, and you can check https://pragmatic-play-online.com/app/ for clarity. Predictor apps have no access to the random number generator and cannot forecast spins. The only legitimate mobile route is the browser.
Myth two: the iPhone needs a dedicated Pragmatic Play iPhone app. It does not. Pragmatic Play slots run inside a web browser or progressive web app, and iPhone and iPad users access slots via Safari with no download at all. A slot web app behaves like a native client without the install step, and player registration and balances remain with the licensed casino operator, not with the supplier. That is the correct architecture, and it is also the safest one.
Myth three: a demo mode tells you what the real game will pay. Demo modes load without an account using fun money, which makes them excellent for measuring volatility factor cheaply and useless for predicting outcomes. Use demo spins to estimate standard deviation, then carry the number to your real money casino session. The distribution is the same; only the settlement differs.
Putting the formula to work without fooling yourself
The formula is only as good as its inputs. Recalculate the volatility factor whenever you change bet size or game, because Pragmatic Play’s profiles are not linear across the range. A slot that behaves like a 12x game at 0.50 may behave differently at 5.00 if the paytable weighting shifts. Treat the factor as a local estimate, not a constant of nature.
Set a session stop-loss at roughly 30 standard deviations below your starting balance and a time limit alongside it. With a 15x factor, that is 450x stake of drawdown tolerance, which on a 500 bankroll at 0.33 per spin is about 148 down. If that number makes you flinch, the stake is too high, not the game too harsh. The formula is a mirror, not a promise.
Finally, remember that no staking method changes expected value. The house edge sits in the RTP, and the formula does not touch it. What the formula does is keep you in the game long enough for the distribution to behave as advertised, which is the only thing stake sizing can honestly deliver. Pelican Payday, Freya 1000, and the rest of the catalogue will pay what they pay. Your job is to still be holding a balance when the tail arrives.
Run the numbers before you spin. Divide your bankroll by 100 times the game’s per-spin standard deviation in stake units, round down to a sensible bet step, and treat that as your ceiling rather than your target. Log your sessions, refine the factor per game, and ignore every app store listing that claims to be the official Pragmatic Play mobile client. The browser is the client, the licensed operator holds your balance, and the arithmetic is the only edge you control.