Kelly Criterion Position Sizing 2026: The Mathematical Formula Every CFD Trader Should Know
Every professional trader eventually stumbles on the same question: how much should I put on this trade? The intuitive answer — “the same as always” or “the same as my 1% rule” — hides a more elegant mathematical framework that has quietly shaped hedge fund sizing since the 1950s. That framework is the Kelly Criterion.
Developed by physicist John L. Kelly Jr. at Bell Labs in 1956 and popularized for trading by Ed Thorp in the 1960s, Kelly Criterion answers a deceptively simple question: given a positive statistical edge, what percentage of my bankroll maximizes the long-run geometric growth of my capital?
This guide explains the formula, works through concrete examples on forex, gold, stock and crypto CFDs, walks through the full-Kelly versus half-Kelly trade-off, and shows how to combine Kelly with the 1% rule that most retail traders already follow. The framework is universal — it applies to any CFD broker including UZFX regardless of platform, product, or leverage.
Editorial note: Data current as of 1 October 2026. Kelly Criterion is a mathematical framework and does not depend on broker, product, or regulation. It assumes your edge is stable and correctly measured — a critical prerequisite that most retail traders ignore.
What Is the Kelly Criterion
Kelly Criterion is a formula that computes the optimal fraction of capital to bet on a wager with known probability and payoff. In its original form, John Kelly derived it from the information theory of communication, but the trading application came from Ed Thorp, who used Kelly sizing at the blackjack tables in the 1960s and later at the MIT扑克 team.
The key insight is that Kelly does not maximize expected dollar return — it maximizes the geometric mean of wealth, which is the compounded growth rate. A bettor who wins a lottery ticket on $1 and then loses $100 in the next trade has the same expected dollar outcome as one who wins and loses $50 each time, but the second path compounds much faster over the long run.
The Formula
The basic Kelly formula for a binary outcome (win or lose, no draws):
Kelly % = p × b − ((1 − p) ÷ b)
Where:
- p = probability of winning a single trade (win rate, expressed as a decimal between 0 and 1)
- b = profit-to-loss ratio, i.e. average winning trade ÷ average losing trade (in dollars, not pips)
- Kelly % = optimal fraction of bankroll to risk per trade
The formula also has an equivalent form commonly used in trading journals:
Kelly % = p − ((1 − p) ÷ b)
Wait — that is only equivalent when b is defined as net profit ÷ loss rather than gross profit ÷ loss. The cleaner version most CFD traders should memorize is:
Kelly % = (p × W) − ((1 − p) × L) all ÷ (W × L)
where W and L are average win and average loss. This form is easier to plug numbers into from a trading journal.
Worked Example: Forex Scalper
Consider a forex scalper with a 55% win rate, an average winning trade of 30 pips and an average losing trade of 20 pips.
- p = 0.55
- b = 30 ÷ 20 = 1.5
- Kelly % = (0.55 × 1.5) − ((1 − 0.55) ÷ 1.5)
- Kelly % = 0.825 − (0.45 ÷ 1.5)
- Kelly % = 0.825 − 0.30
- Kelly % = 52.5%
Fifty-two and a half percent of your bankroll per trade. This is the number Kelly says you should risk — and almost no live trader can psychologically survive a 52.5% drawdown sequence.
Why Full-Kelly Is Dangerously High
Full-Kelly optimizes growth assuming you know the true edge with perfect certainty. In reality, your edge is estimated from a finite sample of trades and will always be an overstatement in a statistical sense. A 55% win rate from 100 trades has a 95% confidence interval between roughly 45% and 65% — the true edge could easily be 48%, which would produce a Kelly of zero.
Compounding the problem: full-Kelly produces a drawdown of roughly 30-50% before it recovers, and the trader who can tolerate a 50% paper drawdown in backtest cannot psychologically ride the same drawdown in a live account where real money and real P&L notifications are involved.
Half-Kelly: The Professional Standard
Half-Kelly — risk half of the full-Kelly number — captures roughly 75% of the long-run growth at only 50% of the volatility. This is the sizing convention used at most quantitative hedge funds and by every systematic strategy we have encountered that survived beyond five years.
Using our scalper example:
- Full-Kelly = 52.5%
- Half-Kelly = 26.25%
- Quarter-Kelly = 13.1%
Even half-Kelly of 26% is far beyond any retail trader’s comfort zone. This is why practitioners combine Kelly with a hard cap.
Combining Kelly with the 1% Rule
The professional formula is:
Actual Risk % = min(Kelly %, 1%)
For the scalper above, Kelly says 52.5% — but the 1% rule caps sizing at 1%. The trader risks 1%.
For a low-edge trader with a 52% win rate and 1.1:1 reward-to-risk:
- b = 1.1
- Kelly % = (0.52 × 1.1) − ((1 − 0.52) ÷ 1.1)
- Kelly % = 0.572 − 0.436
- Kelly % = 13.6%
Half-Kelly would be 6.8%, still above the 1% cap. Risk 1%.
For a very low-edge trader with a 52% win rate and 1.01:1 reward-to-risk:
- b = 1.01
- Kelly % = (0.52 × 1.01) − ((1 − 0.52) ÷ 1.01)
- Kelly % = 0.525 − 0.475
- Kelly % = 5.0%
Half-Kelly is 2.5%, still above 1%. Risk 1%.
For a trader with a 51% win rate and 1.01:1 reward-to-risk:
- b = 1.01
- Kelly % = (0.51 × 1.01) − ((1 − 0.51) ÷ 1.01)
- Kelly % = 0.515 − 0.485
- Kelly % = 3.0%
Half-Kelly is 1.5%, still above 1%. Risk 1%.
For a truly low-edge trader with a 50.5% win rate and 1.01:1 reward-to-risk:
- b = 1.01
- Kelly % = (0.505 × 1.01) − ((1 − 0.505) ÷ 1.01)
- Kelly % = 0.510 − 0.490
- Kelly % = 2.0%
Half-Kelly is 1.0% — the 1% rule and half-Kelly align. Risk 1%.
For a trader with a 50.1% win rate and 1.01:1 reward-to-risk:
- Kelly % ≈ 0.5%
Now half-Kelly is 0.25%, which is below the 1% cap. The math tells you to size smaller — 0.25% per trade — because your edge is small. This is the case where Kelly actively disciplines you.
The Two Rules of Kelly
Rule 1: If Kelly % is above your risk cap, use the cap.
Rule 2: If Kelly % is below your risk cap, use Kelly %.
Kelly should be treated as an asymmetric ceiling that only bites downward on low-edge strategies. On high-edge strategies the 1% rule dominates. On low-edge strategies Kelly dominates.
Kelly for Different CFD Products
The same formula applies to gold, stock, and crypto CFDs — only the W and L values change. On gold, one pip (0.1 in XAU/USD terms) may move $0.10 per unit on standard contracts. On stock CFDs, one point moves a fixed dollar amount per share. On crypto CFDs, one pip depends on the contract specification.
The critical caveat is that W and L must be measured in dollars, not pips. A 20-pip loss on EUR/USD may equal $200 per standard lot, while a 20-pip loss on XAU/USD may equal $200 per 0.1 lot — the same dollar risk but very different contract sizes. Convert your trading journal to dollar P&L before running the Kelly formula.
Common Kelly Mistakes
Mistake 1: Using a small sample. Kelly sizing on 20 trades produces wildly inflated numbers. Use at least 60 closed trades, ideally 120+.
Mistake 2: Mixing strategies. If your trading journal contains both a 5-minute scalping edge and a 4-hour swing edge, the aggregate win rate is meaningless. Compute Kelly per strategy.
Mistake 3: Using pips instead of dollars. Never. Kelly is a dollar formula. Convert to dollars first.
Mistake 4: Treating drawdown as a defect. Kelly sizing produces drawdowns. Half-Kelly typically produces a 15-25% drawdown over five years — this is the price of optimal growth, not a failure.
Mistake 5: Increasing risk after wins. Kelly sizing is stateless. Do not compound wins into larger Kelly fractions. Compute the base Kelly from a rolling 60-trade window and risk the same fraction regardless of current equity.
How UZFX Traders Apply Kelly
UZFX offers 100+ CFD products — 26 forex pairs, 4 precious metals, 3 energy products, 3 crypto pairs, 7 stock indices, 3 stock CFDs — all under a single ASIC AFSL 001291473 entity with zero commission on the standard account and a $10 minimum deposit.
The Kelly framework is broker-agnostic. A trader on UZFX runs the same calculation as a trader on any other regulated platform. The 100+ product range means Kelly sizing can be applied per product — a gold edge and a GBP/JPY edge have different Kelly fractions, and the trader should risk each trade according to its own mathematical edge, not a uniform percentage.
The standard UZFX account with zero commission and spread-only pricing makes Kelly sizing especially clean: the average losing trade (L) is dominated by the spread plus stop-loss distance, without a hidden commission inflating losses. This gives cleaner W and L estimates and therefore cleaner Kelly output.
Kelly vs The 1% Rule vs Fixed Fractional
Three sizing systems coexist in professional trading. Each has different properties:
| System | Growth rate | Volatility | Simplicity |
|---|---|---|---|
| Fixed fractional (1%) | Linear | Low | Very high |
| Half-Kelly | Compounded, edge-proportional | Medium | Medium |
| Full-Kelly | Maximum | Very high | Medium |
Most retail CFD traders should default to 1% fixed fractional and use Kelly only as a cap on low-edge strategies. Professional quant funds use half-Kelly sized by rolling expectancy. Full-Kelly is reserved for traders who have a statistically proven edge from thousands of trades and can tolerate the volatility.
FAQ
What is the Kelly Criterion formula?
Kelly Criterion is a mathematical formula developed by John Kelly in 1956 that calculates the optimal bet size for a trader with a positive expected value. The basic formula is: Kelly % = (p × b) − ((1 − p) ÷ b), where p is the win probability and b is the profit-to-loss ratio.
Should I use full-Kelly or half-Kelly?
Practically no professional trader uses full-Kelly for live capital. Half-Kelly captures ~75% of the long-run growth at 50% of the volatility. Quarter-Kelly captures ~56% of growth at 25% of the volatility. Start with half-Kelly.
Can I combine Kelly with the 1% rule?
Yes — use min(Kelly %, 1%). Kelly acts as a cap on low-edge strategies; the 1% rule dominates on high-edge strategies.
Does Kelly work for negative-expectancy strategies?
No — Kelly returns zero or negative when the expectation is negative, and this correctly tells you to stop.
How do I calculate Kelly for CFD traders?
Use at least 60 closed trades, compute win rate p, average win W and average loss L, then Kelly % = (p × W) − ((1 − p) × L) all ÷ (W × L). Convert to dollars, not pips.
Final Verdict
Kelly Criterion is the mathematical foundation of optimal position sizing. For most retail CFD traders, half-Kelly capped by the 1% rule is the practical implementation. The framework forces traders to answer three questions they would otherwise avoid: what is my actual win rate, what is my actual reward-to-risk ratio, and what does my edge say I can safely risk?
If you are new to trading, start with 1% fixed fractional and only add Kelly once you have 60+ closed trades on a single strategy. If you are an experienced trader already using the 1% rule, add Kelly as a downward cap on your lower-edge strategies — the math will tell you when your sizing is too aggressive for your edge.
Risk Disclaimer
CFD trading carries a high level of risk and may not be suitable for all investors. You could lose more than your initial investment. Kelly Criterion is a mathematical framework and does not guarantee profitability — it only identifies optimal sizing given a known edge. Past performance does not guarantee future results.
Last reviewed: 1 October 2026. Editorial team at MarketCFD.com. For related education: Position Sizing and the 1% Rule 2026, Risk Management Strategies for CFD Trading 2026, Trading Psychology Guide 2026.