Regret Minimization Scorer
When the odds are unknowable, aim to avoid kicking yourself later. Enter each option’s payoff in every scenario and find the choice with the smallest worst-case regret.
Worst-case regret
| Option | Max regret |
|---|---|
| Aggressive ★ | 60 |
| Balanced | 80 |
| Safe bet | 160 |
Minimax-regret pick: Aggressive — the option you're least likely to regret, whatever happens.
The logic of least regret
Some decisions come without a probability sheet: you can picture the scenarios — a recession, a steady year, a boom — but you genuinely can’t say how likely each is. The minimax-regret criterion, introduced by the statistician Leonard Savage, offers a principled way through. Instead of chasing the best average, it asks which option keeps your worst possible regret as small as possible.
For every scenario the tool spots the best result on offer, then scores how far each option trails it. The option whose largest shortfall is smallest is the defensive pick — the one that never leaves you too far behind whatever the future does. Jeff Bezos used a related idea, the “regret minimization framework,” to decide whether to start a company: he pictured himself at 80 and chose the path he’d regret least.
Frequently Asked Questions
What is regret in decision theory?
Regret is the gap between the payoff you actually got and the best payoff you could have got in that scenario, had you chosen differently. If a downturn hits and your option earns 40 while the best option would have earned 100, your regret is 60. Minimising regret is about avoiding the sinking feeling of 'I should have picked the other one.'
How does minimax regret work?
For each scenario the tool finds the best payoff any option achieves, then measures how far each option falls short — that's the regret. Each option's worst regret across all scenarios is its maximum regret, and the recommended choice is the one with the smallest maximum regret. You're picking the option you're least likely to seriously regret, no matter which future arrives.
When should I use this instead of expected value?
Use regret minimization when you genuinely can't assign probabilities to the scenarios — deep uncertainty rather than measurable risk. If you do have credible odds, expected value uses more of that information. Regret minimization is the tool for 'I don't know how likely each future is, but I want to avoid a decision that could look foolish in hindsight.'
Is minimizing regret always the right goal?
No. Minimax regret is conservative and can steer you away from bold options with a big upside, because it focuses on avoiding shortfalls rather than chasing gains. Compare its recommendation with an optimistic (best-case) and a pessimistic (worst-payoff) view before committing, and weigh how much a large regret would actually cost you.
A decision aid, not a decision. Minimax regret is one lens on uncertainty; compare it with best- and worst-case views and your own judgment before committing.