Position Sizing: Fixed Fractional, Kelly, and the Case for Fractional Kelly

Position sizing determines how much capital to deploy on each trade, balancing profit potential against the risk of ruin. This paper examines fixed fractional sizing (the 2% rule and variants), the Kelly criterion and its mathematical basis, and why deploying full Kelly capital can lead to drawdowns that exceed practical tolerance, even when the expected edge is mathematically correct.

Core concept

Position size affects both the magnitude of wins and losses and, critically, the probability of portfolio ruin. A single large loss on an oversized position can trigger forced liquidation before a sound edge can compound. The tension is between deploying enough capital to profit from an edge and deploying so much that a normal losing streak obliterates the account.

Fixed fractional sizing (allocating a fixed percentage of capital per trade, typically 1-2%) is simple and lets the position size grow as capital accumulates. The Kelly criterion offers an optimization: it calculates the theoretically optimal bet size given the win rate, average win size, and average loss size, maximizing long-run logarithmic wealth. However, practical trading rarely matches Kelly's assumptions perfectly, and the formula's optimal sizing often leads to volatility and drawdowns that traders cannot tolerate psychologically or operationally.

Mechanics

Fixed fractional sizing works like this: allocate a constant percentage of account equity to risk per trade, typically 1-2%. If the account has $100,000 and the rule is 2% risk per trade, and a trade's stop-loss is 100 basis points away from entry, then position size is calculated as:

Position size = (Account equity × risk percentage) / stop-loss distance

For a $100,000 account, 2% risk, and a 100 bp stop: position size = ($100,000 × 0.02) / 0.01 = $200,000 notional, or 2 times leverage on the base equity. If the stop is hit, $2,000 is lost. If the stop is 50 bp away, position size doubles; if the stop is 200 bp away, it halves. The method is self-adjusting: as the account grows, so does absolute risk per trade, but risk as a percentage stays constant.

The Kelly criterion answers: what fraction of wealth should be wagered on each bet to maximize long-term growth? The formula is:

f* = (bp − q) / b

where:

• f* is the optimal fraction of capital to bet
• b is the ratio of win size to loss size (e.g., if wins average $100 and losses $50, b = 2)
• p is the win rate (probability of profit)
• q is the loss rate (1 − p)

Example: win rate 55%, average win $1, average loss $1 (so b = 1). Then f* = (1 × 0.55 − 0.45) / 1 = 0.10, meaning bet 10% of capital per wager. Over time, with this edge and independent trials, Kelly maximizes the geometric mean (compounding) of returns.^[1]

Why full Kelly fails in practice involves several mechanical and statistical issues:

1. Parameter uncertainty: real traders rarely know p, b, and the win/loss ratio with precision. Kelly is sensitive to input errors; overestimating win rate or reward ratio by even a few percent can recommend a size far larger than justified. Small estimation errors compound into severe over-sizing.

2. Non-independence and market regime shifts: Kelly assumes bets are independent. Financial markets exhibit momentum, reversals, and correlation shocks. A series of correlated losses is more likely than Kelly assumes, and a single such streak at full Kelly sizing can wipe out a majority of capital.^[2]

3. Drawdown magnitude: full Kelly produces substantial peak-to-trough declines even when the edge is real. Empirical study of Kelly-sized portfolios shows drawdowns of 30-50% are common.^[3] For a trader or fund manager, a 40% drawdown can trigger margin calls, client redemptions, or forced liquidations regardless of whether the edge is sound.

4. Time horizon mismatch: Kelly optimizes logarithmic wealth over an infinite or very long time horizon. Most traders operate with finite horizons, capital constraints, and risk tolerance. A drawdown that Kelly accepts as temporary can be catastrophic in a trader's practical timeframe.

Worked example

Consider a commodity futures trader with a 52% historical win rate, average win of $1,500, and average loss of $1,400 (b = 1,500 / 1,400 ≈ 1.07).

Full Kelly: f* = (1.07 × 0.52 − 0.48) / 1.07 ≈ 0.016, or 1.6% of capital per trade.

On a $500,000 account, Kelly recommends risking $8,000 per trade. If typical trade duration is 5-10 days and the trader can execute 100-150 trades per year, running at full Kelly means accepting that a streak of, say, 15 consecutive losses (a 1-in-2,000 probability under independence, but not unreasonable in a volatile market year) would cut capital by roughly $120,000, a 24% drawdown.

In the 2008 financial crisis, many statistical arbitrage funds with mathematically sound edges experienced drawdowns of 20-40% in weeks, forcing redemptions and liquidation.^[4] Even with correct models, the path to ruin under Kelly is narrower than market stress allows.

A conservative alternative: deploy fractional Kelly, say 25% of Kelly, or 0.4% per trade. On the same account, risk is $2,000 per trade. The same 15-loss streak costs $30,000, a 6% drawdown, tolerable for most traders. Long-term compounding is slower, but so is drawdown. In practice, many professional traders use 25-50% Kelly sizing.

Fixed fractional at 2% risk per trade (not 2% of capital as total position, but 2% as loss cap) often lands close to 0.5-1.5% Kelly, depending on win/loss geometry.

Limitations

This analysis assumes stationary win rates and win/loss magnitudes, which markets rarely sustain. Traders who have measured a 52% win rate in one regime may face 48% in the next; Kelly sizing based on backtest data is unreliable. Also, psychological tolerance is ignored; Kelly assumes the trader can stay disciplined through large drawdowns, but human behavior rarely tolerates volatility that theory recommends. Transaction costs, slippage, and leverage constraints also compress the practical edge, making estimated Kelly larger than the true optimal size.

The fixed fractional approach is simpler and more solid to model misspecification but is arbitrary; why 2% and not 1.5%? There is no first-principles justification, only convention. Some traders use different percentages for different strategy types or risk environments, but this adds complexity and is often discretionary rather than systematic.

Kelly criterion derivations also often assume a single bet; in portfolio sizing, positions are not independent and Kelly's application is contested. Some academics argue Kelly is not appropriate for finite-horizon problems or when ruin is costly.^[5] The formula is elegant but fragile when assumptions fail.

Summary

Position sizing is a mechanical lever that amplifies edge into profit but also translates estimation error into loss. Fixed fractional sizing (1-2% risk) is simple, self-correcting, and psychologically tractable but lacks optimization. The Kelly criterion provides a mathematical optimum for long-run compounding but is sensitive to input error and produces drawdowns that exceed most traders' operational tolerance. Full Kelly sizing is most useful as a theoretical upper bound; most practitioners use fractional Kelly (25-50% of optimal) or fixed fractional sizing to balance growth against operational feasibility and psychological endurance.

Key definitions

Kelly criterion: A formula that calculates the optimal fraction of capital to wager on each bet to maximize long-term logarithmic growth, given the probability and magnitude of wins and losses.

Fixed fractional sizing: A method that allocates a constant percentage of account equity to risk per trade, independent of edge magnitude; typical ranges are 1-2% per trade.

Fractional Kelly: Deploying a fraction (e.g., 25%, 50%) of the Kelly-optimal bet size, reducing drawdown and sensitivity to estimation error in exchange for slower compounding.

Peak-to-trough drawdown: The largest percentage loss from a previous high point to a subsequent low point in the equity curve.

Position size (notional): The total market value of a position; often expressed as a multiple of account equity (leverage).

Win rate: The percentage of closed trades that are profitable; does not account for magnitude of wins vs. losses.

References

1. MacLean, L.C., Thorp, E.O., and Ziemba, W.T., "The Kelly Capital Growth Investment Criterion: Theory and Practice", World Scientific (2011). Https://doi.org/10.1142/7599

2. Vince, R., The Handbook of Portfolio Mathematics: Formulas for Optimal Portfolio Construction and Estimation, Wiley (2007).

3. Poundstone, W., Fortune's Formula: The Untold Story of the Scientific Betting System That Beat the Casinos and Wall Street, Hill and Wang (2005).

4. Khandani, A.E. And Lo, A.W., "What Happened to the Quants in August 2007?", Journal of Investment Management (2011). SSRN: https://ssrn.com/abstract=1015987

5. Thorp, E.O., "The Kelly Criterion in Blackjack, Sports Betting, and the Stock Market", Proceedings of the 10th International Conference on Gambling and Risk Taking (1997).

6. Investopedia, "Position Sizing", https://www.investopedia.com/terms/p/positionsizing.asp

Educational research on historical data only. Not investment advice, not a signal, and never a performance promise. Past results do not predict future performance. Every reference is link-verified before publication and every paper is re-audited weekly against the library's editorial standard.

Last reviewed by the PropLedger research pipeline: 2026-10-18. Educational research on historical data, not financial advice.