Your Backtest Is Too Optimistic: Building an Honest Slippage Model

Why Zero-Slippage Backtests Are Fiction

Every backtest that assumes zero slippage is lying. When you send a market order, you get filled at the best available price, which is not the price you saw when you decided to trade. The gap between decision price and fill price is slippage, and it always works against you.

S20 models slippage as a function of three variables: instrument liquidity (measured by typical spread), order size relative to typical volume, and time of day. During London/NY overlap, EURUSD slippage averages 0.1 pips. During the Asian session, it can reach 0.5 pips. These differences matter across thousands of trades.

The -2.1R Adjustment

Applying S20's slippage model to the full backtest reduced total R by 2.1, from approximately 536R gross to 533.9R net. That 0.4% reduction seems small, and it is. The reason is that V7 primarily trades liquid instruments during liquid sessions, which naturally minimizes slippage.

The model showed that 80% of slippage cost came from 20% of trades. These were entries during session boundaries, exits during news events, and crypto trades where liquidity is structurally thinner. S16 (spread filter) already blocks many of these situations, which is why the net slippage impact is small.

Honest Accounting Builds Trust

The purpose of S20 is not to improve performance. It is to make backtest results trustworthy. When I report +533.9R over 7.5 years, that number includes realistic slippage estimates. A competitor reporting +536R without slippage modeling is reporting a fiction. The 2.1R difference is small, but the methodology difference is large. Investors and traders who evaluate systems need to know whether the numbers they are looking at account for real-world execution costs. S20 ensures that V7's reported numbers can survive the transition from backtest to live trading without unpleasant surprises.