Calculator
Kelly Criterion Calculator
Kelly Criterion Calculator
100% = Full Kelly (maximum growth, maximum volatility). Most practitioners use 25–50% to tame drawdowns.
Optimal Stake (50% Kelly)
16.25%
Sizing Analysis
The strategy shows strong edge of 65.0% per trade. Applying 50%-Kelly (the fractional setting), you would stake 16.25% of the bankroll — $8,125 of $50,000. A Kelly stake in this range is consistent with a strong edge (65.0% expected value per unit risked), where the math and the volatility are both manageable.
*The Kelly Criterion maximises long-run geometric growth under idealised assumptions: exact knowledge of win probability and payoffs, sequential independent bets, and no psychological cost to drawdowns. Real trading edges are estimated, not known. Educational only, not trading or investment advice.
What the Kelly Criterion Solves
Most traders ask which trade to take; the Kelly Criterion answers the harder follow-up: how much of your money should go on it? In 1956, Bell Labs researcher John Kelly derived the formula for the stake fraction that maximises the long-run growth rate of a repeated betting bankroll. The answer depends on just two numbers — the win probability and the win-to-loss payoff ratio — and produces a stake that is provably optimal: betting more grows the account slower in expectation and risks ruin, while betting less simply leaves growth on the table. Ed Thorp carried the formula from blackjack tables to markets, and it still underpins the sizing logic of quantitative desks today. For Australian retail investors running CFD, futures or direct share systems, Kelly is the antidote to two universal sizing mistakes: betting far too big after a winning streak, and far too small on genuinely edged setups. It also exposes an uncomfortable truth — a strategy with only a small edge deserves a small stake no matter how confident you feel, and overbetting a real edge is the fastest way to blow up an otherwise profitable system. This calculator computes full and fractional Kelly stakes, the expected value per trade, the breakeven win rate and the dollar position size for your trading capital, translating an abstract growth-optimal fraction into an order you can actually place.
How the Kelly Formula Works
The Kelly fraction is f* = (b×p − q) / b, where p is the win probability, q is the loss probability (1 − p), and b is the payoff ratio of average win divided by average loss. The numerator b×p − q is the expected value per dollar risked across many trades; dividing by b normalises it into a fraction of capital to stake. If the formula returns zero or less, the strategy has no edge and the correct position size is nothing — the formula refuses to size a losing game, which is one of its most valuable properties. The growth logic maximises the expected logarithm of wealth — the same objective that compounds a bankroll fastest over a long series of independent bets. A practical consequence: the optimal stake equals your edge divided by the odds, so halving your estimated edge halves your stake. Because full Kelly produces stomach-turning volatility — drawdowns of 50% or more are statistically normal even with a genuine edge — most practitioners bet a fraction of it, typically 25–50%, called fractional Kelly. Fractional Kelly sacrifices some growth for dramatically shallower drawdowns, and the calculator's expected-value and breakeven readouts show what each configuration really implies about your system's quality.
Expert Insights
Overbetting Punishes Harder Than Underbetting
The Kelly curve is asymmetric: doubling the optimal stake does not double growth — it roughly halves it, and beyond that turns it negative, so overbetting destroys wealth while underbetting merely slows accumulation. Edge estimates are always uncertain, so when in doubt err toward half-Kelly. That is exactly why the calculator carries a fraction slider.
Your Edge Estimate Is the Fragile Input
Kelly is only as honest as the win rate and payoff numbers you feed it, and backtests flatter both: fills are worse than modelled, ASX liquidity thins out in size, and samples are smaller than they feel. A system backtesting at a 60% win rate may live at 52%. Halve your estimated edge or drop to quarter-Kelly as insurance against the gap between model and reality.
Kelly Assumes Repeated Bets, Not Single Shots
The optimality proof needs many independent repetitions with stable odds — the setting of a rules-based trading system, not one conviction trade. Applying full Kelly to a single takeover-arb punt or earnings play misuses the maths because there is no 'long run' for that bet. For one-off positions, size by the maximum loss you can tolerate instead.
Actionable Tips
- 1
Track Your Real Win Rate Before Sizing Anything
Log every closed trade with outcomes and dollar amounts for at least 50 to 100 trades, then compute the actual win rate and payoff ratio. Those numbers — not intuition — go into the calculator. Systems without a recorded history are being sized by hope, and the Kelly output inherits whatever optimism the inputs carried.
- 2
Cap the Stake With a Hard Risk Ceiling
Even when Kelly says 15%, most risk frameworks cap single-trade risk at 1–2% of equity, and the smaller figure should win. Check the position size output against your cap and override if needed. Kelly maximises growth, but the ceiling maximises survival — and surviving a bad edge estimate matters more than theoretical maximum compounding.
- 3
Start at Half-Kelly or Below
Set the fraction slider to 25–50% when first deploying a system and increase only after live results confirm the backtest. Half-Kelly retains most of the growth while cutting drawdown depth roughly in half. As live statistics accrue, recalibrate inputs and consider stepping up — never the other way around after losses.
Real-World Examples
Victor Cut His Size Before It Cut Him
Victor's SPI 200 futures system backtested to a 45% win rate with a 2.5:1 payoff, which full Kelly sized at 23% of his $80,000 account per trade. He leveraged up on paper confidence, and a six-loss streak drew the account down almost 60% before the edge returned. Re-running at quarter-Kelly, the same streak would have cost about 15%. He kept the system — which was genuinely profitable — and changed only the sizing, the variable that had actually failed him.
Naomi Discovered Her System Had No Edge
Naomi plugged her discretionary small-cap record into the calculator: a 38% win rate with winners averaging just 1.2 times losers. The formula returned zero — Kelly said do not bet at all, even though the trades had felt right. Confronting the negative expectancy, she rebuilt her exit rules until winners ran past twice the size of losers, and only then did the Kelly stake turn positive and worth sizing.
Eli Let the Fraction Slider Decide
Eli ran his swing-trading stats — a 54% win rate at 2:1 payoffs — and saw full Kelly calling for a 31% stake of his trading capital. Uncomfortable, he dropped to half-Kelly (about 15%) then capped it under his 2%-per-trade risk rule. Over a full year the capped stake compounded steadily through drawdowns he could actually endure, and the system's live edge held close to estimate because oversized-position stress never crept into his exits.
Glossary of Terms
- Kelly Criterion
- The formula f* = (bp − q)/b giving the stake fraction that maximises long-run geometric growth for a repeated bet with known odds.
- Fractional Kelly
- Betting a set proportion of the full Kelly stake — typically 25–50% — trading slower growth for substantially shallower drawdowns and protection against estimation error.
- Payoff Ratio (b)
- The average win divided by the average loss per trade — the odds the strategy collects when right against what it surrenders when wrong.
Frequently Asked Questions
Everything you need to know about this topic.
Ivy Sinclair-Wren
Financial Chaos Analyst
Ivy Sinclair-Wren is a Financial Chaos Analyst covering investing, AI, wealth psychology, and the emotional consequences of opening finance apps during market crashes. Based in Melbourne, she specializes in demystifying the Australian tax code and helping users navigate the intersection of spreadsheet logic and human irrationality.