⚠️ The Kelly Criterion: Why the Math That “Maximizes Growth” Will Destroy Your Account
There is a formula that sounds like the answer to every trader’s prayers. It was developed in 1956 by a scientist at Bell Labs. It tells you exactly how much to risk on every trade, based on your win rate and your average payoff. It promises to maximize the long-term growth of your account. It is mathematically proven. It is used by hedge funds, professional gamblers, and quantitative trading desks. It is called the Kelly Criterion. And if you use it as a retail trader, it will blow up your account. Not might. Not could. Will. Here is why the most mathematically elegant position sizing formula ever devised is a trap for anyone trading their own money — and why the people who actually use it successfully do so in ways that look nothing like what the formula spits out. The Seduction: Why Kelly Sounds Like the Answer The Kelly Criterion is beautiful on paper. It takes three inputs — your win rate, your average win, and your average loss — and outputs a precise percentage of your account to risk on each trade. The formula: Kelly % = (Payoff Ratio × Win Rate − Loss Rate) ÷ Payoff Ratio Where: Simple. Elegant. Mathematically proven to maximize the compound growth rate of your capital over time. Let’s run an example. A trader with a 60% win rate, a $200 average win, and a $100 average loss has a payoff ratio of 2. Kelly says: (2 × 0.6 − 0.4) ÷ 2 = 0.8 ÷ 2 = 0.4, or 40%. Risk 40% of your account on every trade. Stop and read that again. Forty percent. On a single trade. According to the math, this is the “optimal” amount to maximize long-term growth. Now ask yourself: what happens to a $10,000 account that risks 40% per trade and hits three consecutive losses? Trade 1: Lose $4,000. Account = $6,000. Trade 2: Lose $2,400. Account = $3,600. Trade 3: Lose $1,440. Account = $2,160. Three losses. That is all it takes to turn $10,000 into $2,160. A 78% drawdown. From three trades. With a strategy that wins 60% of the time — a genuinely good strategy. Three consecutive losses with a 60% win rate happens roughly 6.4% of the time. That is about one in every sixteen three-trade sequences. Over a trading career of hundreds of trades, it is not a question of if this sequence will hit. It is a question of when. And Kelly says this is the optimal amount to risk. 🧪 The Lab vs. The Real World John Kelly developed his formula at Bell Labs while working on signal noise in telephone lines. He was not a trader. He was not a gambler. He was an engineer solving an abstract problem about information theory. The formula assumes three things that are never true in real trading: 1. You Know Your Exact Edge Kelly requires precise inputs. A 60% win rate, not “around 60%.” A $200 average win, not “roughly $200.” The formula is extremely sensitive to small errors in these estimates. If your true win rate is 55% but you think it is 60%, Kelly might tell you to risk 40% when the correct amount is closer to 25%. If your true win rate is 50% — breakeven — Kelly says risk nothing, but you think it is 60% and you are risking nearly half your account on a strategy with no edge at all. No retail trader knows their exact edge. Edge estimates are noisy, based on small sample sizes, and change over time as market conditions shift. The formula demands precision that does not exist. 2. You Have Infinite Time Kelly maximizes growth over an infinite number of trials. It does not care if you go broke in the short term, because over infinite time, the math works out. You do not have infinite time. You have rent. You have bills. You have a finite lifespan and a finite amount of capital. A strategy that produces the highest possible account balance after 10,000 trades is irrelevant if you are broke by trade 7. And with full Kelly sizing, trade 7 is not a worst-case scenario. It is an inevitability. 3. You Can Tolerate the Drawdowns Full Kelly produces drawdowns that approach 100% over a long enough timeline. The math is clear on this. The account will, at some point, lose almost everything before recovering. The formula does not care. It assumes you will hold on and keep trading. Real humans do not do this. Real humans see their account down 78% and quit. Or panic-trade. Or revenge-trade. Or abandon the strategy entirely. The psychological reality of trading makes full Kelly impossible to execute, even if the math were perfect — which it is not, because your edge estimates are wrong. 📉 The “Half Kelly” Trap At this point, defenders of Kelly will say: “Nobody uses full Kelly. You use Half Kelly. Or Quarter Kelly. That solves the problem.” Half Kelly means dividing the Kelly percentage by two. In our 60% win rate example, Half Kelly says risk 20% per trade instead of 40%. In theory, this is better. It is still suicidal for a retail trader. Three consecutive losses at 20% risk: Trade 1: Lose $2,000. Account = $8,000. Trade 2: Lose $1,600. Account = $6,400. Trade 3: Lose $1,280. Account = $5,120. A 49% drawdown. From three trades. With a “good” strategy and “conservative” Half Kelly. Now consider that most retail traders overestimate their edge. If the true win rate is 50% instead of 60%, and the true payoff ratio is 1.5 instead of 2, the real Kelly number is much smaller — but the trader does not know that. They are risking 20% based on bad estimates, and the drawdowns will be even worse than the math predicts. Half Kelly does not solve the problem. It just makes the destruction slower. 🏦 How the Professionals Actually Use Kelly (It Is Not What You Think)