The Kelly Criterion for Crypto Trading
The Kelly Criterion is a mathematical formula that tells you the optimal percentage of your account to risk on any given trade. Used by professional gamblers and quant funds for decades, it's one of the most powerful — and most misunderstood — tools in trading.
What the Kelly Criterion is
The Kelly Criterion, developed by physicist John Kelly in 1956, answers one question: given your edge and your odds, what fraction of your capital should you bet to maximize long-term growth?
Bet too little and you grow slowly. Bet too much and volatility destroys you before your edge plays out. Kelly finds the mathematical optimum.
It was originally developed for information theory at Bell Labs, then adopted by blackjack card counters in the 1960s, then by commodities traders, and now by quantitative hedge funds. The math doesn't change between blackjack and BTC/USDT perpetuals.
The Kelly formula
Where:
f* = fraction of capital to risk
W = probability of winning (win rate)
L = probability of losing (1 − W)
R = average win ÷ average loss (R:R ratio)
In plain English: Kelly balances your win rate against your reward-to-risk ratio to find the size that grows fastest over time.
Crypto trading example
Let's say you've backtested a strategy on Bybit over 200 trades and you have the following statistics:
Applying Kelly:
f* = (1.10 − 0.45) ÷ 2
f* = 0.65 ÷ 2
f* = 0.325 = 32.5%
Kelly says to risk 32.5% of your account on each trade. That sounds terrifying — and for good reason. Full Kelly is almost never used in practice.
Why most traders use Half-Kelly (or Quarter-Kelly)
Full Kelly maximizes long-term growth mathematically — but it assumes perfectly accurate win rate and R:R estimates. In practice, your statistics are based on historical data that may not reflect future conditions. Estimation error is deadly at full Kelly.
If your real win rate is 50% instead of 55%, full Kelly would still tell you to risk too much. The Kelly formula is very sensitive to input errors.
The mathematical insight from Kelly is profound: Half-Kelly gives you 75% of the growth rate of Full-Kelly with roughly half the volatility and drawdown. It's one of the best risk-adjusted points on the curve.
Practical application for crypto traders
Step 1 — Build a track record first
Kelly requires accurate win rate and R:R statistics. You need a minimum of 100 trades to get reliable numbers. Fewer trades and your sample size is too small for meaningful Kelly calculation.
Step 2 — Calculate your Kelly fraction
Use your actual trading statistics — not backtested, not hoped-for. Your live win rate and actual average win/loss amounts.
Step 3 — Apply Half or Quarter Kelly
Divide your Kelly fraction by 2 or 4 for safety. Given the volatility of crypto markets and the uncertainty in any trading system, Quarter Kelly is often the right starting point.
Step 4 — Recalculate regularly
Your edge changes over time. Strategies stop working. Market conditions shift. Recalculate your Kelly fraction every 50–100 trades.
Important: Kelly assumes independent trades — each trade doesn't affect the next. In crypto, trades are not truly independent. Correlated market conditions can create losing streaks longer than Kelly's assumptions allow. This is another reason to use a reduced Kelly fraction.
What if Kelly says risk 0% or less?
If your Kelly fraction is zero or negative, your strategy has no mathematical edge. You should not be trading it with real money. This is one of Kelly's most useful properties — it explicitly tells you when a system doesn't work.
f* = (0.40 × 1 − 0.60) ÷ 1 = −0.20
Result: No edge. Do not trade this system.
Limitations of Kelly in crypto
- Requires accurate statistics. If your win rate or R:R estimates are wrong, Kelly gives dangerous answers. Garbage in, garbage out.
- Assumes continuous compounding. Kelly was derived for continuous betting. Discrete trades with different sizes and timeframes create deviations from theoretical optimal growth.
- Ignores correlation. A string of correlated losses (all triggered by the same macro event) creates larger drawdowns than Kelly's iid (independent and identically distributed) assumption expects.
- Ignores fees and slippage. In crypto futures, fees of 0.05–0.15% per side plus funding rates reduce your actual edge compared to theoretical edge.
- Psychological reality. Even mathematically optimal Kelly can produce 30–40% drawdowns over extended losing streaks. Most traders cannot tolerate this emotionally and abandon the system at the worst time.
The main takeaway from Kelly for most retail traders isn't to use the exact formula — it's to understand that there is a mathematically optimal bet size, that betting too much destroys compounding, and that position sizing matters as much as entry and exit timing.
Try it: your Kelly fraction
FAQ
Calculate your risk per trade
Use our free Position Size Calculator to apply consistent risk management on every trade.
Open Position Size Calculator →
