How do execution assumptions distort backtests?

Execution assumptions distort backtests whenever the simulated trade path is cleaner than the trade path a real operator could actually achieve. The classic shortcuts are zero slippage, perfect fills at the close or midpoint, no liquidity constraint, and no delay between the signal and the trade.

Those shortcuts do not damage every strategy equally. The more often a strategy trades, the smaller the expected edge per trade, or the thinner the market, the easier it is for a pretty historical curve to be mostly borrowed from unrealistic execution.

The most dangerous assumptions are usually the ones that quietly improve every trade a little bit. A tiny fill improvement on a high-turnover system can create a backtest that looks fundamentally different from the strategy you would actually run.

Fast and frequently rebalanced systems live closer to the cost boundary. That means the execution model is often part of the strategy, not just a small overlay. If the edge per trade is thin, slightly worse fills can erase the entire thesis even when the raw signal was directionally right.

Stress the assumptions instead of optimizing around them. Run worse fill scenarios, delay the entry slightly, cap participation, and compare the strategy under realistic turnover. Then push the survivors into walk-forward and paper phases so the clean historical path does not get the last word.