Complete guide · Portfolio management

Kelly criterion:
the mathematical formula
for sizing your positions

The Kelly criterion is the formula used by the best quantitative investors — from Warren Buffett to Edward Thorp — to calculate the optimal size of a position. It maximises long-term capital growth by accounting simultaneously for your probability of winning and your win/loss ratio.

Updated: May 2026
Reading time: 14 min
Level: Intermediate

What is the Kelly criterion?

The Kelly criterion (also the Kelly Criterion, or Kelly formula) is a money-management rule that calculates the optimal fraction to invest in a given opportunity, by maximising the logarithmic growth of capital over the long term.

It was developed in 1956 by John L. Kelly Jr., a mathematician at Bell Labs, in a paper on information transmission. The analogy with betting — then with the stock market — was popularised by Edward Thorp (author of Beat the Dealer and a pioneer of quantitative trading) in the 1960s and 1970s.

The intuition is simple: investing too little holds back capital growth. Investing too much exposes you to catastrophic losses. Kelly mathematically calculates the optimal balance between the two.

f* = (p × b − q) / b p = probability of winning · b = ratio of average gain / average loss · q = 1 − p

Where:

Interactive Kelly calculator

Kelly calculator — Position sizing
Raw Kelly fraction
Enter your parameters
Applied fraction
After multiplier
Amount to invest
On this trade

Which Kelly fraction should you choose?

Full Kelly theoretically maximises growth, but it is rarely used as-is in practice. The reasons are as much psychological as mathematical: the drawdowns generated by full Kelly are often too painful to sit through.

¼
Quarter-Kelly
Very conservative. Ideal for beginners or illiquid markets (smaller small caps). Low volatility, slow growth.
Full Kelly
Maximum theoretical growth but potentially severe drawdowns. Reserved for experienced investors with a high risk tolerance.

Rule of thumb: Edward Thorp, who used Kelly to generate annual returns of 20% over 30 years, systematically recommends half-Kelly: "Full Kelly is too volatile to be psychologically sustainable over the long term. Half-Kelly gives 75% of the growth for half the volatility — it's the ideal trade-off."

Kelly applied to the screener's signals

Screener Small Caps integrates Kelly directly into its Portfolio module. Here is how the parameters are calculated:

Kelly parameterSource in the screenerTypical value
p (win rate)Track record of J+30 predictions logged in the database~62%
b (gain/loss ratio)Ratio of average gains / average losses on confirmed BUY signals~1.30
FractionHalf-Kelly by default (adjustable in Pro)50%
Regime adjustmentkelly_mult × 0.7 in a bear market / ×1.0 in a bull marketVariable

With these parameters (p=0.62, b=1.30), the raw Kelly is:

f* = (0.62 × 1.30 − 0.38) / 1.30 = (0.806 − 0.38) / 1.30 ≈ 32.8% Half-Kelly applied → 16.4% per position

This result means that with capital of €10,000, the Portfolio module recommends investing around €1,640 on each high-confidence BUY signal — subject to sector-diversification constraints and the minimum cash reserve.

Limitations and precautions

Kelly assumes that bets are independent

The Kelly formula was developed for successive independent bets. In the stock market, positions are correlated — a market correction pushes all your small caps down at the same time. The screener's Portfolio module includes a sector-diversification constraint to limit this correlation risk.

The parameters p and b are uncertain

You never have certainty about the true probability of winning. The p and b estimated from the track record can change if the market shifts regime. That is why half-Kelly is preferable: it is more robust to errors in estimating the parameters.

Kelly does not handle liquidity constraints

On the least liquid European small caps, the position size recommended by Kelly can exceed the daily volume available. The Portfolio module includes a minimum-liquidity filter.

Important: The Kelly criterion is a decision-support tool, not a mechanical rule to follow blindly. It does not constitute investment advice. Past performance from the track record is not a guide to future returns.

Kelly built into the Portfolio module

Automatic Kelly builder, per-signal sizing, diversification constraints. Win rate and ratio b calculated from the real track record. Available on Premium and Pro.

Open the screener →

Go further

Frequently asked questions about the Kelly criterion

The Kelly criterion is a mathematical formula that calculates the optimal fraction of capital to invest in a trade, by maximising the long-term growth of the portfolio. Developed in 1956 by John L. Kelly Jr., it is used by the best quantitative managers to size their positions rigorously.
Kelly formula: f* = (p × b − q) / b. p = probability of winning (win rate), b = ratio of average gain / average loss, q = 1 − p. Example: win rate 60%, ratio b = 1.4 → f* = (0.60 × 1.4 − 0.40) / 1.4 = (0.84 − 0.40) / 1.4 ≈ 31% of capital. Use the interactive calculator above for your own figures.
Full Kelly maximises theoretical growth but generates large drawdowns that are hard to bear psychologically. Half-Kelly (fraction × 0.5) halves volatility while keeping about 75% of the expected growth. It is the balance recommended by most practitioners, including Edward Thorp.
A fixed size (e.g. always 5% of capital per trade) ignores the quality of the signal. Kelly allocates more capital to trades with a high probability of winning and reduces the allocation to less certain opportunities. Over the long term, this dynamic allocation generates higher capital growth than any fixed-size strategy, for the same level of risk.