Expected Value (EV)

Key Takeaways

• EV is the probability-weighted average outcome of a bet — the single number that summarizes whether, and how fast, a wager loses money on average.
• For casino games, EV and house edge are the same fact: a 2.7% edge means every $1 staked has an EV of −$0.027.
• EV says nothing about any single session; variance governs the short run. Casinos sell variance and collect EV.
• Almost every casino bet is negative EV. The rational frame is EV as the price of entertainment, computable in advance as expected loss per hour.
• Comparing games by EV per hour — not by win stories — is the closest thing to an objective "which game is cheaper" answer.

How Expected Value Works

Take every outcome a bet can produce, multiply each by its probability, sum the results, and subtract the stake:

EV = Σ (probability × payout) − stake

A fair coin flip paying $2 on heads for a $1 stake has EV = (0.5 × $2) − $1 = $0: a break-even game. Casino games shift the payout or the probability slightly below fair. Single-number European roulette pays 36-to-1 odds worth of return (35-to-1 plus stake) against a true chance of 1 in 37:

EV = (1/37 × $36) − $1 = −$0.027  →  −2.7% of stake

That −2.7% is the house edge; equivalently the bet's RTP is 97.3%. Every casino wager can be reduced to this computation, however elaborate its presentation. A slot is thousands of outcomes instead of two, but the sum is the same construction — which is exactly how studios compute the RTP printed on the info screen.

The law of large numbers

EV is invisible in one bet and inescapable in a million. If you make n independent $1 bets at −2.7% EV, your expected total is −$0.027 × n, while the standard deviation grows only with √n. The ratio of noise to drift shrinks as n grows: short sessions are dominated by luck, long aggregates by EV. This asymmetry is the entire business model of a casino — individual players experience variance; the operator, aggregating millions of bets, experiences EV almost exactly.

The Math: expected loss per hour

EV becomes practical when converted to a session cost:

expected loss / hour = average bet × rounds per hour × house edge

Two things jump out. First, pace matters as much as edge: slots are not the worst edge on the list, but their speed multiplies the cost. Second, within slots, the RTP tier difference (96% vs 92%) doubles the hourly price of an identical experience — the practical argument for checking RTP before playing.

EV vs Variance

EV is the average; variance is the spread around it. Two bets with identical EV can be wildly different products: a −4% EV slot can deliver its average as a slow drip of small losses (low volatility) or as long droughts punctuated by rare big hits (high volatility). Neither structure changes the average cost — only the shape of the experience and the bankroll needed to survive it. The full treatment is in Variance vs Volatility; the one-line summary: choose games by EV for price, by variance for feel.

Where positive EV actually exists

• Full-pay video poker — a handful of paytables exceed 100% return with computer-perfect play; operators know where these machines are.
• Promotions and bonuses — a bonus whose value exceeds the expected cost of its wagering requirements is positive EV; bonus terms exist largely to prevent this.
• Progressive jackpots past a break-even seed — the headline EV can turn positive while the experienced EV for any individual remains a near-certain loss plus a lottery ticket.
• Skill edges in poker (against other players) and advantage play in blackjack — outside the scope of standard casino EV, and actively countered.

The pattern: positive EV is rare, small, effortful, and unwelcome. Building play habits around hunting it is a profession, not a pastime.

For Players

• Price your session before playing. Bet × pace × edge = expected cost per hour. If the number reads acceptable as an entertainment price, the game is honestly priced for you; if not, no strategy changes it.
• No staking system alters EV. Martingale and its relatives rearrange when losses occur, not their average size — every bet in the sequence keeps the same negative EV, and betting more after losses only concentrates risk.
• Winning sessions are consistent with negative EV. Expect them regularly; they are variance, not evidence the math has changed for you.
• Responsible-gambling lens: EV guarantees that money spent gambling is, on average, spent — treat it as a leisure budget, never as income or recovery of losses. If the "win it back" frame appears, that is chasing losses, the canonical marker of harm.

For the Industry

• EV per hour is the honest unit economics of game supply. Portfolio comparisons on RTP alone ignore pace; edge × velocity determines actual hold per seat-hour, which is why turbo modes and round-time UX changes have direct revenue significance.
• Bonus EV modelling is a core risk function. Every promotion is a transfer of EV to players; wagering requirements, game weighting and max-bet rules exist to keep the transferred amount below the acquisition value. Mispriced promotions attract precisely the players who can compute EV.
• Communicating EV honestly is becoming table stakes. Regulators increasingly require expected-loss style disclosures; products that frame cost transparently (as this article does) align with where YMYL-era compliance is heading.