decision-tree-analysis

Maps a decision and the uncertain events that follow it into a tree, then solves it for the sequence of choices that maximizes expected value. Catches the failure where options are weighed by gut feel: the tree forces which moves are yours, which are the world's, and what every outcome is worth into one diagram, and folding back turns that into a defensible recommendation plus a price on resolving the uncertainty.

The node types, the fold-back arithmetic, break-even and sensitivity analysis, expected value of perfect information, and the risk-attitude rule live in references/decision-tree-analysis.md. Load that file before working and follow the fold-back and probabilities-sum-to-one rules exactly.

When to use

The user faces a choice whose outcome depends on uncertain events, and wants the option that maximizes expected value over a horizon. Decision tree analysis fits when there is genuine uncertainty with probabilities you can estimate, and especially when a decision now is followed by later decisions once events resolve (launch now, test first, or shelve).

• If there is no real uncertainty and the choice is between options scored on fixed criteria, it is a deterministic ranking, so use weighted-decision-matrix instead.
• If the hard part is estimating the probability of one event from how similar cases turned out, use reference-class-forecasting, then feed its number into a chance node here.
• If the future has too many unknowns to pin to probabilities and you need to plan against distinct futures instead, use scenario-planning.
• If you have settled on a plan and want to surface how it could fail before committing, use pre-mortem.

Language

Conduct the session and write the report in the language the user is writing in. The reference file is English; keep the framework keyed to it regardless of output language.