Regret is the difference between an option result and the best result available in the same scenario. Minimax regret favors the option with the smaller largest gap when scenario consequences are clearer than their probabilities.
Higher values must be better. Use one unit and the same scenario meaning for both options.
| Shared future scenario | Option A value | Option B value |
|---|---|---|
The result will calculate the avoidable gap for both options in every scenario, then compare maximum and average regret.
Within each scenario, the better entered result becomes the benchmark. Regret equals that benchmark minus the option's result. The scenario winner has zero regret, while the other option shows the value it leaves behind.
Minimax regret compares each option's largest regret and favors the smaller one. It is useful when the scenarios are meaningful but their probabilities are too weak for an expected-value calculation. It does not predict which future will occur, and it does not measure emotional disappointment.
A larger facility returns 70 under low demand, 110 under expected demand, and 160 under high demand. A smaller facility returns 90, 105, and 120. The large facility's regrets are 20, zero, and zero. The small facility's regrets are zero, 5, and 40. Minimax regret favors the larger facility because its worst avoidable gap is 20 rather than 40. If a value below 80 would be unaffordable, however, the large facility fails the downside requirement before regret is considered.
If scenario probabilities become defensible, compare the result with the expected value calculator. If downside boundaries matter more than opportunity gaps, use the decision risk range calculator.
Learn what regret means mathematically, when minimax regret fits, and why hard constraints still come first.
For each scenario, regret is the best entered result minus the result for the option being evaluated. The better option in a scenario has zero regret. This is an opportunity gap, not an estimate of future emotion.
Learn the regret-minimization frameworkIt is useful when you can describe plausible scenarios and consequences but cannot defend precise probabilities. It favors protection from the largest avoidable gap rather than the highest average.
Compare robust and optimal decisionsNo. Remove an option that fails a genuine requirement before comparing regret. A favorable result in another scenario cannot repair an unacceptable safety, legal, affordability, or compatibility failure.
Separate constraints from preferences