Every casino game has a house edge. You probably know that already. What you might not know is that two games with the exact same house edge can feel completely different to play, one grinds your bankroll down slowly, the other whipsaws it up and down like a heart monitor. The difference is variance. This page explains what variance and volatility actually measure, how they differ across every major casino game, and how to use the numbers to size your bankroll and pick games that match how you actually want to play.
Short answer: they overlap, but they’re not identical.
Variance is a precise statistical measure, the average of squared deviations from the mean. Standard deviation is its square root. For any game with known rules and payouts, these have exact, calculable values.
Volatility is the casino industry’s shorthand for risk profile, how bumpy the ride feels. For table games and video poker, where the maths is fully transparent, the two terms are effectively interchangeable because you can calculate exact SD figures from published paytables.
For slots, the terms diverge. Developers and operators use “volatility” to bundle together hit frequency, payout concentration, and max-win potential into a single risk label. Gambling mathematicians have criticised this usage because hit frequency and standard deviation are related but separate metrics. A game could hit frequently yet still be extremely volatile if most of those hits are tiny and the real returns are concentrated in rare bonus features.
The verdict: when you see a specific SD number on this page, that’s the precise mathematical figure. When a slot is described as “high volatility,” treat it as a risk profile that combines several factors. The distinction matters most when comparing across game types, which is what the next section does.
| Game | Typical Bet | SD / Unit Bet | What This Means for You |
| Pai Gow Poker | Base game | 0.75 | Smallest swings in the casino. Sessions feel like a slow, steady grind. |
| Baccarat (Banker/Player) | Banker or Player | 0.93 | Very smooth. Near even-money with a tiny edge. Minimal surprises. |
| French Roulette | Even money (la partage) | 0.99 | Smooth ride. Small wins and losses oscillate predictably. |
| Single-Zero Roulette | Even money | 1.00 | Baseline variance. Textbook near-even-money bet. |
| Craps | Pass Line only | ≈1.00 | Smooth on Pass alone. Adding odds increases variance significantly. |
| Blackjack | Basic strategy | 1.14 | Moderate. Doubles and splits add slight extra swing. |
| 3-Card Poker | Ante & Play | 1.64 | Moderate-high. Bonus payouts for premium hands add swing. |
| Craps | Pass + 3-4-5× Odds | ≈3.0 | Much more volatile. Odds bets carry zero edge but high variance. |
| Video Poker: 9/6 JoB | Max coin | 4.42 | High. Royal flush concentration drives a wide result spread. |
| Ultimate Texas Hold’em | Ante only | 4.94 | High. Large raises on strong hands create big swings. |
| Let It Ride | Base bet | 5.17 | High. Bonus-heavy paytable concentrates returns in rare hands. |
| Video Poker: Deuces Wild | Max coin | ≈5.07 | Very high. Wild cards push more value into top-end payouts. |
| Slots | Varies by game | VI 8–15+ | Wild swings. Bankroll can evaporate or multiply in minutes. |
| Keno | Varies by ticket | Very high | Extreme. Lottery-like payout structure with rare, large wins. |
A few things jump out. Pai Gow Poker, baccarat, and even-money roulette all cluster below SD 1.0, these are the games where a $500 bankroll can comfortably sustain a long evening. At the other end, Let It Ride, video poker variants, and slots produce SD figures four to six times higher. Same casino, same house edge range, completely different experience.
Within a single game, bet selection changes everything. Craps on the Pass Line alone sits at SD ≈1.0. Add full odds and it jumps to roughly 3.0. The house edge on the odds bet is zero, but the variance cost is real, you need a much bigger bankroll to ride the swings.
For table games and video poker, you can calculate SD directly from published paytables. Slots don’t work that way. The exact probability of every outcome is locked inside the game’s PAR sheet (Probability Accounting Report), which developers aren’t required to publish in most jurisdictions. That’s why the industry uses a separate metric: the Volatility Index (VI).
The VI scale runs from 0 to 20. It’s derived from simulating millions of spins and measuring the standard deviation of returns over a defined window (typically 10,000 spins per session in the simulation). The convention breaks down roughly as:
Players can’t calculate VI themselves because the underlying probability tables aren’t public. What you can do: check the game’s info screen or paytable (some developers display a volatility rating), look at operator ratings (BetMGM, for example, uses a 1–5 scale), or test the game in free play over several hundred spins to get a rough feel.
Some modern slots offer adjustable volatility, letting you choose between payout profiles within the same game. Worth looking for if you want control over your session experience.
Slot examples by tier:
Low volatility: Starburst (NetEnt) – RTP ~96.09%, max win ~500× stake.
High volatility: Gates of Olympus (Pragmatic Play) – RTP ~96.5%, max win ~5,000× stake.
The max-win multiple is one of the most reliable shorthand indicators. A game capped at 500× is almost certainly low-to-medium volatility. A game advertising 5,000× or 10,000× is high volatility by definition, those payouts can only exist if the game withholds returns from the majority of spins and concentrates them into rare events.
Required bankroll (95% confidence) = 2 × √(variance) × √(number of hands) × bet size
This gives you the bankroll that, 95% of the time, will survive the session without going bust. Three worked examples:
Blackjack, basic strategy. Variance per hand ≈1.30. Bet size $10. Session of 400 hands (roughly four hours at a table). Required bankroll = 2 × √1.30 × √400 × $10 = 2 × 1.14 × 20 × $10 = $456.
9/6 Jacks or Better video poker. Variance per hand = 19.51. Bet size $1.25 (five quarters). Session of 600 hands (roughly one hour). Required bankroll = 2 × √19.51 × √600 × $1.25 = 2 × 4.42 × 24.49 × $1.25 = ≈$271.
Double Double Bonus video poker. Variance per hand ≈28.26. Same bet and session length. Required bankroll = 2 × √28.26 × √600 × $1.25 = 2 × 5.32 × 24.49 × $1.25 = ≈$326.
Double Double Bonus requires roughly 20% more bankroll than Jacks or Better for the same session, even though both are video poker at similar RTPs. That’s the variance difference expressed in pounds and dollars.
Bankroll Quick-Reference Table
| Game | Variance | Bet Size | Hands (Session) | Bankroll (95%) |
| Blackjack (basic strategy) | 1.30 | $10 | 400 (≈4 hrs) | $456 |
| Craps (Pass + 3-4-5× Odds) | ≈9.0 | $10 | 400 (≈4 hrs) | $1,200 |
| 9/6 Jacks or Better VP | 19.51 | $1.25 | 600 (≈1 hr) | $271 |
| Double Double Bonus VP | 28.26 | $1.25 | 600 (≈1 hr) | $326 |
| High-Volatility Slot (est.) | ≈80–120 | $1.00 | 500 (≈1 hr) | $400–$500 |
Slot estimate uses a typical high-volatility VI range converted to approximate variance. All figures are illustrative guides, not guarantees. Sources: Casino Player Magazine.
The formula doesn’t guarantee anything, it gives you a 95% survival threshold, meaning 1 in 20 sessions you’ll still bust through it. But it moves bankroll planning from guesswork to something grounded in actual math.
Risk of ruin is the probability that you’ll lose your entire bankroll before either quitting or hitting a target. It’s the practical consequence of variance applied over time.
The core tension is straightforward. Standard deviation grows with √n (the square root of hands played), but expected loss grows with n (linearly). Over a short session, variance dominates, results scatter widely, and winning sessions are common. Over a long stretch, the house edge accumulates faster than the variance spread widens. The edge overtakes the noise. This is why short gambling trips produce winners, but the casino’s annual P&L is almost boringly predictable.
What this means for you: for any casino game without an exploitable edge, the risk of ruin over infinite play is mathematically 100%. Not because the games are rigged, but because a small persistent negative expectation compounds over enough trials to consume any finite bankroll. Session limits work. Lifetime targets don’t.
Variance is not your enemy. It’s the mechanism that makes winning sessions possible despite a negative edge. Understanding how it interacts with the house edge over time is what separates deliberate play from hopeful play. For broader context on how these ideas connect, see our casino fairness and math overview. And for practical guidance on session planning and limits, our responsible gambling resources are a good next step.
Common Misconceptions About Variance
“High-volatility games have higher RTP.”
Volatility and RTP are independent measurements. A high-volatility slot can have an RTP of 94%, and a low-volatility slot can return 97%. Volatility describes how returns are distributed, not how much the game returns overall.
“I’m due for a win after a long losing streak.”
The gambler’s fallacy. Each spin, hand or roll is independent. The game has no memory. A slot that hasn’t paid out in 200 spins is no more likely to pay on spin 201 than it was on spin 1. Variance produces streaks, that’s what it does, but those streaks carry zero predictive information about what comes next.
“Casinos change slot volatility during busy periods.”
Regulated gaming jurisdictions require that a slot’s parameters, RTP, hit frequency, and volatility profile remain fixed as approved. Changing them requires a new certification submission to the regulator. The notion that casinos “tighten” machines on Friday nights is a persistent myth with no basis in how the regulatory framework operates.
“Betting bigger reduces volatility.”
Bet size scales the dollar value of your swings but doesn’t change per-unit variance. If a game has SD 5.0 per unit bet, betting $1 gives you $5 swings per wager and betting $10 gives you $50 swings. The ratio is identical. Bigger bets make the ride more volatile in absolute terms, not less.
“You can tell a slot’s volatility by playing a few dozen spins.”
Not reliably. Volatility Index figures are derived from simulating millions of spins. A sample of 30 or 50 spins tells you almost nothing, you could hit a bonus feature on spin 5 of a high-volatility game and wrongly conclude it pays frequently. If the game or operator doesn’t publish a volatility rating, free play over at least 500 spins gives a rough feel, but even that is a crude estimate compared to the actual simulated data.
Scenario 1: “You’re on holiday with $300 for the casino floor tonight.”
Your goal is session time. You want the money to last the evening, not vanish in 20 minutes. Low-variance games are your friends: baccarat Banker (SD 0.93), even-money roulette (SD 1.00), or blackjack at a modest minimum. The bankroll formula says a $10-minimum blackjack session needs ≈$456 for four hours, so either find a $5 table, plan for two hours, or stick to the even-lower-variance options. Avoid high-volatility slots, your $300 could be gone in under 100 spins.
Scenario 2: “You play video poker weekly and manage a monthly bankroll.”
You’re a regular who can absorb short-term losses across many sessions. The bankroll formula becomes your session-planning tool: know your game’s variance (9/6 Jacks or Better = 19.51, Double Double Bonus ≈28.26), size each session accordingly, and track results over time. Medium-to-high variance is perfectly workable here as long as your total monthly bankroll supports the per-session requirement across enough visits.
Scenario 3: “You want a shot at a big win and accept the risk.”
You’re buying lottery tickets, not commuting. High-volatility slots, progressive jackpot games, or aggressive video poker variants fit this goal. Go in with eyes open: the bankroll requirement is higher, dry spells are longer, and the most likely outcome of any single session is a loss. Set a hard session limit, the amount you’re prepared to spend for the experience, and treat any win as the exception it statistically is.
Each approach is valid. None is more “correct” than the others. Variance is a planning tool, it tells you what kind of experience to expect so you can choose deliberately rather than discover it the hard way.