Why one-minute bar backtests flatter anything near a news release

[Amanhasnoname]

Any strategy that trades within ten minutes of a release needs suspicion before its backtest gets believed. Bar-based testing hides what matters most in those minutes.

Take a one-minute bar with a 20 pip range, where the take-profit sits 5 pips above entry and the stop 5 pips below. Both levels are inside the bar, and the tester has to guess which was touched first. Depending on the method, it often picks whichever favours the strategy. Real order flow might have hit the stop first, with a 3 pip slip on top.

The assumptions underneath are equally friendly. Spread is usually fixed at a calm-market value, fills happen at the exact stop price, and no order is ever rejected. During a release, none of that holds.

A better process for these strategies:

- Use real tick data with variable spread if available
- Add a slippage penalty on stop fills, perhaps 1 to 3 pips around events
- Exclude or separately report the minutes surrounding major releases

Then compare the equity curves. If the edge vanishes after the penalties, it was a bar artifact, not a strategy. If it survives, the result is far more credible.

How much does adding realistic news slippage cut the profit factor of your backtests?