Leverage & Margin in CFD Backtests

Leverage Is Not Edge

A strategy that makes +2 pips per trade makes +2 pips per trade at any leverage; leverage decides how many euros each pip is worth for your account size — and how quickly a losing streak becomes ruin. So in backtesting, leverage enters in exactly two places: position-sizing feasibility (could you legally and practically hold this size?) and risk compounding (what does the losing streak do to equity?). Everything else about the strategy is leverage-neutral.

The Sizing Formula (Use It in Every Backtest)

lots = (account × risk per trade) ÷ (stop distance in pips × pip value per lot)

Example: €10,000 account, 1% risk (€100), 25-pip stop on EUR/USD (pip value ≈ €9.2/lot at EUR-denominated accounts; $10/lot in USD):

lots = 100 ÷ (25 × 9.2) ≈ 0.43 lots

Backtests sized this way produce results in R-multiples and percent, which transfer to any account size. Backtests sized in fixed lots produce results that silently assume an account — and usually an unrealistic one.

The Margin Feasibility Check

After sizing, check the position fits your margin. Required margin = notional ÷ leverage cap:

Jurisdiction (retail)	Major FX	Minors/Gold/Indices	Crypto
EU (ESMA) / UK	30:1	20:1	2:1
Australia (ASIC)	30:1	20:1	2:1
US (forex)	50:1	20:1	n/a (varies)
Offshore brokers	up to 500:1+	varies	varies

The 0.43-lot example: notional ≈ €43,000 → margin at 30:1 ≈ €1,430, or 14% of the account. Fine. But the same check kills plenty of backtest fantasies: three concurrent 1%-risk positions with 15-pip stops can demand more margin than an EU retail account has — meaning the backtest’s “take every signal” rule was untradeable during clustered signals. Add a margin column when testing strategies that hold multiple concurrent positions.

Drawdown Compounds Through Sizing

Risk-percent sizing interacts with the losing streak: at 1% risk, an 8-loss streak draws down ~7.7%; at 3% risk, ~21.6%. Same strategy, same streak — triple the sizing turns a survivable stretch into an account-threatening one (and a guaranteed prop-firm breach). When you backtest, record the streaks; when you size, let the streaks — not optimism — set the risk percent.

Margin Calls Don’t Belong in (Good) Backtests

If a simulated position would have hit margin call / stop-out (brokers commonly stop out at 50% margin level), the lesson isn’t to model liquidation mechanics — it’s that the sizing rule is broken. Fix the rule, re-run. A strategy that flirts with stop-out levels in backtesting has already failed the test that matters.

Related: the full cost audit · metrics that matter · prop-firm rules

Frequently Asked Questions

Does higher leverage improve backtest results?

No. Leverage doesn't appear in per-unit results at all — a strategy's pips and R-multiples are identical at 5:1 and 500:1. It only scales position size, which scales both profits and the speed of ruin symmetrically. Any 'improvement' from leverage in a backtest is just larger sizing, with proportionally larger drawdowns.

What risk per trade should I use in a backtest?

1% is the conventional starting point; 0.5% for tight-stop high-frequency styles or prop-firm rules; above 2% needs justification from an exceptional losing-streak profile. Whatever you choose, validate it against the backtest's worst streak: streak length × risk% should stay well inside any drawdown limit you must respect.

Do CFDs and spot forex margin work the same way?

Mechanically yes — required margin is notional divided by the leverage cap, per instrument class. The caps differ by asset (indices and gold are usually 20:1 retail in the EU, crypto 2:1), so multi-asset strategies need per-instrument margin checks, not one blanket number.

Should I backtest with compounding or fixed account size?

Both views are useful: fixed-size (every trade risks 1% of the starting balance) isolates the strategy's edge; compounding (1% of current equity) shows the realistic equity curve. Report expectancy from the fixed view and drawdown experience from the compounding view.