Gambling Odds and Probability, Explained
Most gambling losses are not bad luck — they are bad math, accepted in advance without being read. Odds, probability, payout, expected value and variance are five different numbers, and casinos profit from players who blur them together. This guide separates them with worked examples from real games: dice at a flat 1% edge, roulette at 2.70% and 5.26%, and blackjack paying 3:2 on a natural. None of it requires more than arithmetic, and all of it changes how you read a betting screen.
See the math live on dice →Probability, odds and payout are three different numbers
Probability is the chance an event happens, written as a fraction or percentage: a single number on a European roulette wheel hits with probability 1/37, about 2.70%. Odds express the same fact as a ratio of failures to successes: 36 to 1 against. Payout is what the casino actually pays you — and it is the only one of the three the casino controls. Roulette pays a single number at 35 to 1, one unit short of the true 36-to-1 odds. That gap between true odds and paid odds is the entire business model of every casino ever built. When you evaluate any bet, the question is never 'what are the odds?' in isolation — it is 'how far does the payout fall short of the true odds?' A game can have terrible win probability and still be fair value if the payout matches, and a game can win half the time and still bleed you if it pays under even money.
Expected value: the one number that decides everything
Expected value (EV) is the average result per bet if you could repeat the bet forever: multiply each outcome by its probability and add them up. A worked example from Llamabet dice, where the payout formula is 99 divided by your win chance. Choose a 50% win chance and the payout multiplier is 99 ÷ 50 = 1.98x. Bet 1 SUI: half the time you receive 1.98 SUI, half the time you receive nothing. EV = (0.50 × 1.98) + (0.50 × 0) − 1 = −0.01 SUI. You expect to lose one hundredth of every unit wagered — a flat 1% house edge, and it stays 1% at every win chance you pick, because the formula scales the payout to the probability. Pick 10%: 99 ÷ 10 = 9.9x, and EV is (0.10 × 9.9) − 1 = −0.01 again. Negative-EV bets are the price of the entertainment; the skill is knowing exactly what that price is before you pay it.
Worked edges: roulette and blackjack
European (single-zero) roulette has 37 pockets and pays single numbers at 35 to 1. EV per unit on a straight bet: (1/37 × 36) − 1 = −0.027, a 2.70% house edge — and remarkably, almost every bet on the layout carries that same 2.70%, because the payouts are all scaled to a 36-pocket wheel while the real wheel has 37. American (double-zero) roulette adds a 38th pocket without touching the payouts, nearly doubling the edge to 5.26%. Same game, same felt, twice the price. Blackjack is harder to reduce to one number because your decisions matter, but the payout rules move the edge directly: a natural blackjack paying 3:2 rather than 6:5 is worth roughly 1.4% to the player, which is why 3:2 tables (Llamabet pays 3:2) are the ones worth sitting at. Played with sound basic strategy, blackjack's edge lands around half a percent — among the cheapest games in any casino.
Variance: why the average is not what happens to you
House edge tells you the long-run average; variance tells you how violently individual sessions swing around it. In plain language: a 1% edge does not mean you lose 1% every session. It means that across millions of bets, losses converge toward 1% of turnover — while any given hour can be +80% or −60%. Standard deviation is just the typical size of a swing. A 50% dice bet has modest swings; a 2% dice bet at 49.5x has brutal ones, even though both carry the identical 1% edge. High variance means longer droughts and bigger spikes, which means you need a larger bankroll relative to your bet size to survive the same game. This is why two players can play the same game at the same edge and one goes broke while the other grinds along: they did not choose different odds, they chose different variance against the same bankroll.
The gambler's fallacy: dice have no memory
The gambler's fallacy is the belief that past independent results change future probabilities — that after five reds, black is 'due.' It is false, and provably so: each spin, roll and shuffle is an independent event. The wheel does not know what it did last time, and on a provably-fair platform you can literally verify this, because each outcome is derived from a fresh server-seed commitment published before your bet, not from any history. The fallacy is dangerous because it feels like insight. It powers loss-chasing ('I'm due'), the Martingale system (doubling after losses until a finite bankroll meets an unbounded losing streak), and most betting 'systems' sold online. No staking pattern can turn a negative-EV game positive; systems only rearrange when the losses arrive. The one honest counterexample is games with memory — a blackjack shoe genuinely depletes as cards leave it — but that is card removal, not luck being owed.
Why short-term results lie to you
The law of large numbers guarantees convergence to the expected value — eventually. Over 100 bets, the noise dwarfs the edge: a dice player at 50%/1.98x is meaningfully likely to be up after 100 bets despite the negative EV. Over 100,000 bets, the 1% edge is nearly inescapable. This asymmetry cuts both ways. A winning weekend is not evidence you have a system; a losing weekend is not evidence the game is rigged. Both are ordinary variance around a small negative mean. Casinos understand this deeply: they welcome short-term winners because short-term winners keep playing, and volume is the house's only strategy. Your defense is the same arithmetic — judge games by their edge and your exposure by total amount wagered, not by how last night went. If you wager 10,000 SUI across a month at a 1% edge, your expected cost is 100 SUI regardless of the path.
Comparing games by house edge
Once you think in edge, comparing games becomes shopping. Dice at a flat 1% is the benchmark. Blackjack with 3:2 naturals and sound play sits near it. Single-zero roulette costs 2.70%; double-zero costs 5.26% — avoid the extra zero whenever you have the choice. Poker against other players is different in kind: the house takes rake, not an edge on outcomes, so skill can be positive-EV. Two caveats. First, edge is a percentage of amount wagered, so a low-edge game played fast can cost more per hour than a high-edge game played slowly. Second, edge assumes the game is honest — on-chain commit-reveal fairness is what makes the published number trustworthy rather than a brochure claim, since you can recompute every outcome yourself. Know the price, pick the variance that fits your bankroll, and never confuse a hot streak with an edge.
Frequently asked questions
What is the difference between odds and probability in gambling?
Probability is the chance of winning expressed as a fraction of all outcomes — 1/37 for a single roulette number on a European wheel. Odds express the same event as failures to successes: 36 to 1 against. The payout is a third, separate number set by the casino, and the house edge lives in the gap between the true odds (36 to 1) and the paid odds (35 to 1).
How do you calculate expected value on a casino bet?
Multiply each outcome by its probability, sum them, and subtract your stake. On a dice bet with a 50% win chance paying 1.98x: (0.50 × 1.98) − 1 = −0.01, so you expect to lose 1% of each bet on average. That figure is the house edge, and it holds over the long run regardless of short-term streaks.
Is the gambler's fallacy real — are numbers ever due to hit?
No outcome is ever due. Each spin or roll is independent, so five reds in a row leave the next spin at exactly the same probability as always. On a provably-fair casino you can verify independence directly, because each result derives from a seed committed on-chain before your bet, not from previous results.