What is the difference between fixed-lot and risk-based sizing?
Fixed-lot sizing trades the same volume on every trade regardless of stop distance, so a trade with a 100-pip stop risks ten times more than one with a 10-pip stop. Risk-based sizing fixes the amount at risk — typically a percentage of equity — and derives the volume from the stop distance, so every trade contributes comparably.
That difference has a large and under-appreciated consequence for backtesting. Under fixed lots, your risk per trade swings with volatility whether you intended it or not: wide-stop trades in volatile conditions become your largest bets automatically.
Backtests using fixed lots therefore frequently attribute to the strategy what is actually an artefact of which trades happened to have tight stops. If your tight-stop trades cluster in a particular regime, the result is describing that regime's behaviour rather than the rule set's.
Risk-based sizing is what makes trades comparable across instruments too. A one-lot position in gold and a one-lot position in EUR/USD are entirely different bets; a 1%-risk position in each is the same bet expressed twice.
Why is R the honest unit for backtest results?
R — the amount risked on a trade — is the honest unit because it makes results comparable regardless of instrument, volatility, account size or leverage.
A +2R trade means the same thing on gold as on EUR/USD as on an index. Expressing the same outcome in dollars means the comparison depends on how large the position happened to be, which is a fact about your account rather than about the strategy.
R also makes the strategy's shape visible. A system with a 40% win rate and an average winner of +2.5R is a coherent, tradeable proposition whose behaviour you can reason about — you know it will lose more often than it wins and you know roughly how the losses cluster. The same win rate expressed in dollars across differently-sized trades tells you almost nothing.
The practical benefit is that R-based results transfer. A strategy averaging +0.3R per trade means the same thing on a $2,000 account and a $200,000 one, which is exactly what you want from a measurement.
How does position sizing affect drawdown?
Position sizing affects drawdown roughly proportionally, while affecting the probability of ruin far more than proportionally — which is the asymmetry that matters.
Doubling risk per trade roughly doubles drawdown depth in percentage terms. But the chance of hitting a level from which recovery is impractical rises much faster than that, because losses compound against a shrinking base. A 50% drawdown requires a 100% gain to recover; a 20% drawdown requires 25%.
This is why the same strategy at 1% risk and 4% risk are genuinely different propositions rather than the same one scaled. The first survives its worst historical stretch; the second may not survive a slightly worse one that has not happened yet.
The backtest implication is that drawdown figures are meaningless without the sizing rule that produced them. A 12% maximum drawdown is an excellent number or an alarming one depending entirely on whether it came from 0.5% risk per trade or 3%.
Should position size vary with volatility or confidence?
Varying size by volatility is defensible and usually improves consistency; varying it by confidence is where most sizing schemes quietly become a second, untested strategy.
Volatility-based sizing is really just risk-based sizing done properly. If your stop is placed relative to recent range, then a volatile period produces a wider stop, and risk-based sizing automatically reduces volume to keep the money at risk constant. That is the mechanism working as intended rather than an additional layer.
Confidence-based sizing — trading larger on setups you rate more highly — is a different proposition, because it embeds a claim: that your grading actually predicts outcome. That claim is testable and frequently false. Before sizing on it, split your historical trades by your own confidence rating and check whether the high-confidence group genuinely outperformed. Many traders discover their A-setups and B-setups produce statistically indistinguishable results, which means confidence sizing was adding variance and no expectancy.
The same applies to any scaling rule tied to recent performance — increasing size after wins, cutting after losses. These feel prudent and are testable, and testing them against a flat-size baseline on the same trade sequence usually settles the argument quickly.
What position size should you use when backtesting?
Use risk-based sizing at a fixed percentage of equity — commonly between 0.5% and 2% per trade — because it reflects how disciplined traders actually operate and makes every trade in the result comparable.
Keep it constant across every test you intend to compare. This is the single most important rule in this article: never compare two backtests run under different sizing schemes. The differences you observe will be dominated by sizing, not by the strategies, and the conclusion will be confident and wrong.
Be cautious with compounding in the backtest itself. Sizing off current equity is realistic, but it means early results disproportionately shape the equity curve — a good first year magnifies everything after it, and a bad one suppresses it. Testing on fixed-equity sizing as well separates the strategy's behaviour from the path it happened to take.
And keep sizing separate from the strategy logic. A strategy that only works with a particular sizing scheme has embedded a bet on that scheme, and it is worth knowing which of the two is doing the work.