The Kelly Criterion, developed by mathematician J.L. Kelly Jr. in 1956, is a formula for sizing bets to maximize the long-term geometric growth of a bankroll. It answers the question: given a known edge and odds, what percentage of your bankroll should you risk to grow your money as fast as possible?
Why Half Kelly is popular. Full Kelly maximizes long-run growth in theory, but it comes with severe drawdowns along the way. With full Kelly, a losing streak can cut your bankroll by 50% or more even when you have a real edge. Half Kelly sacrifices about 25% of the peak growth rate but cuts variance roughly in half, leading to a far smoother bankroll curve. Most professional bettors and traders use half Kelly or less for exactly this reason.
The trade-off: growth vs. risk. Every step below full Kelly reduces both growth rate and drawdown. Quarter Kelly is extremely conservative — growth is slow, but ruin is nearly impossible over a reasonable sample. Full Kelly generates the most wealth in the long run but requires iron discipline to stomach the swings.
Key limitations. Kelly assumes your probability estimate is accurate. Overestimating your edge is dangerous — betting more than full Kelly on a losing bet accelerates ruin exponentially. Always use conservative estimates and consider fractional Kelly as a buffer against model error.
Practical application in crypto poker. Use Kelly to decide how much of your stack to commit in high-edge spots, how much to sit down with at a table relative to your bankroll, or how aggressively to move up in stakes. It works best when you can accurately estimate your win rate and the effective odds you are getting.
Note: Kelly Criterion assumes independent bets with known, constant probabilities. Real poker and casino games involve correlated outcomes, estimation error, and variable odds. Always verify your edge estimate before applying full Kelly sizing.