Take the exact same 6,000 trades and apply two different sizing methods, and you get two unrecognizably different equity curves. Position sizing is not a detail. It is half the system.
The two methods
Fixed lots: every trade uses the same size, say 0.5 lots, regardless of account balance or stop distance. Simple, predictable in dollar terms per pip.
Percent risk: every trade risks a fixed percentage of current equity, with position size calculated from the stop distance. A $100,000 account risking 1% loses $1,000 if stopped, whether the stop is 20 pips or 80 pips away.
Why percent risk compounds and fixed lots do not
With fixed lots, a growing account trades relatively smaller and smaller: your risk per trade shrinks as a percentage of equity, so growth is roughly linear. With percent risk, position size grows with the account, so growth is geometric. Over 11 years, our backtest at 1% percent risk compounds to over +23,000%. The identical trades at a static lot size produce a small fraction of that, because every winning year fails to feed the next.
Why percent risk also protects better
The same mechanism works in reverse. During a drawdown, percent-risk positions shrink automatically: after a 10% decline, each trade risks 1% of the smaller balance, slowing the bleed. Fixed lots keep hammering at full size into the drawdown, so the same losing streak digs a deeper relative hole. Percent risk is anti-fragile in both directions: it accelerates recovery and decelerates decline.
The stop distance detail everyone gets wrong
Percent risk requires calculating size from the stop distance, not using a fixed size with a percent-of-balance margin. A trade with a 60-pip stop must be roughly one third the size of a trade with a 20-pip stop for both to risk the same amount. Systems that skip this (and many cheap EAs do) have wildly inconsistent real risk per trade even when the marketing says fixed risk. Our systems calculate size per trade from the exact stop distance, which is why the backtest risk numbers scale cleanly when you change the risk setting.
Bottom line
Fixed lots make sense in exactly one case: very small accounts where minimum lot sizes leave no room to scale down. Everyone else should be on percent risk. It is the difference between a system that compounds and a system that just accumulates.
The hybrid approaches, and why they underdeliver
Between fixed lots and percent risk live several hybrids that traders invent hoping for the best of both. Stepped sizing (increase lot size at each $10,000 of balance) is percent risk with lag: it captures some compounding but always trades yesterday's size, and the steps create odd risk discontinuities. Equity-curve sizing (increase risk after wins, decrease after losses) feels adaptive but is mathematically a bet on streak persistence, and in our testing it degrades expectancy on systems whose wins and losses arrive independently, which is most systems. Volatility-scaled sizing (risk more in quiet markets, less in wild ones) has a genuine institutional pedigree, but done crudely it just adds a parameter that flatters the backtest. The consistent finding across our testing: plain percent risk with the size recalculated from stop distance on every trade captures nearly all the available benefit at nearly zero added complexity, and complexity is never free. The hybrids mostly exist because plain percent risk feels too simple to be the answer. It is the answer anyway.
Implementation details that quietly change your risk
Even pure percent risk gets corrupted by implementation sloppiness, so audit these four details on any system you run. Rounding: lot sizes round to the broker's step (usually 0.01); on small accounts that rounding can move actual risk 20%+ from target, one more reason for the capital floors discussed elsewhere on this blog. Commission and swap: risk calculated on price distance alone ignores several dollars of round-trip cost per lot, a small but permanent drag that honest backtests include. Simultaneous positions: three concurrent 1% trades are not 1% portfolio risk; correlation between them decides whether it behaves like 1.5% or nearly 3%, which is why our pair selection keeps the systems' concurrent exposures structurally unrelated. Slippage on the stop: percent risk assumes the stop fills at its price; fast markets add a fraction beyond it. None of these is a scandal individually. Together they are the difference between a system whose live risk matches its label and one that is quietly 30% hotter than advertised. Our published backtests price all four in, which is the level of accounting you should demand anywhere.