Expected Value and Tradeoffs

If you face a choice between two options and you know both the possible outcomes and their probabilities, there is a mathematically precise way to compare them: expected value. It is one of the most powerful tools in rational decision-making, and one of the most frequently misapplied. This lesson covers what expected value actually is, how to compute it, when it is the right tool, and — crucially — what it cannot do.

What Is Expected Value?

The expected value of an option is the probability-weighted average of all its possible outcomes. Formally: EV(option) = sum over all outcomes of [P(outcome) x value(outcome)] A simple example: a game costs $5 to play. You flip a fair coin. Heads: you win $12. Tails: you win nothing. Should you play? EV = P(heads) x $12 + P(tails) x $0 = 0.5 x 12 + 0.5 x 0 = 6 + 0 = $6 The expected value of playing is $6. The cost is $5. Since EV > cost, playing has positive expected value — in the long run, this is a profitable game. Over many repetitions, you expect to net $1 per play. Notice the word 'expected.' Expected value is not what will happen in any single trial. On any given play, you either win $12 or $0 — you never receive exactly $6. Expected value describes the long-run average, not the single-instance outcome. This matters enormously for how you apply it.

Expected value calculations become more complex — and more realistic — when outcomes are not binary. Example: you are deciding whether to start preparing now for an optional competition. Three scenarios: Scenario A: you prepare, compete, and place in the top three. Probability: 15%. Value: +100 points (career boost, prize money). Scenario B: you prepare, compete, but do not place. Probability: 55%. Value: -10 points (time cost, mild disappointment). Scenario C: you prepare but withdraw due to illness. Probability: 10%. Value: -15 points (time lost). Scenario D: you skip and redirect time to other work. Probability: 20% weight — this is the counterfactual. Value: +5 points (moderate productivity gain). EV(compete) = 0.15(100) + 0.55(-10) + 0.10(-15) = 15 - 5.5 - 1.5 = 8 points EV(skip) = 1.0(5) = 5 points On expected value, competing wins. But notice how sensitive this is to the 15% success probability — if you reassess it as 5%, EV(compete) = 5 - 5.5 - 1.5 = -2. The decision flips. Your probability estimates are load-bearing, and should be scrutinized carefully.

Tradeoffs: When Values Conflict

Expected value requires you to assign a single numerical value to every outcome. In practice, outcomes have multiple dimensions — financial, social, ethical, emotional — and these dimensions do not always convert cleanly into one number. A tradeoff is a situation where improving along one dimension necessarily worsens another. Taking a challenging internship might increase your skills and career prospects but cost you time with family and friends. There is no position that is better on both dimensions simultaneously; you must sacrifice something. Facing tradeoffs honestly requires identifying your values and their relative weights. Most people resist this because explicit prioritization feels harsh — it forces you to say, in effect, 'I value this more than that.' But implicit prioritization happens anyway, in your actual choices. Making it explicit just means the prioritization is deliberate and examined rather than accidental. Multi-attribute utility theory offers a framework: decompose outcomes into their dimensions, rate each option on each dimension, multiply by dimension weights, and sum. This does not eliminate the judgment about weights — it just surfaces it so it can be inspected and debated.

Where Expected Value Has Limits

Expected value is not a complete theory of rational decision-making. Three important limitations: First: diminishing marginal utility. $1,000 is worth more to a person with $500 than to a person with $500,000. The expected monetary value of a bet may be positive, but the expected utility — the value of the money to you — may not be. This is why rational people decline actuarially fair gambles: the pain of losing $100 is often larger than the pleasure of gaining $100. Expected utility theory accounts for this by replacing dollar values with utility values that reflect actual preferences. Second: catastrophic or irreversible outcomes. Even if the expected value of an action is positive, an outcome that is catastrophic and irreversible may make the action unwise. Imagine a surgery with a 90% chance of full cure and a 10% chance of death. If you value your continued existence far more than you value any monetary equivalent, a naive EV calculation misses the point. For decisions involving rare, catastrophic outcomes, expected value should be supplemented with explicit attention to tail risks. Third: distributional fairness. Expected value aggregates outcomes into a single number, erasing who bears the costs and who receives the benefits. A policy that produces $1 million in expected value to one person while imposing $10,000 costs on 100 people may have high EV — but it raises fairness questions that EV alone cannot answer.