Concept/Probability and Statistics/No. 0343
Expected Value
The repair shop offers a $180 protection plan, and you weigh that price against repairs you might never need.
Also called Mathematical Expectation
- Evidence
- Well established
- Read
- 6 min
- Links
- 14 connections
01You've seen this when…
- in life
Your flight is oversold. The airline offers $400 to give up your seat, but missing your connection could mean a hotel bill and a lost day’s pay.
- at work
A client offers a rush job with a generous fee. Your team estimates the profit if everything goes smoothly, then subtracts the chance-weighted cost of overtime and penalties.
- out in the world
A town compares flood defenses costing $2 million with the damage they might prevent. Most years bring no flood, but one bad year could destroy an entire neighborhood.
02The idea
A protection plan has one price. Going without it has several possible prices: nothing if the equipment works, a small repair bill if one part fails, a large bill if it breaks completely. Expected value puts those possibilities on a common footing.
Multiply each outcome by its probability, then add the results:
Expected value = sum of each outcome × its probability.
A free draw pays $100 with a 10% chance and nothing with a 90% chance. Its expected payout is:
0.10 × $100 + 0.90 × $0 = $10.
The draw pays either $100 or nothing, so the $10 average falls between the two possible payouts. The expected value is the distribution’s average, not a prediction for your next attempt. The most likely outcome can differ from that average, which may fall outside the set of available outcomes.
The quantities must be comparable. You can calculate expected dollars, hours, repairs, or lives saved. Combining dollars and hours requires deciding how to value them. For a decision about money, use net outcomes: what remains after subtracting the relevant costs from the headline payout.
For more complicated choices, a decision tree helps keep the possible paths and their probabilities straight.
03Why it matters
Expected value brings outcomes hidden by an attractive headline into the calculation. A large prize with a tiny chance may be worth less, on average, than a modest payment you receive reliably. A rare loss can outweigh many ordinary gains.
Expected value is especially useful for planning across many comparable cases, from warranty claims and delivery delays to insurance losses and a portfolio of projects. The law of large numbers explains why, under suitable conditions, averages from repeated trials tend toward their expected value. A short run can still produce an average far from the expected value.
Expected values also add. The expected total cost of ten repairs is the sum of their expected costs, even if the repairs are correlated. But correlation changes the risk of several expensive repairs arriving together.
Used in cost-benefit analysis, the calculation makes an outcome’s likelihood and consequences explicit so you can compare the choice with other uses of the same resources. That last comparison is the opportunity cost. An option with a positive expected return can still be less attractive than another option that offers more for the same commitment.
04A worked example
Consider an illustrative print shop choosing between a routine order and a rush order for the same production slot. It has $6,000 in cash reserves and no emergency borrowing available. The routine order guarantees $600 in net profit. The rush order has an estimated 80% chance of producing $3,000 in net profit and a 20% chance of producing an $8,000 net loss, including rework and late-delivery penalties.
What it looks like The rush order is more attractive. Its expected profit is 0.80 × $3,000 + 0.20 × −$8,000 = $800, compared with $600 for the routine order. The difference is $200.
What’s actually going on $800 correctly describes the rush order’s average under those assumptions. Each possible result differs from that average. The shop faces a one-in-five chance of a loss larger than its reserves. Choosing the higher expected profit could leave it unable to pay its bills. The average advantage comes with that risk.
What would have helped Calculating expected profit first, then examining the downside separately. The shop could take the routine order or reduce the rush order’s downside by negotiating a cap on penalties or buying backup capacity. A sensitivity analysis also helps: if the chance of the $8,000 loss rises to 25%, the rush order’s expected profit falls to $250. The recommendation depends heavily on a probability that may be only a rough estimate.
05Where people trip up
- Treating expected as promised. An expected repair cost of $100 describes an average. Prepare for the possible bills by looking at both their amounts and their probabilities.
- Confusing the average with the typical case. A few enormous payouts can pull the expected value far above what most people receive. Mean vs. median is the useful distinction: the mean reflects the total spread across everyone; the median marks the middle outcome.
- Mixing payouts with profits. A 10% chance of receiving $100 has an expected payout of $10. If entry costs $12, the expected net result is a $2 loss. Include costs consistently and avoid counting the same cost twice.
- Giving guesses a false air of precision. The arithmetic can be exact while the inputs are poor. Use a plausible range of probabilities and losses, then check whether the preferred option changes.
- Leaving out inconvenient branches. An average based on ordinary success and failure may cover the wrong set of possibilities if the calculation leaves out cancellation, liability, or a missed deadline. Check that outcomes cover the relevant cases and probabilities add to 100%.
- Assuming repetition removes danger. A single loss can end your ability to continue, even when you’re repeating a favorable bet. Check risk of ruin and whether shared conditions could make several bets fail together.
06Where it doesn’t settle the decision
Expected value summarizes one feature of a distribution. Two options can have the same expected value and very different spreads, captured partly by their variance. One pays $500 reliably; another pays nothing most of the time and occasionally pays a fortune. Their averages can match without making them interchangeable.
Maximizing expected dollars and maximizing well-being can lead to different choices. Losing $5,000 may threaten your rent, while gaining $5,000 merely improves your vacation. Expected utility theory handles this by averaging how valuable outcomes are to the person choosing. Paying for insurance can therefore be reasonable even when it raises your expected monetary cost.
Some distributions also have no finite expected value. More practically, rare extremes can make an estimate unstable. When the largest plausible loss dominates the answer and its probability is poorly known, report that uncertainty so the decision accounts for the estimate’s limits.
07Roots
Christiaan Huygens visited Paris in 1655 and learned about mathematical questions surrounding games of chance. One problem was especially revealing: players had put money into a game, but play stopped before either had won. How should they divide the pot?
Blaise Pascal and Pierre de Fermat had tackled this problem in their correspondence the previous year. Dividing the money fairly required calculating each player’s chance of eventually winning from the current score. If equally matched players need three wins, and the score is two to one, the leader has a three-in-four chance of winning: win the next round, or lose it and win the deciding round. A fair division gives the leader three-quarters of the pot.
Huygens turned reasoning about wagers into a systematic account, published in Latin in 1657 as De ratiociniis in ludo aleae. He treated an uncertain payoff as something with a fair present worth. That made games of chance a laboratory for a much broader idea: possible future outcomes can be valued before any one of them happens.
The distinction between money and its value emerged later. In 1738, Daniel Bernoulli argued that a person’s wealth affects how much an additional sum is worth to them. Expected monetary value remained mathematically useful, while a rule for choosing could also account for the value of money to the person.
08How solid is this?
Expected value’s well-established properties follow from its mathematical definition. Its usefulness depends on the outcomes and probabilities supplied; whether maximizing expected money leads to a good decision depends on the circumstances.
09Connections
- Helps counter Probability Neglect
- Part ofCost-Benefit Analysis, Decision Tree, Expected Utility Theory
- Includes Mean vs. Median
- See also BATNA, Sensitivity Analysis, Variance, Risk of Ruin, Law of Large Numbers, Opportunity Cost, Gambler’s Fallacy, Optimal Stopping, Risk vs. Uncertainty
+ 4 more in the list
10Origin and sources
Christiaan Huygens gave an early systematic account of the value of uncertain wagers in De ratiociniis in ludo aleae (1657), following Pascal and Fermat’s correspondence on games of chance (1654).
- [1]Hald, A. (1990). A History of Probability and Statistics and Their Applications before 1750. John Wiley & Sons.
- [2]Blitzstein, J. K., & Hwang, J. (2019). Introduction to Probability (2nd ed.). CRC Press.
- [3]Bernoulli, D. (1954). Exposition of a New Theory on the Measurement of Risk. Econometrica, 22(1), 23–36.
Suggest an edit· Updated 2026-10-02