How to calculate expected value
Multiply each possible outcome by its probability and add the products. For a discrete outcome, expected value is the probability-weighted average of its values.
Expected value = Σ (probability × outcome).
If an outcome is worth 100 units with probability 0.5 and 0 units with probability 0.5, its expected value is 0.5 × 100 + 0.5 × 0 = 50 units. In One set of outcomes, enter the probabilities and one row of outcomes. The weighted terms are shown with the exact result.
How to calculate expected monetary value for decisions
In Compare decisions, enter one probability for each possible state and a row for every action. A state is an event that can occur after you choose. Use a set of mutually exclusive states that covers all possibilities, and use the same probabilities for every action. The worksheet supports up to eight states and six actions.
Expected monetary value, or EMV, is expected value with outcomes measured in money. Compare the expected payoffs when higher outcomes are better. Compare the expected costs when lower outcomes are better. Use a consistent currency and measurement period throughout the worksheet.
How to calculate the expected value of perfect information
Find the best outcome in each state, then weight those outcomes by the state probabilities. This is the expected result you could obtain if you knew the state before choosing. Compare it with the expected result of the best action you could choose in advance.
EVPI for payoffs = expected payoff with perfect information − highest fixed expected payoff.
EVPI for costs = lowest fixed expected cost − expected cost with perfect information.
In the starting example, two states have probability 0.5 each. Action 1 pays 100 or 0, for an expected payoff of 50. Action 2 pays 40 in either state, for an expected payoff of 40. Knowing the state would let you choose 100 in the first state and 40 in the second, for an expected payoff of 70. EVPI is 70 − 50 = 20 units.
Expected opportunity loss and the gap between choices
For each state, opportunity loss is the distance from that state's best outcome. The expected opportunity loss is the weighted average of those distances. In the example it is 20 for Action 1 and 30 for Action 2. The gap from the best fixed choice is 0 for Action 1 and 10 for Action 2. The expected opportunity loss of a best fixed choice equals EVPI.
Check the assumptions before interpreting a result
Probabilities must add up to exactly 1, or 100%. If rounded entries such as three copies of 33.33% total 99.99%, enter 1/3 in fraction view for each state. Changing the display between fractions and percentages preserves the original probability. A zero-probability state contributes zero to the weighted total.
Expected value describes a weighted average under the inputs you provide. For a single decision, a possible outcome can differ substantially from its average. Consider the range of outcomes and how you estimated their probabilities. EVPI assumes the state is known in time to act. Forecasts or partial information require a different calculation.
See the UC Berkeley explanation of value of perfect information and the Dartmouth decision analysis examples. For probability rules, use the Probability Calculator.