Play chips onlyNo accountWorks with no API key

Preflop odds on fame.

You are dealt one real person and one year to have backed them from. Put your whole stack behind that decade, or pass. Then a solver that already knows how every decade turned out tells you exactly what the right answer was — and precisely what yours cost.

The surprise is the lesson

A card running 24% is a fold in every poker game you have played. Here it is often a shove, and the reason is worth the two minutes: the cards that call you are not an average slice of the deck, they are the strongest fifth of it. So the price you have to clear is far lower than the one poker taught you, and the number was on screen before you clicked.

Nothing here is a guess

The chart you are scored against was solved offline and certified: two independent best-response calculations bracket the value of the game, so the exploitability is a proof rather than a claim that a solver converged.

Perfect hindsight is worth nothing

You might reasonably ask why replaying settled history is not trivially winnable. Because you are priced against the whole range of cards the other seat could hold, and your score is written before the stretch is drawn. Someone who had memorised every trajectory would tie the chart and could never beat it. That is not a limitation of the game — it is the proof that the game is a real one.

What is measured, and what is not

The people are real. The money is not measured — it is modelled, from a small hand-built roster, and every figure in it was entered from memory rather than read off a source. So nothing here is a claim about anybody’s finances, and the ranks you are shown are positions among 24 people we chose. A different roster is a different game, and the method page says so at length rather than in a footnote.

Read the method