FINITE MATHEMATICS • DECISION ANALYSIS

Expected Value & Opportunity Loss — Compute expected value and expected opportunity loss

Quantify rational choices under uncertainty by weighting outcomes with probabilities and measuring the cost of imperfect decisions.

Historical Context & Motivation

The notion of systematically weighing uncertain outcomes stretches back to the correspondence between Blaise Pascal and Pierre de Fermat in 1654, where the two mathematicians wrestled with fair divisions of stakes in interrupted games of chance. Their exchange gave rise to the concept of expected value — the probability-weighted average of all possible payoffs — which became the cornerstone of rational decision-making under uncertainty. Over the next three centuries, economists, statisticians, and operations researchers refined this idea, eventually formalizing it within decision analysis, a discipline that pairs expected value with complementary measures such as expected opportunity loss (EOL) to guide choices when the state of nature is unknown.

1654
Pascal–Fermat Correspondence
Blaise Pascal and Pierre de Fermat develop the foundational idea of mathematical expectation while solving the problem of points — how to fairly divide stakes in an unfinished game of chance.
1738
Bernoulli's Utility Theory
Daniel Bernoulli publishes his St. Petersburg paradox paper, arguing that people maximize expected utility rather than expected monetary value, introducing risk aversion into the framework.
1951
Savage's Minimax Regret
Leonard Savage formalizes the concept of regret (opportunity loss) in decision theory, proposing the minimax regret criterion as an alternative to expected value maximization.
1961
Raiffa & Schlaifer's Decision Analysis
Howard Raiffa and Robert Schlaifer publish Applied Statistical Decision Theory, establishing payoff tables, opportunity loss tables, expected value of perfect information (EVPI), and the relationship EOL* = EVPI as standard tools.

The central question decision analysis addresses is deceptively simple: given several possible actions and several possible states of nature — each with an estimated probability — which action should a rational decision-maker choose? Expected value provides one lens: pick the action with the highest (or lowest, for costs) probability-weighted payoff. Expected opportunity loss provides another, equally powerful lens: pick the action whose average regret, measured against what you could have done if you knew the future, is smallest. As we will demonstrate, these two criteria always agree on the optimal action — a remarkable and elegant equivalence.

Core Principles & Definitions

Before diving into calculations, it is essential to establish the structural vocabulary of decision analysis. Every problem begins with a payoff table (also called a decision matrix), which enumerates every combination of decision alternatives and states of nature. Each cell of the table records the payoff (profit, revenue, or cost) that results if a particular action is taken and a particular state occurs. An opportunity loss table is then derived from the payoff table by computing, column by column, the difference between the best possible payoff in that column and the payoff of each action — this difference is the regret incurred by not having chosen optimally for that particular state.

1

Decision Alternative (aᵢ)

A course of action available to the decision-maker. Examples: launch product A, invest in fund B, or do nothing. The set of all alternatives is denoted {a₁, a₂, …, aₘ}.
2

State of Nature (sⱼ)

An external condition or scenario beyond the decision-maker's control. Examples: strong demand, weak demand, recession. Each state sⱼ has an associated prior probability P(sⱼ), and all probabilities sum to 1.
3

Payoff (Vᵢⱼ)

The quantitative outcome — usually profit or cost — that results when alternative aᵢ is chosen and state sⱼ occurs. Higher payoffs are preferred when measuring profit; lower payoffs are preferred for costs.
4

Opportunity Loss (Lᵢⱼ)

The regret from not having selected the best alternative for state sⱼ. Computed as Lᵢⱼ = max(Vⱼ) − Vᵢⱼ for each column j. The opportunity loss of the optimal action in any column is always 0.
5

Expected Value of Perfect Information (EVPI)

The maximum price a decision-maker should pay for a perfect forecast. EVPI equals the expected value under certainty (EV|CI) minus the best expected value (EV*), and it always equals the minimum EOL: EVPI = EOL*.
KEY TAKEAWAY
Think of opportunity loss as the feeling you get when, after choosing a restaurant and eating a mediocre meal, you discover there was a far better restaurant next door. You cannot change the past, but the size of your regret depends on how much better the alternative was. Decision analysis quantifies that regret for every possible future, weights each by its probability, and picks the action that minimizes your average regret — the expected opportunity loss.

Visual Explanation — Payoff Table to EOL

The diagram below illustrates the complete decision-analysis pipeline: we start with a payoff table on the left, derive the opportunity loss table in the center, and compute expected values and expected opportunity losses on the right. Follow the arrows to trace how each number is generated. Note that the column maxima in the payoff table become the reference points from which all opportunity losses are measured.

The pipeline from a payoff table (left) to the opportunity loss table (center) and the final expected values and EOLs (right). The golden box at the bottom verifies the fundamental relationship EVPI = EOL*.

In the diagram above, notice how the opportunity loss for each cell is computed by subtracting the cell's payoff from the column maximum. For alternative a₁ under state s₂, the opportunity loss is 60 − 30 = 30 because, had the decision-maker known s₂ would occur, she would have chosen a₂ (earning 60 instead of 30). The expected opportunity loss for each alternative is then the probability-weighted sum of its row in the opportunity loss table. The alternative with the smallest EOL is optimal — and it is always the same alternative that has the largest expected value.

Mathematical Framework

Let there be m decision alternatives a₁, a₂, …, aₘ and n mutually exclusive, collectively exhaustive states of nature s₁, s₂, …, sₙ with prior probabilities P(s₁), P(s₂), …, P(sₙ) that sum to 1. Let Vᵢⱼ denote the payoff when alternative aᵢ is chosen and state sⱼ occurs. The following formulas constitute the mathematical backbone of expected-value decision analysis.

EXPECTED VALUE (EV)
EV(aᵢ) = Σⱼ₌₁ⁿ P(sⱼ) × Vᵢⱼ
EV(aᵢ) is the weighted average payoff of alternative aᵢ across all states of nature. The optimal alternative under the EV criterion is a* = argmax EV(aᵢ).
OPPORTUNITY LOSS
Lᵢⱼ = max₍ₖ₎ Vₖⱼ − Vᵢⱼ
For each state sⱼ, find the maximum payoff across all alternatives (max₍ₖ₎ Vₖⱼ), then subtract the payoff of alternative aᵢ. The result Lᵢⱼ ≥ 0 measures the regret of not having chosen optimally for that particular state.
EXPECTED OPPORTUNITY LOSS (EOL)
EOL(aᵢ) = Σⱼ₌₁ⁿ P(sⱼ) × Lᵢⱼ
The probability-weighted average of the regrets for alternative aᵢ. The optimal alternative under the EOL criterion is a* = argmin EOL(aᵢ). This always coincides with argmax EV(aᵢ).
EXPECTED VALUE UNDER CERTAINTY & EVPI
EV|CI = Σⱼ₌₁ⁿ P(sⱼ) × max₍ₖ₎ Vₖⱼ ; EVPI = EV|CI − EV* = EOL*
EV|CI imagines a clairvoyant who always picks the best alternative for each state. EVPI is the gap between this ideal payoff and the best achievable expected value. The equality EVPI = EOL* provides a consistency check: the minimum expected opportunity loss always equals the expected value of perfect information.
💡 Why EV and EOL Always Agree
One can prove the equivalence algebraically. Note that EOL(aᵢ) = Σ P(sⱼ) × [max Vₖⱼ − Vᵢⱼ] = Σ P(sⱼ) × max Vₖⱼ − Σ P(sⱼ) × Vᵢⱼ = EV|CI − EV(aᵢ). Since EV|CI is a constant (it does not depend on i), minimizing EOL(aᵢ) is equivalent to maximizing EV(aᵢ). The minimum EOL therefore equals EV|CI − max EV(aᵢ) = EVPI.

Detailed Breakdown — From Payoff to Decision

To solidify the procedure, consider a three-alternative, three-state-of-nature scenario. A manufacturer must decide whether to build a small plant (a₁), a medium plant (a₂), or a large plant (a₃). The states of nature are low demand (s₁), moderate demand (s₂), and high demand (s₃) with prior probabilities 0.2, 0.5, and 0.3 respectively. Payoffs are net profits in millions of dollars.

Payoff table — net profits in $M
Alternatives₁ Low (0.2)s₂ Moderate (0.5)s₃ High (0.3)
a₁ Small plant101010
a₂ Medium plant−21515
a₃ Large plant−10825
Column max101525
Bar chart comparing the expected values of all three alternatives (left cluster) alongside their expected opportunity losses (right cluster). Alternative a₂ (medium plant) achieves the highest EV and the lowest EOL simultaneously, confirming the EV–EOL equivalence.

The opportunity loss table derived from the payoff table above is shown next. Each cell is obtained by subtracting the alternative's payoff from the column maximum for that state.

Opportunity loss table with EOL computation
Alternatives₁ Low (0.2)s₂ Moderate (0.5)s₃ High (0.3)EOL
a₁ Small10 − 10 = 015 − 10 = 525 − 10 = 150(0.2)+5(0.5)+15(0.3) = 7.0
a₂ Medium10 − (−2) = 1215 − 15 = 025 − 15 = 1012(0.2)+0(0.5)+10(0.3) = 5.4
a₃ Large10 − (−10) = 2015 − 8 = 725 − 25 = 020(0.2)+7(0.5)+0(0.3) = 7.5

Verification: EV|CI = 0.2(10) + 0.5(15) + 0.3(25) = 2 + 7.5 + 7.5 = 17.0. The best EV is EV(a₂) = 0.2(−2) + 0.5(15) + 0.3(15) = −0.4 + 7.5 + 4.5 = 11.6. Wait — let us recalculate carefully: EV(a₁) = 0.2(10) + 0.5(10) + 0.3(10) = 10.0; EV(a₂) = 0.2(−2) + 0.5(15) + 0.3(15) = −0.4 + 7.5 + 4.5 = 11.6; EV(a₃) = 0.2(−10) + 0.5(8) + 0.3(25) = −2 + 4 + 7.5 = 9.5. So EV* = 11.6, and EVPI = 17.0 − 11.6 = 5.4 = EOL(a₂). The identity EVPI = EOL* holds perfectly.

Worked Example — Investment Portfolio Decision

An investor must allocate funds among three portfolios: Conservative (a₁), Balanced (a₂), and Aggressive (a₃). An economist forecasts three possible economic conditions: recession (s₁, probability 0.3), stability (s₂, probability 0.5), and boom (s₃, probability 0.2). Annual returns (in thousands of dollars) are given below.

Annual returns in $000s
PortfolioRecession (0.3)Stability (0.5)Boom (0.2)
a₁ Conservative304035
a₂ Balanced−105565
a₃ Aggressive−3040100
Full Decision Analysis
1
Step 1 — Compute Expected ValuesEV(a₁) = 0.3(30) + 0.5(40) + 0.2(35) = 9 + 20 + 7 = 36. EV(a₂) = 0.3(−10) + 0.5(55) + 0.2(65) = −3 + 27.5 + 13 = 37.5. EV(a₃) = 0.3(−30) + 0.5(40) + 0.2(100) = −9 + 20 + 20 = 31.
EV* = EV(a₂) = 37.5 ($000s) — Balanced portfolio is optimal by EV criterion.
2
Step 2 — Identify Column MaximaFor s₁ (recession): max{30, −10, −30} = 30. For s₂ (stability): max{40, 55, 40} = 55. For s₃ (boom): max{35, 65, 100} = 100.
Column maxima: 30, 55, 100
3
Step 3 — Build Opportunity Loss Tablea₁: L₁₁ = 30 − 30 = 0, L₁₂ = 55 − 40 = 15, L₁₃ = 100 − 35 = 65. a₂: L₂₁ = 30 − (−10) = 40, L₂₂ = 55 − 55 = 0, L₂₃ = 100 − 65 = 35. a₃: L₃₁ = 30 − (−30) = 60, L₃₂ = 55 − 40 = 15, L₃₃ = 100 − 100 = 0.
4
Step 4 — Compute Expected Opportunity LossesEOL(a₁) = 0.3(0) + 0.5(15) + 0.2(65) = 0 + 7.5 + 13 = 20.5. EOL(a₂) = 0.3(40) + 0.5(0) + 0.2(35) = 12 + 0 + 7 = 19. EOL(a₃) = 0.3(60) + 0.5(15) + 0.2(0) = 18 + 7.5 + 0 = 25.5.
EOL* = EOL(a₂) = 19 ($000s) — Balanced portfolio has the minimum regret, confirming the EV result.
5
Step 5 — Verify with EVPIEV|CI = 0.3(30) + 0.5(55) + 0.2(100) = 9 + 27.5 + 20 = 56.5. EVPI = EV|CI − EV* = 56.5 − 37.5 = 19. This equals EOL* = 19. ✓ The investor should pay at most $19,000 for a perfect economic forecast.
EVPI = EOL* = 19 ($000s) ✓

Strengths, Limitations & Decision Criteria Compared

The EV and EOL criteria are powerful, but they are not the only frameworks for decision-making under uncertainty. Other classical criteria — such as maximax (optimistic), maximin (conservative), and minimax regret — do not require probability estimates at all. Understanding when each criterion is appropriate depends on the information available and the decision-maker's attitude toward risk.

Comparison of classical decision criteria
CriterionRequires Probabilities?StrengthsLimitations
Expected Value (EV)YesUses all available information; mathematically elegant; provides a single optimal actionSensitive to probability estimates; ignores variance/risk; may recommend actions with catastrophic worst cases
Expected Opportunity Loss (EOL)YesAlways agrees with EV; directly yields EVPI; intuitive 'regret' framingSame sensitivity to probabilities as EV; requires constructing an additional table
MaximaxNoSimple; captures optimistic outlookIgnores all but the best outcome; highly risky
MaximinNoConservative; protects against worst caseIgnores all but the worst outcome; may sacrifice large gains for marginal protection
Minimax RegretNoUses opportunity loss concept; no probabilities neededOnly considers worst-case regret; sensitive to addition/removal of alternatives
KEY TAKEAWAY
EV and EOL are the workhorse criteria of decision analysis under risk — situations where probabilities are known or can be reasonably estimated. When no probability information is available (decision-making under uncertainty in the strict sense), criteria like maximin and minimax regret become appropriate. In engineering reliability or public safety contexts, where catastrophic outcomes must be avoided, maximin or expected-utility approaches with risk aversion may override a pure EV recommendation.

Connections to Advanced Theory

The expected value and expected opportunity loss framework introduced in this lesson serves as the entry point to several deeper theoretical constructs. In a Bayesian decision-analysis course, you will learn how to update prior probabilities with new data (e.g., market research, diagnostic tests) using Bayes' theorem, which leads to revised expected values and a quantity called the expected value of sample information (EVSI). In expected utility theory, the linear payoff function is replaced by a concave utility function to model risk aversion, and the decision criterion becomes maximization of expected utility rather than expected monetary value. In multi-stage settings, decision trees and influence diagrams generalize the single-stage payoff table into a sequential framework.

From single-stage EV/EOL to advanced decision analysis
This Lesson (Single-Stage, EV/EOL)Advanced Extensions
Prior probabilities P(sⱼ) are fixed and givenBayesian updating: posterior probabilities P(sⱼ | data) computed via Bayes' theorem
Payoffs Vᵢⱼ are monetary valuesExpected utility theory: payoffs mapped through a utility function U(Vᵢⱼ) to capture risk preferences
EVPI = maximum value of a perfect forecastEVSI = value of an imperfect forecast; always satisfies 0 ≤ EVSI ≤ EVPI
Single decision point (payoff table)Sequential decisions modeled by decision trees with multiple chance and decision nodes
Finite, discrete states of natureContinuous probability distributions over states; integrals replace summations

Despite its simplicity, the EV/EOL framework provides the conceptual scaffolding for all of these extensions. The identity EVPI = EOL*, for example, generalizes naturally: in the Bayesian framework, EVSI can be computed as the reduction in expected opportunity loss achieved by observing sample data before making the final decision. Mastering the mechanics of payoff tables, opportunity loss tables, and expected-value computations is therefore essential preparation for the more sophisticated tools you will encounter in operations research and management science.

Practice Problems

PROBLEM 1CONCEPTUAL
Explain in your own words why the alternative that maximizes expected value is always the same alternative that minimizes expected opportunity loss. Your explanation should reference the relationship EOL(aᵢ) = EV|CI − EV(aᵢ).
PROBLEM 2BASIC CALCULATION
A company chooses between two strategies under two states of nature. Payoffs (in $000s): V₁₁ = 80, V₁₂ = 20; V₂₁ = 50, V₂₂ = 60. P(s₁) = 0.45, P(s₂) = 0.55. Compute EV and EOL for each alternative, identify the optimal action, and verify EVPI = EOL*.
PROBLEM 3INTERMEDIATE
A retailer must decide how many units to stock: 100 (a₁), 200 (a₂), or 300 (a₃). Demand is low (s₁, P = 0.25), medium (s₂, P = 0.40), or high (s₃, P = 0.35). Profit table (in $000s): a₁ row: {15, 15, 15}; a₂ row: {−5, 30, 30}; a₃ row: {−20, 10, 50}. Determine the optimal stocking level using both EV and EOL. What is the EVPI?
PROBLEM 4APPLIED
A hospital administrator must decide whether to lease MRI equipment (a₁), purchase MRI equipment (a₂), or outsource MRI services (a₃). Annual net benefits (in $000s) depend on patient volume: low (s₁, P = 0.20), moderate (s₂, P = 0.50), high (s₃, P = 0.30). Payoffs: a₁ = {100, 150, 160}; a₂ = {−50, 200, 350}; a₃ = {120, 120, 120}. (a) Determine the optimal decision. (b) Interpret the EVPI in the context of the hospital's decision.
PROBLEM 5CRITICAL THINKING
Suppose you have a payoff table with m alternatives and n states of nature. Prove that adding a dominated alternative (one whose payoff is less than or equal to some other alternative's payoff in every state) does not change the optimal EV or EOL decision, nor does it change the EVPI. Then discuss: can adding a non-dominated alternative change the EVPI? Provide a concrete 2 × 2 example to support your argument.

Lesson Summary

In decision analysis under risk, the expected value (EV) of an alternative is the probability-weighted sum of its payoffs across all states of nature: EV(aᵢ) = Σ P(sⱼ) × Vᵢⱼ. The rational decision-maker selects the alternative with the maximum EV (for profits) or minimum EV (for costs). The opportunity loss Lᵢⱼ measures the regret of choosing aᵢ when state sⱼ occurs, computed as the difference between the best possible payoff in that column and the actual payoff: Lᵢⱼ = max Vₖⱼ − Vᵢⱼ.

The expected opportunity loss (EOL) is the probability-weighted average regret. The alternative that minimizes EOL is always identical to the one that maximizes EV, because EOL(aᵢ) = EV|CI − EV(aᵢ), where EV|CI is a constant. This relationship also establishes that the expected value of perfect information (EVPI) equals the minimum EOL, providing both a consistency check and a practical ceiling on what a decision-maker should pay for additional information. These tools — payoff tables, opportunity loss tables, EV, EOL, and EVPI — form the foundation of quantitative decision-making and extend naturally into Bayesian analysis, expected utility theory, and sequential decision trees.

Varsity Tutors • Finite Mathematics • Expected Value & Opportunity Loss