Expected Value: The Math Behind Rational Choice

Expected value is a foundational concept in decision theory. The expected value of an action is the average outcome you would get if you took the action many times: the sum of each possible outcome, weighted by its probability.

For a coin flip where you win $10 on heads and lose $5 on tails, the expected value is (0.5 × 10) + (0.5 × -5) = $2.50. On average, each flip is worth $2.50 to you. Take the bet enough times and you will come out ahead.

Why it’s useful

Expected value gives you a way to compare uncertain options on a single scale. Two job offers with different probabilities of success and different compensation can be compared by their expected values. Two investments with different risk profiles can be compared. Two policy options with different outcomes can be evaluated against each other.

The framework is general enough to apply almost anywhere there’s uncertainty about outcomes. Wherever you can estimate probabilities and values, expected value provides a coherent way to combine them.

When it’s the right tool

Expected value is the right framework when several conditions hold.

The decision will be repeated, or is equivalent to a sequence of decisions. Insurance companies make decisions on expected value because they make many similar bets. An individual considering insurance may not, but if the decision is one of many similar decisions over a lifetime, the aggregate behaves like a repeated bet.

The stakes are similar across outcomes. Expected value treats a $1,000 gain and a $1,000 loss symmetrically. For decisions where the loss would be catastrophic (loss of all your savings, say), expected value alone doesn’t capture the right weighting — you should use it alongside other considerations.

The probabilities are at least roughly estimable. Expected value requires probability estimates. If you cannot estimate them, the calculation produces a precision that is not warranted.

Why people fail to use it

Most people have heard of expected value. Far fewer apply it in their actual decisions. The gap reflects several factors.

Expected value calculations require effort. The instinctive system for evaluating decisions is fast and intuitive. The expected value system is slow and analytical. In most situations, the fast system runs first and produces a verdict before the slow system has a chance to engage.

Expected value calculations produce counter-intuitive conclusions. The intuitive system is loss-averse and probability-distorting. Expected value reasoning often says “accept this bet even though you might lose,” which conflicts with the intuitive aversion to potential losses.

Expected value calculations require numerical inputs that are uncomfortable to produce. Most decisions don’t come with explicit probabilities and values. The decision-maker has to estimate them, often based on limited information. The estimation feels arbitrary even when it tracks reality better than the alternatives.

When intuition beats it

There are cases where expected-value reasoning is the wrong tool.

Decisions with rare catastrophic outcomes — where the worst case would be ruinous — cannot be made on expected-value terms alone. The expected value might be positive, but if accepting the bet exposes you to outcomes you cannot recover from, you should usually decline. Personal finance literature calls this “risk of ruin” reasoning, and it modifies pure expected-value calculations in important ways.

Decisions in domains where you cannot estimate probabilities reasonably also fail the expected-value framework. If you genuinely don’t know how likely each outcome is, plugging in guesses produces calculations that look rigorous but aren’t.

For the broad middle — decisions of modest stakes, with estimable probabilities, made under repeated conditions — expected value remains the cleanest framework. Most everyday financial and consumer decisions fall in this middle. Practitioners who use the framework consistently outperform those who don’t.