Kelly Criterion

Crypto & Digital Assets

Updated Apr 2026 • Has calculator

Calculates the optimal fraction of capital to allocate per trade or bet to maximize long-run portfolio growth.

What is Kelly Criterion?

The Kelly Criterion, derived by mathematician J.L. Kelly Jr. in 1956, specifies the fraction of capital that maximizes the expected logarithmic growth of a portfolio over a series of independent bets or trades with known probabilities. Betting the full Kelly fraction is theoretically optimal but leads to extreme drawdowns in practice; most professional traders and fund managers use a 'fractional Kelly' — commonly half-Kelly — to reduce volatility while preserving most of the growth advantage. A negative Kelly result indicates the expected value of the bet is negative and no position should be taken. The formula assumes independent, identically sized opportunities — a simplification that limits real-world application to systematic strategies with well-estimated edge.

Formula

f* = (b × p − q) / b

Worked Example

Backtested 500-trade sample

Step 1 Win rate (p): 60% | Loss rate (q): 40%

Step 2 Net odds (b): 2.0 (win $2 for every $1 risked)

Step 3 f* = (2 × 0.60 − 0.40) / 2 = 0.80 / 2 = 40.00%

Step 4 → Full Kelly: risk 40% of capital per trade

Step 5 → Half-Kelly (recommended): risk 20% to reduce drawdown

Source: Kelly, J.L. (1956) — Bell System Technical Journal (1956-07-01)

Calculate Kelly Criterion

Win Probability (p)

Probability of a winning outcome (0 to 1). E.g. 0.60 = 60% win rate.

Net Odds (b)

Net amount won per unit bet. E.g. 2.0 means you win $2 net for every $1 risked.

Kelly Position Size

—

Not investment advice.

How to Interpret Kelly Criterion

< 0

Negative Edge — expected loss; do not enter position

0 – 10

Small Edge — conservative position (0–10% of capital)

10 – 25

Moderate Edge — consider half-Kelly (5–12.5%)

25 – 50

Strong Edge — half-Kelly recommended (12.5–25%)

> 50

Very Strong Edge — quarter-Kelly to manage drawdown

📚 Advanced Risk — Complete the path

Sources