RTP, Variance, and House Edge: The Math Behind Every Casino Game

Return-to-player is the long-run statistical return of the game expressed as a percentage of wagered turnover. A 96% RTP slot returns, in aggregate across millions of spins by all players combined, $96 of player winnings for every $100 of player turnover. The remaining $4 is the operator's gross gaming revenue. RTP is calculated by the game provider during the design and testing phase of the title, mathematically derived from the game's pay table, the probability distribution of outcomes, and any bonus mechanics. The result is verified by independent testing labs (eCOGRA, iTech Labs, GLI) before the game ships to operators.

RTP is a long-run, aggregate-across-players number. It is not what an individual player should expect to receive in any individual session. A single player's session at a 96% RTP slot can return anywhere from 0% (losing the full bankroll without a meaningful hit) to several thousand percent (catching a high-variance bonus round). Over millions of sessions, the aggregate result converges to 96%; over a single session, the result is dominated by variance, not by the RTP.

The RTP is published on the game's information panel — accessible from a 'i' icon, settings menu, or paytable button on the game's screen. Players should always verify the RTP before serious play. Some slot titles ship at multiple RTP variants (the same Starburst can be 96.05%, 94.05%, or 88.12% depending on the operator's chosen variant), so the published RTP is per-operator and per-title combined.

Higher RTP is structurally better for the player at constant variance. A 97% RTP game returns more of the wagered turnover than a 95% RTP game over the long run, by definition. However, RTP differences of 1-2% are dwarfed by variance in any reasonable-length session — a player at a high-volatility 97% slot can easily lose more than a player at a low-volatility 95% slot in the same session length, simply because the variance distribution overlaps. RTP matters; variance matters more in many practical contexts.

Variance describes how much individual outcomes deviate from the long-run RTP average. Two games can have identical RTPs but very different variance profiles. A low-variance slot pays small wins frequently (often, on every few spins) and rarely produces large wins; a high-variance slot pays small wins rarely (long dry spells) and produces large wins as the headline mechanic. Both return the same RTP over the long run; the distribution of how the RTP is paid out differs.

Variance is typically labeled qualitatively on the game's info panel: low, medium-low, medium, medium-high, high, or very high. Some providers publish a numerical hit frequency (the percentage of spins that produce any win at all — typically 25-30% for low-variance slots, 15-25% for medium, and 8-15% for high). The numerical hit frequency, combined with the maximum win potential (max win expressed as a multiple of bet size), gives a fuller picture of the variance distribution.

Low-variance slots are appropriate for players with smaller bankrolls relative to the session length, players who want frequent emotional reinforcement, and players who want predictable bankroll consumption per session. High-variance slots are appropriate for players with larger bankrolls relative to session length, players who can tolerate long dry spells in pursuit of bonus features, and players whose goal is the rare large-multiplier hit.

A practical heuristic: if your bankroll is less than 100x your average bet size, you should play low-to-medium variance slots, because high-variance slots have routine dry spells of 100+ spins between meaningful wins. If your bankroll is 500x+ your average bet, you can afford to ride out the variance of high-variance slots to chase the headline bonus events.

House edge is the complement of RTP — the percentage of wagered turnover the operator retains over the long run. A 96% RTP slot has a 4% house edge. A 99% RTP video poker variant has a 1% house edge. European roulette has a 2.70% house edge across all bet types. American roulette has 5.26%. Live blackjack with full basic strategy has approximately 0.5%. Baccarat banker bet has 1.06%.

House edge applied to total turnover gives the operator's expected revenue from the player. A player wagering $10,000 in turnover at a 4% house edge has an expected loss of $400 — the long-run integral of the player's wagering against the operator's edge. Note: this is expected loss, not certain loss. Variance can produce any actual result around this expectation.

Different house edges suit different player objectives. Players maximizing playing time per dollar of bankroll should target the lowest available house edges — live blackjack (0.5%), video poker at full-pay tables (1% or less), baccarat banker (1.06%). Players prioritizing entertainment and excitement over efficient bankroll burn typically choose mid-edge slot play (3-6% house edge typical) where the variance is the entertainment value. Players chasing rare large hits typically choose high-variance slot or game show play (3-6% house edge with variance distributed toward rare large events).

House edge per unit time matters as much as house edge per bet. A slot at 4% house edge taking 5 seconds per spin generates $4 of expected loss per $100 of average bet per minute (at 12 spins per minute, that's $4 expected loss per minute at $1/spin). A live blackjack table at 0.5% house edge taking 60 seconds per hand at $25 average bet generates $25 × 0.5% / 60 = $0.21 expected loss per minute. The live blackjack player loses 20x slower per minute of play, despite playing at higher single-bet amounts.

Slot RTPs across the segment vary by provider, by title, and by RTP variant chosen by the operator. Some provider-level patterns are worth knowing. NetEnt typically ships titles at 96.0% to 96.5% RTP in their default variants, with low-RTP variants (93-94%) available for operators in higher-tax jurisdictions. Play'n GO ships at 96.0% to 96.3% as default with similar low-RTP variants. Pragmatic Play ships at 96.0% to 96.5% as default; Pragmatic also ships specific high-RTP variants (97.5%+) for tournament play.

Premium slot providers (Yggdrasil, Push Gaming, Quickspin) tend to ship slightly higher default RTPs (96.5% to 97.0%) as a market positioning choice. High-volatility providers (Hacksaw Gaming, Nolimit City) ship default RTPs at 96.0% to 96.5% — the variance does the heavy lifting, not the RTP.

Crypto-friendly providers (BGaming, GameArt) ship RTPs at 96.0% to 97.0%, often with provably-fair RNG for player verification. Some BGaming titles ship at 98%+ for tournament or VIP-only play.

Low-RTP variants exist across most major providers. An operator's choice to integrate a low-RTP variant (88-92%) instead of a default variant is sometimes a tax-jurisdiction necessity (the operator pays high GGR taxes and needs to recover them) and sometimes a margin choice. Always check the RTP on the game info panel before serious play, regardless of how the title is advertised on the lobby. The same title at 88% versus 96% is a fundamentally different game in expected return terms.

Table games have a much tighter house-edge range than slots. The math is simpler — fewer outcome combinations, no bonus features, no random multipliers — so the house edge is more uniformly distributed across operators and titles.

Blackjack: house edge varies from approximately 0.4% to 1.5% depending on the rules of the variant. Six-deck shoe with dealer standing on soft 17, double after split allowed, split aces with single card, no surrender: house edge 0.48% with full basic strategy. Dealer hits soft 17 adds 0.22%. No double after split adds 0.14%. No re-splits adds 0.04%. Single-deck (rare online) reduces by 0.48%. Live dealer variants typically run at 0.5-0.7% house edge depending on the table's rules.

Baccarat: banker bet 1.06% house edge, player bet 1.24%, tie bet 14.36% (always avoid). Pair side bets typically 10-11% house edge. Lightning Baccarat and similar variants with random multipliers shift the house edge slightly higher (typically 1.2-1.5%) due to the multiplier mechanic.

Roulette: European single-zero 2.70% house edge, American double-zero 5.26%, French with la partage or en prison rules 1.35% on even-money bets (2.70% on inside bets). Lightning Roulette and similar variants run slightly higher than European (about 2.99%) due to the multiplier mechanic.

Video poker: house edge varies dramatically by paytable. Full-pay Jacks or Better has a 0.46% house edge with perfect strategy. Common short-pay variants (9/6 Jacks or Better, 8/5, 7/5) run at 0.46%, 2.7%, and 5.0% respectively. Players who play video poker should verify the paytable on the title before playing — short-pay variants are common online and the player should always seek full-pay or near-full-pay tables.

Live dealer and RNG versions of the same table game (live blackjack vs RNG blackjack, live roulette vs RNG roulette) usually have slightly different house edges. RNG versions typically run 0.05-0.15% lower house edge than live versions of the same rules, because the RNG has no live-studio operational overhead built into the math and can enforce strictly optimal shoe-management rules.

Live blackjack: typical house edge 0.5% on standard six-deck rules with basic strategy. RNG blackjack: typical house edge 0.35-0.45% on equivalent rules. The difference is real but small.

Live roulette: 2.70% house edge on European, 5.26% on American. RNG roulette: identical, since the math is determined by the wheel layout and not by the dealer.

Live baccarat: 1.06% on banker. RNG baccarat: 1.06% on banker, same math. Live game shows (Crazy Time, Monopoly Big Baller, Funky Time): no RNG equivalent at most operators; the live format is the product.

Players seeking lowest-edge play often gravitate toward RNG blackjack at high-RTP operators. Players prioritizing live experience accept the slightly higher edge in exchange for the studio experience and game show category access.

RTP describes long-run aggregate return. It tells you the expected percentage of turnover the game returns over millions of spins or hands. It does not tell you, for any specific spin, what will happen. It does not tell you whether you will hit a bonus on the next spin or the spin after 200. It does not tell you whether your session will produce a win or a loss.

The probability of any individual spin's outcome is independent of every previous spin. This is the structural property that distinguishes casino games from skill games — the past results carry no information about the future results. A slot that has produced 50 consecutive losing spins is not 'due' for a win; the probability distribution on spin 51 is identical to spin 1. Beliefs to the contrary (the 'gambler's fallacy') are a well-documented cognitive bias, not a property of the game.

Variance is what produces the streaks and dry spells. In a high-variance slot, dry spells of 100+ spins between meaningful wins are routine and expected within the game's mathematical model. The RTP is preserved across the full distribution; the dry spells are matched by occasional outsized hits. The player who quits during a dry spell, never returning to claim the high-variance hits the model anticipates, will experience an apparent RTP much lower than the published number.

Practical implication: RTP is most relevant for high-volume play across many sessions. A casual player whose session lengths are short relative to the variance window of the game will see realized returns that can deviate significantly from the published RTP, in either direction. The published RTP is a guide to the structural math of the game, not a prediction of any individual player's experience.

Variance dictates the bankroll a player should bring to a session relative to the average bet size. A practical framework: 'units' of bet size, with bankroll measured in units, and session length capped at a defined number of units. A standard recommendation: 100 to 200 units of bankroll for a single session at a medium-variance slot, with the session ending at either a 30% drawdown (60 units down from start) or a 50% gain (150 units total). High-variance slots require larger unit counts to weather expected dry spells; low-variance slots require smaller.

Average bet size determines effective bankroll. A player with $200 bankroll playing $1/spin slots has 200 units, suitable for medium-variance slot play. The same player playing $5/spin slots has 40 units — too small for medium-variance play, suitable for low-variance play only. Adjusting bet size to bankroll is the fundamental bankroll management discipline.

Session-end conditions matter as much as starting bankroll. A player who sets a stop-loss at 50% drawdown ($100 of a $200 starting bankroll) preserves capital for future sessions; a player who plays until the bankroll is exhausted has surrendered the entire amount to operator turnover. Cumulative across many sessions, players who set and honor stop-losses lose meaningfully less than players who play without.

Pace of play matters too. A player playing at high pace (high-volume slot spins, high-pace live blackjack) burns through the bankroll quickly even at favorable house edges. A player playing at moderate pace (slot at deliberate pace, live table at default speed) extracts more entertainment per unit of expected loss. Slowing the pace is the simplest way to extend entertainment value relative to the operator's structural edge.

The structural math of every casino game in 2026 is that the expected long-run result for the player is negative. This is a feature of the segment, not a bug. The math is the source of the operator's revenue and underlies every legitimate operator's business model. Choosing to play any casino game requires accepting this math and treating the activity as paid entertainment within a budget.

Within that frame, choosing games whose math matches your budget and entertainment preferences is the player's primary decision. Long sessions on small bankrolls call for low-variance, low-edge games (live blackjack, video poker, low-variance slots). Short sessions on larger bankrolls allow higher-variance play (high-variance slots, game shows, side bets). Sessions where the goal is the rare large win require very high variance and very deep bankroll relative to bet size.

The operator's house edge does not depend on game choice. Slots, table games, live dealer, instant-win — every category has a house edge, and the operator profits from the integral of player turnover across all of them. The choice between game types is therefore a player-side optimization based on preference (variance, pace, social, math) rather than a strategic move that beats the operator's edge.

Use the operator's player tools (deposit limits, session limits, time-outs, self-exclusion) to enforce the budget you set in advance. The single most reliable predictor of sustainable play is whether the player set limits in advance and honors them throughout the session, not which specific game category the player chose. The bureau's responsible gambling reference at /responsible covers the full tool set.