A formula for optimal position sizing based on edge and odds. It tells traders what fraction of their bankroll to risk on each trade to maximize long-term growth while minimizing the risk of catastrophic loss.
A formula for optimal position sizing based on edge and odds. It tells traders what fraction of their bankroll to risk on each trade to maximize long-term growth while minimizing the risk of catastrophic loss.
The Kelly criterion is a mathematical formula that answers a deceptively simple question: how much of your money should you bet on a single trade? Rather than betting all-in on your highest-conviction trade, or always betting the same tiny amount regardless of opportunity, the Kelly criterion suggests an optimal bet size that depends on both your estimated edge (probability of winning) and the odds you're being offered. It's named after J.L. Kelly Jr., who published it in 1956, though the concept has roots in information theory. The formula itself is elegant: f = (bp - q) / b, where f is the fraction of bankroll to bet, b is the odds you're getting, p is your probability of winning, and q is your probability of losing (1 - p). The practical result is a disciplined approach that lets you grow your bankroll as quickly as possible without risking catastrophic loss.
For prediction market traders, the Kelly criterion has become a touchstone concept because it bridges the gap between theoretical edge and real-world bankroll management. In traditional gambling and sports betting, sophisticated bettors have used Kelly sizing for decades to avoid overbetting high-conviction trades and to avoid underbetting when odds are heavily in their favor. Prediction markets like Polymarket introduce a similar dynamic: you identify an outcome you believe is mispriced, but you need to decide how much capital to deploy. The Kelly criterion offers an answer grounded in mathematical optimality—specifically, it maximizes the expected log of your bankroll over many trades, which is equivalent to minimizing your ruin risk while achieving the fastest possible compounding growth.
When trading on Polymarket, you might encounter the Kelly criterion in a few ways. First, some traders build it into their position-sizing logic: before placing a trade, they estimate the market's implied probability, compare it to their own private assessment, and then use Kelly to decide order size. For example, if you believe a US political outcome has a 65% chance but the market is pricing it at 40%, you've identified an edge, and Kelly tells you how much to risk given that edge and the available odds. Second, algorithmic traders and more sophisticated community participants often discuss Kelly sizing in forums and analysis, treating it as a standard concept that separates disciplined position sizing from reckless overbetting. Third, because prediction markets reward both accuracy and capital deployment, Kelly provides a mathematically grounded way to think about the tradeoff between maximizing expected returns and preserving capital to trade again. The actual Polymarket UI doesn't calculate Kelly for you, so traders must do it themselves or use external tools.
The Kelly criterion is often misunderstood in practice. One misconception is that Kelly sizing guarantees profit or minimizes losses—it doesn't. It optimizes expected log-utility under the assumption that your probability estimates are correct. If your estimate of an outcome's true probability is wrong, Kelly sizing won't save you. Another common error is full-Kelly sizing, which can be psychologically volatile; a series of losses will shrink your bankroll rapidly, and if your edge estimates are slightly off, full Kelly can lead to drawdowns of 20%, 30%, or more. Many experienced traders use fractional Kelly—half-Kelly, third-Kelly, or even quarter-Kelly—to smooth out volatility and give themselves a safety margin if their probability estimates are slightly miscalibrated. Additionally, the Kelly criterion assumes you can take any bet size and that you have a long enough horizon for the law of large numbers to kick in; on Polymarket, liquidity limits may force you to size smaller than Kelly suggests, and a trader with a short time horizon might prefer a different approach. Finally, newer traders sometimes apply Kelly mechanically without validating whether their edge is real, which can lead to outsized losses if they're simply overconfident.
The Kelly criterion sits within a broader ecosystem of concepts traders use to manage risk and size positions. It's intimately related to the idea of edge, your probabilistic advantage over the market. It also connects to expected value, since a positive edge implies positive expected value, and Kelly helps you capitalize on it optimally. Portfolio theory, particularly the Sharpe ratio and the concept of diversification, also touches on optimal bet sizing, though from a different angle—Kelly focuses on a single bet's compounding power, while portfolio theory looks at correlations across many bets. Risk management more broadly, including concepts like bankroll management, loss limits, and position limits, all serve similar purposes to Kelly, though Kelly is more quantitative. In prediction markets, Kelly sizing is sometimes contrasted with simpler heuristics like betting a fixed percentage of bankroll on every trade or using fixed dollar amounts; these approaches are easier to implement but may not extract maximum long-term value from your edge.
Suppose you believe the probability of a Bitcoin price crossing $65,000 by June 30 is 70%, but the market on Polymarket is offering Yes at 50¢ (implying 50% probability). You have identified an edge. Using Kelly criterion, with p=0.70, q=0.30, and b=1 (even money), you calculate f = (1 × 0.70 - 0.30) / 1 = 0.40, meaning you should allocate 40% of your trading bankroll to this position.