This forum post explains a dual-engine decision-making framework for investing that combines Bayesian inference (how to update beliefs from new evidence) with the Kelly criterion (how to size bets given a probability estimate). Using stories of Edward Thorp beating Las Vegas blackjack and the base-rate fallacy in medical testing, the author illustrates Bayesian updating with worked examples, then derives the Kelly formula f* = (bp - q)/b, showing that a 60%-edge coin flip warrants a 20% bet. A five-step decision loop is presented: set a prior probability, update via Bayes as news arrives, estimate odds from target and stop-loss prices, compute the Kelly position, and rebalance as new information arrives. The post stresses practical caveats: use fractional Kelly (half or one-third) because probability estimates carry error and over-betting is punished asymmetrically; odds are dynamic; correlated holdings require multi-asset extensions; and emotions are the biggest execution risk. Framed as 'knowledge (Bayes) plus action (Kelly),' the essay argues that maximizing log returns, not expected single-period returns, is what ensures long-run survival and compound growth, referencing Kelly's 1956 paper and Thorp's later work.
A Compass for an Uncertain World: When Bayes Meets Kelly
Summary
This forum post explains a dual-engine decision-making framework for investing that combines Bayesian inference (how to update beliefs from new evidence) with the Kelly criterion (how to size bets given a probability estimate). Using stories of Edward Thorp beating Las Vegas blackjack and the base-rate fallacy in medical testing, the author illustrates Bayesian updating with worked examples, then derives the Kelly formula f* = (bp - q)/b, showing that a 60%-edge coin flip warrants a 20% bet. A five-step decision loop is presented: set a prior probability, update via Bayes as news arrives, estimate odds from target and stop-loss prices, compute the Kelly position, and rebalance as new information arrives. The post stresses practical caveats: use fractional Kelly (half or one-third) because probability estimates carry error and over-betting is punished asymmetrically; odds are dynamic; correlated holdings require multi-asset extensions; and emotions are the biggest execution risk. Framed as 'knowledge (Bayes) plus action (Kelly),' the essay argues that maximizing log returns, not expected single-period returns, is what ensures long-run survival and compound growth, referencing Kelly's 1956 paper and Thorp's later work.
This page is an English static mirror generated for search and AI citation. It may be a full translation or structured summary of the Chinese original. Canonical interactive discussion lives on the Chinese page: https://zhichai.net/topic/177168868