Technical revision: 1 October 2026 · Educational material
Position size changes how a sequence of trading outcomes affects an account. It does not turn an unvalidated entry rule into a profitable one. Compare methods using the same trade sequence and clearly stated assumptions.
Start with the exposure being controlled
| Method | What stays constant | Main limitation |
|---|---|---|
| Fixed lots | Requested volume | Money exposure changes with stop distance, symbol and account equity. |
| Fixed cash budget | Estimated account-currency loss to stop | The same amount becomes a larger share of shrinking equity. |
| Fixed fractional | Estimated loss as a fraction of chosen equity or balance | Execution and portfolio losses can exceed individual estimates. |
| Volatility-adjusted | A chosen measure of price movement or exposure | The volatility estimate and its regime can change. |
For stop-based sizing, define entry, stop, estimated costs and the account measure before calculating volume. Round down to the broker’s step. If the minimum lot exceeds the budget, an acceptable result is no trade.
Understand drawdown recovery
If an account falls by a fraction d, recovering to its previous level requires a gain of d / (1-d) on the reduced equity. A 10% decline requires about 11.1%; 20% requires 25%; 50% requires 100%. These are arithmetic relationships, not forecasts.
In an idealized model where every loss is exactly a fraction r of current equity, equity after n losses is E0 × (1-r)^n. For 20 losses at 1%, the remaining fraction is approximately 0.818, or an 18.2% decline. This hypothetical model excludes gaps, costs, broker minimums and simultaneous positions.
Include simultaneous exposures
Several open positions can share currency or market drivers. Counting each position’s estimated stop loss is a useful starting exposure measure, but it is not a guarantee that the account loss cannot exceed the sum. Market gaps and correlated movement affect several positions together.
Do not use an arbitrary correlation multiplier to claim a precise portfolio cap. A portfolio model needs defined exposures, covariance assumptions, a time horizon and stress scenarios. When those inputs are uncertain, report the uncertainty.
Treat estimated edge cautiously
Kelly-style sizing depends on a specified outcome model and estimated probabilities/payoffs. Estimates from a small or optimized backtest can be unstable. A formula result should not be presented as a personal risk recommendation, and there is no universal percentage suitable for every account.
Separate an experiment in sizing from a deployment decision. First establish that the signal, costs and execution are measured consistently. Then compare sizes across adverse sequences and realistic broker constraints.
Build an auditable sizing experiment
- Keep the entry and exit sequence fixed while comparing methods.
- Record volume, equity, stop distance and estimated costs before each trade.
- Include rejected below-minimum sizes instead of silently increasing them.
- Report maximum equity drawdown, exposure overlap and margin stress alongside returns.
- Re-run with adverse costs and an untouched chronological validation period.
The MQL5 sizing lesson implements a conservative volume estimate. The backtesting checklist describes how to record a comparison. All examples here are educational and hypothetical.
