Expected Value

Definition and Mathematical Formalism

Expected value (EV) is the probability-weighted average of all possible outcomes of a random variable. For a discrete random variable X that can take values x_i with probabilities p_i, the expected value is defined as E[X] = Σ_i p_i x_i, provided the sum converges. For a continuous random variable with distribution function F, the corresponding definition is E[X] = ∫_{-∞}^{∞} x dF(x) when the integral exists. The concept of expectation is linear: for random variables X and Y and constants a and b, E[aX bY] = aE[X] bE[Y]. This linearity property is central to many analyses in games and gambling, as it allows decomposition of complex wagers into simpler components.^[1]

In gambling contexts, the expected value of a single wager expresses the average gain or loss per play in the long run. A wager with positive EV is favorable to the player (in expectation), while negative EV indicates an expected loss. Casinos typically operate games with negative EV for players; the magnitude of that negative EV is often described as the house edge or casino advantage. Practically, expected value is commonly expressed on a per-unit-bet basis, for example, as dollars per $1 wagered or as a percentage return to player (RTP = 1 EV when EV is given per unit bet).

Two common scenarios illustrate EV computations. First, a simple discrete example: consider a game where a fair coin is flipped; if heads the player receives $2, if tails the player loses $1. The outcomes are x_1 = 2 with probability p_1 = 0.5, and x_2 = −1 with probability p_2 = 0.5. The expected value is E[X] = 0.5·2 0.5·(−1) = 1 − 0.5 = 0.5. On average the player gains $0.50 per play. Second, a casino example: in European single-zero roulette (37 pockets), a straight-up number bet pays 35 to 1. The EV per unit bet is E[X] = (1/37)·35 (36/37)·(−1) = −1/37 ≈ −0.027027..., which corresponds to a house edge of ≈ 2.70%.

Expectation interacts with other probabilistic measures. Variance (Var[X]) quantifies dispersion around the expectation and is defined as Var[X] = E[(X − E[X])^2] = E[X^2] − (E[X])^2. High variance implies that actual outcomes will frequently deviate from EV over short horizons, even though the long-run average will approach EV under the law of large numbers. The strong law of large numbers states that the sample average of independent, identically distributed observations converges almost surely to the expected value as the number of trials tends to infinity; this result justifies using EV as the predictor of long-run performance in repeated play.^[1]

Table 1 gives a compact discrete example of expectation calculation for a hypothetical bet.

In this hypothetical example the expected value is $4.50 per play. The expected value alone does not convey risk: the variance of outcomes remains large because of the 5% chance of a very large payoff and the 60% chance of modest loss. Practitioners thus pair EV analysis with variance and utility theory when making decisions under risk.

Historical Development and Key Events

The mathematical notion of expectation emerged as probability theory developed in the 17th century. The correspondence between Blaise Pascal and Pierre de Fermat in the 1650s addressed the problem of points (how to fairly divide stakes in an interrupted game) and laid the groundwork for formal probability calculations. Christiaan Huygens published 'De Ratiociniis in Ludo Aleae' (On Reasoning in Games of Chance) in 1657, which contains one of the earliest known explicit discussions of expected value in game contexts and methods for computing fair payments for wagers.^[3]

In the 18th century, Daniel Bernoulli introduced the concept of expected utility in 1738 to resolve paradoxes that arise when expected monetary value does not match observed human preferences, most famously the St. Petersburg paradox. Bernoulli proposed that decision-makers maximize expected utility rather than expected monetary value, introducing a concave utility function to capture diminishing marginal utility of wealth. Later formalizations by John von Neumann and Oskar Morgenstern in the 20th century established axiomatic foundations for expected utility theory in 'Theory of Games and Economic Behavior' (1944), linking EV-type reasoning with rational choice under uncertainty and game-theoretic equilibrium concepts.^[4]

Within the gambling industry, the practical recognition of expected value as central to profitability predated rigorous mathematical formalism. Traditional games evolved house rules that guarantee a built-in negative EV for the player (the house edge). Notable historical dates include the formal analysis of specific games during the 18th and 19th centuries and the adoption of probabilistic models in the 20th century as casinos modernized operations and computerized bookkeeping. The analysis of card games such as blackjack advanced further in the mid-20th century when Edward O. Thorp applied probability and card-counting techniques in the 1960s to identify and exploit deviations from negative EV under basic strategy; Thorp's work demonstrated how skillful play can shift EV in a player's favor in certain conditions. Over the decades, expected value calculations also informed regulatory oversight, payout regulations for lotteries, and the design of electronic gaming machines where return-to-player (RTP) settings are implemented to define long-run EV.

'To decide among uncertain prospects one must compare their expectations under an appropriate scale of satisfaction.' - paraphrase of principles articulated in early utility theory and formalized by Bernoulli and later by von Neumann and Morgenstern.^[4]

Modern academic treatments trace the concept through mathematical probability texts and game-theory literature, while industry practice integrates EV into pricing of bets, calculation of promotions, and assessment of advantage play. Historic references and primary sources remain central for scholars studying the philosophical and mathematical evolution of expectation, while casino operators consult EV-based metrics for operational and marketing decisions.

Applications in Games, Casinos, and Wagering Strategy

Expected value is the principal analytic tool used to assess the fairness and profitability of bets in casino games. For any wager, the EV per unit bet can be computed from the payout schedule and the probability of each payout event. Common instances include roulette, blackjack, craps, baccarat, and slot machines. Each game has a characteristic house edge or negative EV for standard play; variations in rules and player skill can alter the effective EV experienced by a player.

Examples and calculations:

Skill-based actions can change EV. For example, in blackjack, adoption of optimal basic strategy minimizes the house edge and thereby maximizes player EV among non-counting players. Card counting can provide a player with a positive expectation under favorable conditions and bet sizing rules, effectively converting a negative-EV game into a positive-EV opportunity when the count indicates a richness of ten-value cards. Similarly, in sports betting, accurate probabilities estimated by skilled handicappers can produce positive expected value bets when the bookmaker's odds are mispriced relative to true likelihoods.

Expected value also guides promotional evaluation. Casinos offer free bets, match play, and loyalty comps; the effective EV of these promotions is the net expected monetary benefit after accounting for wagering requirements, match rates, and game-specific edge. For instance, a 'free bet' with a wagering requirement that forces play on high-house-edge slots may have substantially reduced or even negative realized EV for the player, despite appearing valuable at face value.

From a risk-management perspective, EV must be considered alongside variance and bankroll considerations. The Kelly criterion provides a method to size bets to maximize expected logarithmic growth of capital when an edge is known; it uses EV and variance information to determine a fraction of bankroll to wager. Conservative players may prefer fractional Kelly or fixed-percentage staking to manage drawdown risk. Additionally, the time horizon matters: over short sessions, variance may dominate and cause large deviations from EV; over long horizons, the law of large numbers implies the realized average will converge to EV, subject to independence and stationarity assumptions.

Practical rules and terms commonly used by professionals:

• House edge: long-run average loss as a percentage of initial wager.
• Return to Player (RTP): complement of house edge, typically expressed as a percentage of stake returned on average.
• Advantage play: strategies that alter expected value in favor of the player (e.g., card counting, hole-carding, advantage promotions).
• Expected utility: decision criterion that may replace EV when players have nonlinear preferences over outcomes.

Proper application of EV requires accurate estimation of probabilities. Misestimation leads to incorrect EV calculations and suboptimal decisions. In regulated environments, transparent disclosure of RTP for electronic machines provides a metric of EV for consumers; in unregulated or obscure markets, EV estimates may be unreliable. Operators use EV calculations to set rules and paytables to ensure long-term profitability while managing regulatory compliance and marketing considerations.

Notes

1. [1] Expectation (mathematics) - general reference material on expectation and properties such as linearity and the law of large numbers; see standard treatments in probability theory and the relevant Wikipedia entry.
2. [2] Roulette and casino game mechanics - reference descriptions of European and American roulette rules, payout schedules, and house edge calculations; see casino game reference materials and the roulette Wikipedia entry.
3. [3] Historical sources - Pascal–Fermat correspondence (1654) and Christiaan Huygens' 1657 work 'De Ratiociniis in Ludo Aleae' discuss early probability and expected-value ideas; see history of probability summaries and the relevant Wikipedia entries on the problem of points and Huygens.
4. [4] Utility and game theory - Daniel Bernoulli's 1738 paper on expected utility and the von Neumann–Morgenstern utility theorem (1944) formalize utility-based decision-making; see the Wikipedia entries on the St. Petersburg paradox, expected utility, and the von Neumann–Morgenstern utility theorem.

References above are provided as descriptive citations. For further reading consult standard probability texts and the collection of historical sources summarized in public-domain encyclopedic repositories such as Wikipedia.