Position Sizing: The Skill That Keeps Traders Alive

Position sizing is the rule that decides how much money goes into a single trade or holding. It matters more than entry signals: sized too large, a genuinely good strategy can still ruin you during a normal losing streak; sized sensibly, even a mediocre strategy survives long enough to be judged fairly. This guide covers the main sizing methods, the losing-streak arithmetic that makes sizing the first risk decision, and why the famous Kelly formula is almost always used at half strength or less.

Why position sizing beats entry signals

Losing streaks are not a sign of a broken strategy; they are a statistical certainty of a working one. A strategy that wins 50% of the time should be expected to hit a streak of six or seven consecutive losses somewhere in every 100 trades, and a streak of ten somewhere in a long trading career. The only question is what that streak does to the account, and the answer is set entirely by position size:

• Risking 2% per trade, ten straight losses leave 81.7% of the account: a drawdown of 18.3%, unpleasant and survivable.
• Risking 5% per trade, ten straight losses leave 59.9%: a 40.1% drawdown that most people abandon the strategy inside of.
• Risking 10% per trade, ten straight losses leave 34.9%: a 65.1% drawdown that requires a 187% gain just to get back to even.

Same strategy, same signals, three different outcomes ranging from bruise to burial. That is why professionals talk about sizing before they talk about entries.

Past performance does not guarantee future results. For educational and informational purposes only. Not financial advice. Consult a licensed advisor.

Fixed-fractional position sizing

Fixed-fractional sizing risks a constant percentage of current equity on every trade. Define the risk as the distance from entry to stop, then work backward to the share count. With a $50,000 account risking 1%, the risk budget is $500 per trade. Buy a $120 stock with a stop at $114 and the risk is $6 per share, so the position is 83 shares, about $9,960 of stock. Note the useful distinction: the position is nearly 20% of the account, but the risk is 1%, because the stop defines how much of the position is actually exposed.

Fixed-fractional has a built-in survival property: as the account shrinks, the dollar risk shrinks with it, so a losing streak decelerates its own damage. It is the default method for a reason.

Volatility-based position sizing

Volatility-based sizing replaces the hand-placed stop with a measure of how much the instrument normally moves, most commonly the 14-day average true range (ATR). The logic: a stop inside the stock's normal daily wobble is not a risk control, it is a donation. Set the stop at a multiple of ATR, typically 2x, and size from there. If that $120 stock has an ATR of $3, the stop sits $6 away and the 1% sizing math above follows automatically. A quieter stock with a $1.50 ATR would get a $3 stop and 166 shares: same dollar risk, double the shares, because the instrument moves half as much.

The practical effect is that every position contributes roughly equal risk, so one jumpy name cannot quietly dominate the portfolio. That concentration question, how much of an outcome rides on one position, is one of the four things Agenttrading's investment risk analysis panel reads out in plain English on every backtest.

The Kelly criterion, and why full Kelly is too aggressive

The Kelly criterion computes the mathematically optimal fraction of capital to risk for maximum long-run growth. For a bet with win probability p and even payoff, the formula is 2p minus 1: a 55% win rate with 1-to-1 payoffs gives a Kelly fraction of 10% of equity per trade. That number should immediately look alarming, given the streak math above, and it is. Full Kelly is optimal only in a mathematician's sense: it assumes you know your true win rate exactly, that it never changes, and that you can tolerate the ride. None of those hold in markets:

• The inputs are estimates. A backtest that says 55% might really be 51%, and Kelly sized for 55% while reality delivers 51% is over-betting, which Kelly punishes with drawdowns deeper than the edge can repair.
• The drawdowns are brutal even when the inputs are right. A full-Kelly bettor should expect to spend long stretches down 50% or more from the peak. Almost nobody actually trades through that.
• Half Kelly is the practitioner's compromise. Cutting the fraction in half retains about three quarters of the long-run growth rate while dramatically shallowing the drawdowns. Many professionals go further, to a quarter Kelly, which lands close to the small fixed fractions below.

The 1% to 2% risk-per-trade convention

The widely quoted convention is to cap the loss on any single trade at 1% to 2% of account equity. It is not derived from a formula; it is derived from survival. At 1% risk per trade, it takes roughly 69 consecutive losses to halve the account. At 2%, about 35. Both numbers give a strategy room to be wrong for a long time while you evaluate whether it is broken or merely cold. Newer traders and unproven strategies belong at the bottom of the range; a rule you have tested across decades of data can argue for the top.

Position sizing methods compared

Position sizing and the drawdown math

Sizing and drawdown are the same subject viewed from two ends, connected by an asymmetry: losses require disproportionate gains to repair. A 10% drawdown needs an 11.1% gain to recover, 20% needs 25%, 33% needs 50%, and a 50% drawdown needs a full 100% gain. The curve is gentle at the top and vertical at the bottom, which is precisely why sizing rules exist: their entire job is to keep normal losing streaks in the flat part of that curve. When a backtest shows a strategy's maximum historical drawdown, it is showing what the rule plus its sizing did to capital at the worst moment; we walk through reading that number in maximum drawdown explained.

Past performance does not guarantee future results. For educational and informational purposes only. Not financial advice. Consult a licensed advisor.

Test the sizing, not just the signal

Most people backtest entries and eyeball the rest. The more useful habit is testing the whole rule, entry, exit, and the losing streaks between them, against decades of history, and then asking whether you could actually have sat through the worst stretch at your intended size. Agenttrading's trading strategy tester runs a plain-English rule over 20+ years of split- and dividend-adjusted daily data with a 0.1% cost per trade assumed by default, shades the worst drawdown window on the chart, and counts the trades so a thin sample gets called out. Run the idea, look hard at the shaded stretch, and size like you will meet it again. What you decide from there is your judgment, and this is analysis to inform it, not advice to replace it.