The Game Was Rigged in Their Favor. 28% Went Bust Anyway | Kris Abdelmessih on How to Size Bets

Episode Summary

Executive Summary: The episode explains the Kelly criterion as a way to size bets and decisions to maximize long-term compounded growth, not just expected value. Using a biased-coin experiment and real-world analogies like self-insurance and extended warranties, the discussion shows how overbetting destroys capital, why repeated games matter, and why full Kelly is often too aggressive in practice due to uncertainty in estimating odds.

Main Topics: Kelly criterion basics (Priority: 5/5): The hosts introduce Kelly as a formula for determining what fraction of bankroll to bet in an advantage game, emphasizing that it maximizes long-run compounded growth rather than immediate expected value. Repeated games vs. single bets (Priority: 5/5): A central lesson is that optimal sizing changes when a favorable bet is repeated many times; betting too much can lead to ruin even in a positive-EV game because compounding works both ways. Formula intuition: probability and payout (Priority: 5/5): The Kelly fraction is explained as probability of winning minus probability of losing, adjusted by payout odds (inflating or deflating the losing term depending on reward structure). Real-world applications beyond gambling (Priority: 4/5): The conversation maps Kelly thinking to insurance, self-insuring a home, and buying extended warranties, showing how to translate abstract odds into everyday decisions and bankroll risk. Why full Kelly is often too spicy (Priority: 5/5): Full Kelly maximizes growth mathematically, but it also creates large drawdowns and depends on knowing true probabilities; the speakers argue for fractional Kelly because of uncertainty and risk management. Empirical evidence from the biased-coin study (Priority: 4/5): The transcript reviews a study of educated participants betting on a favorable coin flip, showing that many overbet, fail to capture the game’s edge, and sometimes go bust despite the favorable odds.

Key Arguments: A good expected-value bet can still be a bad decision if the bet size is too large, because compounding and volatility drag can destroy capital. Kelly is best understood as a bankroll-sizing tool for maximizing long-term compounded growth rate, not as a claim that a bet is safe. The key intuition is to bet a constant fraction of bankroll; as bankroll rises, the bet size rises, and as bankroll falls, the bet size shrinks. Even a highly favorable game can produce poor outcomes if the wager fraction is too high over repeated trials. Fractional Kelly is often preferable in practice because true probabilities are uncertain and overestimating edge leads to overbetting. In real life, you can use the same framework to ask whether self-insurance or an extended warranty is worth the risk relative to your total net worth. Overbetting Kelly is especially destructive because once you pass the growth-optimal point, both growth and risk move in the wrong direction. The framework is useful not just for gambling but for any repeated decision under uncertainty where capital preservation matters.

Data Points: Participants in study: 61 people - Finance-oriented subjects in the observed betting patterns experiment on a biased coin Finance professionals in sample: About one quarter - Roughly a quarter of participants worked professionally in finance Starting bankroll per participant: $25 - Each subject was given capital to bet in the simulation Time allowed: 30 minutes - Participants had a limited session to place bets Win probability: 60% - The coin was biased in the bettor’s favor Lose probability: 40% - Complement to the 60% win probability Expected return per flip: 20% - Betting $1 on the favorable coin produces a $1.20 expected value Optimal Kelly fraction in coin game: 20% of bankroll - For a 60/40 coin with even-money payout, the Kelly bet is 20% Outcome if betting 50% of bankroll over 100 flips: Lose 97% of money most likely - Illustrates volatility drag from overbetting a positive-EV game Expected gain at 20% Kelly over the game: About 7.5x money / 650% return - Expected compounded outcome if betting the Kelly fraction Participants expected to hit cap: 95% - Simulation result the researchers said should occur if bettors sized well Participants who hit cap: 21% - Actual outcome in the study Participants who lost all money: 28% - Significant fraction of subjects went bust despite favorable odds Home value example: $500,000 - Used in self-insurance example Annual insurance premium example: $3,000 - Cost being avoided in the self-insurance scenario Implied return from self-insuring: 0.6% - $3,000 saved on $500,000 risked Disaster probability example: 0.2% - Assumed chance of home loss in the self-insurance example Extended warranty example: $100 warranty on $500 item - Used to illustrate Kelly-style decision-making in consumer purchases Kelly drawdown risk: One-third chance of losing two-thirds before doubling - Illustrates the severity of drawdowns under full Kelly

Pivotal Quotes: "A good bet can be bad even if you bet it too big." — Chris Adlamasia: Core takeaway on why sizing matters more than just having an edge "Overbetting Kelly is an absolute disaster." — Chris Adlamasia: Warning that going beyond the growth-optimal fraction destroys value "It's not just do I have an edge, like, how much of my bankroll does this edge deserve to get?" — Matt Ziglar: Framing the practical decision-making lesson at the end of the discussion

Implications: Listeners should focus less on spotting edges and more on sizing them correctly. The Kelly framework can improve investing, insurance, and everyday choices, but only if uncertainty, drawdowns, and total bankroll are respected.