Kelly Criterion Explained Simply (with Real Trading Examples)

The Kelly Criterion sounds intimidating but is one of the most powerful ideas in trading. It answers a very specific question: given the edge I have, how much of my capital should I risk on each bet to maximise long-term growth? This guide breaks down the formula, shows real examples, and explains why almost every professional uses fractional Kelly rather than the full formula.

What Is the Kelly Criterion?

Developed by Bell Labs mathematician John L. Kelly Jr. in 1956, the Kelly Criterion is a formula that calculates the growth-optimal bet size given a known edge. Applied to trading, it takes two inputs — your win rate and your ratio of average win to average loss — and returns the fraction of your account to bet.

The Formula

Kelly % = W − ((1 − W) / B)

• W = win rate (as decimal). 0.55 = 55%.
• B = average win ÷ average loss ratio.

Use the Bizzlytics Kelly Criterion Calculator to compute it instantly.

Beginner-Friendly Example

Suppose you win 60% of the time and your average win is 1.5× your average loss (B = 1.5). Full Kelly = 0.60 − (0.40 ÷ 1.5) = 0.60 − 0.267 = 33.3%. Kelly says risk 33% of your account on each trade to maximise growth. In practice, almost no professional risks that much — see fractional Kelly below.

Worked Example Comparison

Why Full Kelly Is Dangerous

Full Kelly maximises geometric growth if your inputs are exactly right. In practice, your win rate and win/loss ratio are estimates from a small sample — probably wrong by 5-10% either way. Full Kelly with slightly wrong inputs produces catastrophic drawdowns (60-70% is normal).

That is why every serious practitioner — from Edward Thorp to modern hedge funds — uses fractional Kelly. Half Kelly reduces drawdown by roughly half while retaining ~75% of the growth. Quarter Kelly is even safer for volatile assets.

How to Use Kelly Without Blowing Up

1. Track at least 50 trades before trusting your inputs. 100+ is better.
2. Never use Full Kelly. Half Kelly is the standard starting point.
3. Cap the result. Even if half-Kelly says 15%, most pros cap at 5% per trade.
4. Recompute quarterly. Edge changes with market regime.
5. Combine with hard limits from the Position Size Calculator — Kelly is one input, not the only input.

Common Mistakes

• Using Full Kelly — devastating drawdowns even when the math is right.
• Small sample sizes — 20 trades is not enough. Your inputs are basically noise.
• Ignoring correlation — Kelly assumes independent bets. Six correlated tech longs behave like one Kelly-sized position.
• Confusing Kelly % with risk % — Kelly is position size (% of capital), not risk (% of capital at risk).
• Chasing higher Kelly by cherry-picking data — filter the last winning month and Kelly looks great. It is not.

FAQs

Why is my Kelly % negative?

It means your system has no measurable edge. Either improve the strategy, gather more data, or do not trade it live.

Is Kelly used by real professionals?

Yes — sports bettors, quants, and traders like Ed Thorp built fortunes using fractional Kelly. Warren Buffett has referenced its logic when discussing concentration.

Does Kelly work for options?

Yes, if you can estimate expected value per contract and volatility. Options add complications (path dependence, gamma) that reduce the reliability of Kelly for beginners.

Should I recalculate Kelly after every trade?

No — that overreacts to noise. Recompute monthly or quarterly with a rolling window of ~50-100 trades.

Key Takeaways

• Kelly = growth-optimal bet size given a known edge.
• Never use Full Kelly. Half Kelly is the practical standard.
• Trust the formula only after 100+ trades of clean data.
• Cap Kelly % with a hard risk-per-trade ceiling.
• Negative Kelly = no edge = do not trade.

Use Bizzlytics trading tools to calculate risk, position sizing, and improve your decision-making. Try the Kelly Criterion Calculator alongside the Expectancy Calculator to validate your edge first.