Cover of The Perfect Bet

The Perfect Bet

Adam Kucharski

6 ideas

  1. Massachusetts Cash WinFall roll-down exploit

    When the Cash WinFall jackpot hit its cap without a winner, the prize money rolled down to the lower tiers, which briefly gave each ticket a positive expected value. MIT students and a retired Michigan couple (the Selbees) spotted this and bought hundreds of thousands of tickets during roll-down weeks. The lottery's own rule, meant to spread payouts around, created an edge that anyone who did the arithmetic could exploit.

  2. Physics turns roulette into a prediction problem

    Edward Thorp and Claude Shannon built a wearable computer that timed the ball and rotor to estimate where the ball would land, and the Eudaemons later refined the approach. Roulette is not truly random. It is deterministic motion that is sensitive to initial conditions. Measuring those conditions narrows the likely landing sector enough to turn the house edge into a player edge.

  3. Betting edge comes from beating the market's prices

    A profitable bettor does not need to predict outcomes perfectly. They need probability estimates that are more accurate than the odds the market or bookmaker implies. Bill Benter's Hong Kong horse-racing model worked by blending its own forecasts with the public odds, and it profited only where the two disagreed.

  4. Chance as a spectrum of predictability

    Following Henri Poincaré's levels of ignorance, the book treats randomness as a matter of degree. What looks like luck is often a system whose causes are unmeasured, too sensitive to track, or too complex to model. The practical question becomes how much information would be needed to predict an outcome, and whether that information can be obtained.

  5. Regression to the mean in rating teams

    Dixon and Coles built football models that estimate each team's underlying attack and defense strength from past scores. Separating persistent skill from noise lets a model beat punters who overreact to recent form.

  6. Poker bots and game-theoretic unexploitability

    Programs such as Polaris and Cepheus are designed to approach a Nash equilibrium strategy, one that cannot be exploited whatever the opponent does. Human experts, by contrast, often profit by exploiting opponents' specific weaknesses. This shows a trade-off: a strategy that guarantees you won't lose leaves money on the table, and exploiting others' errors exposes you to counter-exploitation.

Save and mark ideas in the app