Slippage: What You Can and Can't Model

What Slippage Is — and Isn’t

Slippage is the difference between the price you intended and the price you got. It is not the spread (that’s a known, always-present cost); it’s the additional movement between decision and execution. Causes: price moving during order transmission, thin order books absorbing your size at worse levels, and — the big one — price gapping across your level at news or market open, where no quotes existed in between.

For retail-sized orders in liquid majors during active sessions, market-order slippage is small and roughly symmetric (sometimes price improves). For stops through fast moves, it is one-sided and can be brutal: the stop fills at the first available price beyond the level, not at the level.

The Stop vs Limit Asymmetry

This is the piece most backtests get wrong:

• Stop orders (including stop-losses): guaranteed to trigger, not guaranteed a price. Through a fast move, the fill is worse than the level. Backtest rule: in gap situations, fill at the gap’s far side, never at your level.
• Limit orders: guaranteed a price (or better), not guaranteed to fill. If price only kisses your limit, the realistic outcomes are “filled at your price” or “missed the trade” — never “filled worse.” Backtest rule: require price to trade through your limit (not just touch) before counting the fill; touched-but-not-pierced limits are missed trades in the pessimistic accounting.

These two rules — pessimistic stops, pessimistic limit fills — cost a little accuracy in calm conditions and prevent the systematic optimism that makes backtests lie.

What You Can Model

1. A flat buffer. Add 0–0.5 pips per round trip on liquid majors (more on crosses, gold, thin-session trading) as a standing slippage allowance. It’s crude, but it biases the test the safe way.
2. Gap handling. Across weekend gaps and news candles, fill stops at the far side of the gap. Tick-level replay helps here: with the real price path visible (e.g. in tick data), you can see exactly where the first tradable quote after the gap printed, instead of assuming.
3. Session discipline as a model input. If the strategy avoids holding tight stops through scheduled red news, write that into the rules and the backtest — you’ve designed away the unmodelable tail rather than pretending to model it.

What You Can’t (and Shouldn’t Pretend To)

Nobody — retail or institutional, manual or automated — can backtest your broker’s execution during a specific news spike that happened years ago. Latency, requotes, liquidity-provider behavior, and your order’s place in the queue are unrecorded. Any tool that claims precise historical slippage modeling is modeling an assumption. The professional response isn’t a better guess; it’s robustness: a strategy whose edge survives a doubled cost stack and pessimistic fill rules doesn’t need slippage precision. One whose edge depends on perfect fills has no edge.

Related: the full cost audit · prop-firm rules simulation (daily-loss breaches via slippage are a classic challenge-killer)

Frequently Asked Questions

How much slippage should I assume per trade?

For retail sizes in liquid majors during active sessions: 0 to 0.5 pips per round trip is a reasonable standing buffer. Increase it for exotics, gold, crypto, thin sessions, and any strategy that trades into scheduled news. The buffer's job is directional honesty, not precision.

Does tick replay eliminate slippage uncertainty?

It removes the price-path guesswork — you see exactly which quotes printed, so stop touches and gap fills resolve against real data. What it can't reproduce is broker execution: latency and queue effects in fast markets. Tick replay plus a small pessimistic buffer is the honest combination.

Are guaranteed stops worth it?

Guaranteed stop-loss orders (GSLOs) remove gap risk for a fee — usually a wider spread or a per-use charge, and only at some brokers. If you'd trade with them live, model the fee in the backtest; for most strategies that simply avoid holding through news and weekends, they're not needed.

Why was my live slippage worse than backtested even in calm markets?

Check execution type first: instant-execution accounts can requote, and market-execution accounts always fill but at the current price after transmission delay. Also verify you backtested ask-side entries for longs — what looks like slippage is often just the spread mechanics being missed in the test.