Expected value (EV) is the probability-weighted average of all possible outcomes of a decision: multiply each outcome by its probability, sum the results. EV thinking helps clarify when risk aversion is costing you real value and when caution is genuinely justified. It is most reliable when decisions repeat over time and when probabilities can be estimated from evidence. It breaks down when outcomes are catastrophic or when probabilities are unknowable.
Where this came from
Expected value has roots in 17th-century probability theory. Blaise Pascal and Pierre de Fermat developed the foundational mathematics in correspondence about gambling problems in 1654. Daniel Bernoulli formalised the concept of expected utility in 1738, distinguishing between the mathematical expected value of an outcome and its subjective utility to the decision-maker, a distinction that anticipates much of modern behavioural economics.
The concept became central to economics, statistics, insurance, and eventually to professional decision-making in fields from military strategy to poker. John von Neumann and Oskar Morgenstern's expected utility theory (1944) provided the formal basis for rational decision-making under uncertainty and remained the dominant framework in economics for decades.
In practice, EV thinking is most visibly applied by professional poker players, quantitative investors, and military planners. Poker player and decision strategist Annie Duke, in her book "Thinking in Bets," popularised the application of EV thinking to everyday decisions, arguing that the quality of a decision should be judged by the quality of the reasoning at the time it was made, not by whether it happened to produce a good outcome. EV thinking and outcome bias are directly in tension: EV thinking is the antidote to judging decisions by their results.
How it works
The calculation is straightforward. For each option you are considering, list the possible outcomes. Assign a probability to each outcome (the probabilities for each option must sum to 1.0). Assign a value to each outcome. Multiply each outcome's value by its probability. Sum the results across all outcomes for that option. The option with the highest expected value is, in theory, the one to choose.
A simple example: you are offered a guaranteed payment of Rs 80,000 or a 50% chance of Rs 2,00,000. The EV of the guaranteed payment is Rs 80,000. The EV of the risky option is 0.5 x Rs 2,00,000 = Rs 1,00,000. The risky option has a higher EV. Whether you should take it depends on your risk tolerance and your financial situation, but EV thinking at least makes the trade-off explicit: you are paying Rs 20,000 in expected value to have the certainty of the guaranteed payment.
The more important application is not to numerical bets but to decisions where probabilities and values must be estimated. When you are deciding whether to leave a stable job to start a company, EV thinking asks you to estimate, explicitly: what is the probability the company succeeds at each level of success? What would that be worth to you in financial terms, in learning, in autonomy? What is the probability of failure, and what is the cost of that failure in financial terms and opportunity cost? Most people make this type of decision on intuition. EV thinking forces the underlying assumptions into the open, where they can be examined and challenged.
EV thinking is used systematically by quantitative investors. Charlie Munger, Warren Buffett's partner at Berkshire Hathaway, has described their investment process in terms that are explicitly EV-based: they are looking for situations where the probability of loss is low and the potential gain is large, producing a high expected value. The discipline is not just to find good bets but to avoid negative-EV bets that feel safe because they are familiar.
When to use it and when it fails
Expected value thinking is most valuable for repeated decisions where the law of large numbers operates over time. An investor making 50 decisions per year benefits from EV thinking even if any single decision is unpredictable, because EV optimisation will compound favourably across the portfolio. The same logic applies to a sales team choosing how to allocate prospecting time, or a product team deciding which features to build.
EV thinking is less reliable for unique, one-time decisions with catastrophic downside potential. If a 5% probability of ruin is involved, no amount of positive EV on the other 95% of outcomes justifies taking the risk, because ruin eliminates the possibility of future decisions. This is the Kelly criterion problem: bet sizing must be calibrated not just to EV but to the variance of outcomes and the size of your bankroll. Never bet the enterprise on a single high-EV play.
EV also breaks down when probabilities cannot be estimated. In genuinely novel situations with no historical reference, probability estimates are more like expressions of confidence than reliable statistics. In those cases, other frameworks, scenario planning, pre-mortem analysis, or the Regret Minimization Framework, are better suited to the uncertainty.
Probability Neglect
People treat small probabilities as either zero (ignoring them entirely) or near-certain (fearing them disproportionately). This is why lottery tickets sell and why people over-insure against remote risks while under-preparing for likely ones. EV thinking requires assigning realistic probabilities, which means neither dismissing a 5% chance as negligible nor treating it as almost certain. Probability neglect distorts the EV calculation at both ends: it leads people to avoid high-EV options with small failure probabilities and to pursue low-EV options with vivid but unlikely upside scenarios.
Run this framework on your actual decision.
DecisionsMatter.ai walks you through a structured 5-step analysis. Your first analysis is free.
How to apply it in practice
Start by listing the options and the possible outcomes for each option. Be specific about outcomes. "Things go well" is not an outcome; "revenue reaches Rs 50 lakh by month 18" is an outcome. For each outcome, assign a probability. If you have data, use it. If you are estimating, use a reference class: what is the base rate for this category of outcome in this context?
Assign a value to each outcome. For financial decisions this can be monetary. For career decisions, use a 1-10 scale for life satisfaction, career advancement, financial security, and other dimensions you care about, then weight them. The goal is not false precision but to force explicit trade-offs onto paper where they can be examined.
Calculate the EV for each option. Then ask: given this EV comparison, am I choosing the option I am leaning toward for sound reasons, or am I choosing it because it feels safe? If the lower-EV option is your preference, name the reason. Risk aversion is legitimate when the downside is catastrophic or when you are making a one-time bet. Risk aversion is irrational when the downside is manageable and the situation will recur.
Finally, record the decision and your probability estimates. Reviewing these after the outcome is the only way to improve your calibration over time. The record tells you whether your probability estimates were realistic, which is information that makes your next EV calculation more reliable.
This is one model from the upcoming Decisions Matter book.
30 mental models. 40 cognitive biases. A 5-step decision system. Get the relevant chapter in your inbox when it publishes.
Frequently asked questions
What is expected value?
Expected value (EV) is the probability-weighted average of all possible outcomes of a decision. You calculate it by multiplying each possible outcome by its probability of occurring and summing the results across all outcomes. For example, a 40% chance of gaining Rs 1,00,000 and a 60% chance of gaining nothing has an expected value of Rs 40,000. Expected value thinking asks you to make the decision that maximises EV across repeated decisions, rather than the one that feels safest in any single instance.
How do you calculate expected value for a real decision?
List the possible outcomes, assign a probability to each (probabilities must sum to 1.0), assign a value to each outcome, multiply each outcome's value by its probability, and sum the results. For a career decision, you might estimate a 30% chance the new role accelerates your career significantly (value: high), a 50% chance it is lateral (value: neutral), and a 20% chance it is a setback (value: negative). You cannot always quantify precisely, but even a rough EV comparison forces you to make your probability estimates explicit, which improves the quality of the decision.
How can expected value thinking help with career decisions?
Career decisions often involve asymmetric upside and downside that intuition handles poorly. Expected value thinking helps by forcing you to estimate probabilities and values rather than relying on gut feel. A job change that feels risky might have an EV significantly higher than staying, once you assign realistic probabilities to the upside and downside scenarios. Conversely, a seemingly attractive opportunity might have a low EV when the probability of success is realistically low. EV thinking also highlights optionality: choices that keep future options open have value beyond their direct payoffs.
When does expected value thinking fail?
Expected value thinking fails in three situations. First, when you cannot estimate probabilities reliably, such as in genuinely novel situations with no historical base rate. Second, when the stakes are so high that a single bad outcome is catastrophic regardless of EV (a 1% chance of ruin should be avoided even if the EV is positive). Third, when you are making a one-time decision rather than a repeated one: EV is most valid as a guide when you will make many similar decisions over time, because averages only emerge over repeated trials. For unique, irreversible, high-stakes decisions, EV should inform but not determine the choice.