Backtests Need a Slippage Ledger, Not One Guess

A fixed one-pip slippage assumption can make a forex backtest look precise while hiding the conditions that decide live profitability. Slippage changes with order type, pair, size, session and news. The model should reflect those drivers or show a range of outcomes.

Start with recorded deals from the intended broker. For each trade, store the decision price, submission time, order type, filled price, filled volume and event label. Separate spread from slippage so the same cost is not counted twice.

Build three execution cases

Positive slippage should remain in the raw sample, but a conservative model should not assume it will offset every bad fill. Stops, market entries and limits have different distributions. A limit may show zero negative price slippage because it simply fails to execute. Counting only filled orders then introduces selection bias.

The live calendar supplies a natural event label. The Fed meets September 15–16 after August CPI rose 0.4% month on month. BOJ follows on September 17–18. Trades around those windows should not share a cost assumption with a quiet mid-session sample.

For strategies trading multiple currencies, convert slippage into the account currency at the transaction timestamp. Then calculate its effect on expectancy: win rate alone cannot show whether execution consumed the edge.

The September 13 weekly forex outlook is a useful calendar-oriented companion. It provides interpretation, while the test should use the trader's own fill data and primary event dates.

The loop closes when the model predicts a range that can be compared with new live trades. Update the ledger monthly, investigate drift, and retire a strategy if real execution consistently falls outside the tested band.

Sources

• MetaTrader 5: order and deal concepts
• FCA: best-execution review
• BLS: August 2026 CPI
• Bank of Japan: meeting schedule

A backtest is a model; it cannot guarantee live fills or future profitability.