Decision Tree Expected Monetary Value: Method and Example

What is a decision tree and expected monetary value (EMV)

A decision tree is a diagram that lays out a decision under uncertainty as nodes, branches and payoffs. The expected monetary value (EMV) of an option is the average of its payoffs, weighted by the probability of each scenario. As a rule, you pick the option with the highest EMV.

On the shop floor it fits decisions such as buying a machine before demand is confirmed, outsourcing a peak in orders or accepting a contract with uncertain volume. The method comes from decision analysis, described by authors such as Howard Raiffa, and belongs to the classic operations research toolkit.

Decision tree visual map

The map summarizes the nodes, the probabilities, the EMV calculation and the comparison of the three options in the example.

Tree elements: nodes, branches and payoffs

• Decision node (square): the point where you choose between options. Each branch leaving it is an option under your control.
• Chance node (circle): the point where the outcome depends on chance, such as high or low demand. Each branch carries a probability, and the branches of one node must add up to 100%.
• Payoff (leaf): the net gain or loss of each path, with investment and costs already deducted.

You read the tree from left to right, but you calculate it from right to left.

EMV formula

EMV = Σ (scenario probability × scenario payoff)

For an option with two scenarios: EMV = p × Payoff_high + (1 − p) × Payoff_low. Probabilities come from sales history, firm orders on hand or, when there is no data, an estimate from the sales team, which should be recorded and reviewed.

Worked example: buy a new press or not

Fictional example. A plant received a proposal for higher volume. High demand has a 60% probability and low demand 40%. The payoffs below are net, in $ thousand, after investment and extra costs.

Option B has the highest EMV, $220k against $200k for A. Buying the press pays more if high demand materializes, but loses $100k if it does not. Option B does not lose money in either scenario.

Step by step: building your tree

1. Define the decision and the time horizon (for example, 5 years).
2. List the options, including "do nothing" as a baseline.
3. List the uncertain scenarios and assign probabilities that add up to 100%.
4. Calculate the net payoff of each path. For long horizons, use the NPV as the payoff rather than a simple sum.
5. Compute the EMV backwards: first the chance nodes, then pick the best branch at each decision node.
6. Test the sensitivity of the probabilities before deciding.

Sensitivity and the value of perfect information

If the probability of high demand is p, the EMVs become straight lines: A = −100 + 500p and B = 100 + 200p. They cross when 300p = 200, that is, at p ≈ 66.7%. Below that, B wins. Above it, A wins. With p = 70%, for example, A = 250 and B = 240.

The expected value of perfect information (EVPI) shows how much it is worth to reduce uncertainty, for instance with a firm customer order or a pilot run. With perfect information you would pick A if demand were high (400) and B if it were low (100):

EVPI = (0.6 × 400 + 0.4 × 100) − 220 = 280 − 220 = $60k

Paying more than $60k to confirm demand does not pay off. Paying less does.

When to use it and how to combine it with other tools

• Capacity investment: combine with the production capacity calculation to estimate the payoff of each scenario.
• Make or buy: see make or buy to build the costs of each branch.
• Safety or quality risks: the risk matrix ranks the hazard, and the tree computes the expected cost.
• Many uncertain variables: Monte Carlo simulation replaces discrete scenarios with distributions.
• Profit per product: the contribution margin feeds the branch payoffs.

Common mistakes and limitations

• Probabilities without a basis: guessed numbers give false precision. Always run a sensitivity analysis.
• Forgetting time: adding multi-year cash flows without discounting favors long investments.
• Ignoring risk: EMV is an average. Option A in the example can lose $100k, and the company may not survive that scenario even with a positive EMV.
• Incomplete branches: scenarios that do not add up to 100% or that overlap distort the result.
• A tree that is too big: more than three levels usually turns into an unreadable spreadsheet. Simplify or use simulation.
• Mixing it up with a fault tree: FTA investigates causes of an unwanted event; it does not choose between options.

Frequently asked questions

What is a decision tree?

A diagram of decision nodes, chance nodes and payoffs that compares options under uncertainty.

How do you calculate expected monetary value?

Multiply each scenario payoff by its probability and add them up. In the example, option B gives 0.6 × 300 + 0.4 × 100 = $220k.

Do I always pick the option with the highest EMV?

As a rule yes, if the company can absorb the worst case. If the possible loss is severe, consider risk beyond the average.

How do I set the probabilities?

Use history, firm orders and the sales team’s judgment, and record the source. Then test how far they can move before the decision changes.

What is EVPI?

The difference between the EMV with perfect information and the best current EMV. In the example, $60k, the most worth paying to reduce uncertainty.

What is the difference between a decision tree and a fault tree?

A decision tree compares options by expected payoff. A fault tree breaks down the causes of a failure.

About the author

Vagner Soares

Lean Manufacturing & Behavioral Management Specialist

Over 20 years in the automotive and metalworking industries (GM and Dana), Lean Manufacturing practitioner since 2006. SENAI instructor and mentor in Brazil’s Brasil Mais Produtivo program, delivering consulting, training and audits for 50+ companies, combining quality, productivity and people development.

LinkedInInstagramFull profile →