Never go full Kelly

Abstract

The Kelly Criterion, a strategy that maximizes the long-term growth of wealth, is often considered too aggressive for most investors. This article proposes two modifications to the Kelly Criterion that aim to make it more practical for real-world applications. First, the Kelly fraction (the proportion of one’s wealth to bet) should be reduced to account for uncertainty about the probability of the event in question. This uncertainty is modeled as a Beta distribution, and a rule of thumb is presented to calculate the appropriate reduction in the Kelly fraction based on the level of uncertainty. Second, the Kelly fraction should be further reduced to account for the information that the market or one’s counterparty has about the event. This is modeled as a weighted average of the investor’s probability estimate and the market’s probability estimate, with the weights reflecting the relative information content of each source. The article argues that these two modifications provide a more realistic framework for applying the Kelly Criterion to real-world situations, where both uncertainty and information asymmetries are present. – AI-generated abstract.