Slippage by hour is the table most traders never build
Posted: Sat Oct 31, 2026 7:44 am
Average slippage across all trades is a nearly useless number. The useful one is slippage by hour, and it takes twenty minutes in a spreadsheet to build.
Export closed trades, compute signed slippage per entry (fill minus clicked, sign-adjusted so negative means worse for you), and group by hour of server time. Add a column for count and another for the worst case. Then look.
Typical shapes: a calm middle of the London session where slippage hugs zero, a ragged edge near the New York open, a spike around news hours, and a dead zone during rollover and the Asian lull where spreads are wide and fills are slower. Which hours are bad varies by pair and by broker, which is exactly why a personal table beats any rule of thumb.
The point is to turn it into rules. If 14:00 to 14:30 server time costs 0.4 pip on average and the strategy's edge is 0.8 pip per trade, that window is eating half the edge. Cut it, and the equity curve will often smooth out without touching the signal logic.
Sample size is the catch. Under 20 trades in an hour bucket, treat the number as a hint, not a result.
Which hour of the day is your worst for fills, and did removing it change results?
Export closed trades, compute signed slippage per entry (fill minus clicked, sign-adjusted so negative means worse for you), and group by hour of server time. Add a column for count and another for the worst case. Then look.
Typical shapes: a calm middle of the London session where slippage hugs zero, a ragged edge near the New York open, a spike around news hours, and a dead zone during rollover and the Asian lull where spreads are wide and fills are slower. Which hours are bad varies by pair and by broker, which is exactly why a personal table beats any rule of thumb.
The point is to turn it into rules. If 14:00 to 14:30 server time costs 0.4 pip on average and the strategy's edge is 0.8 pip per trade, that window is eating half the edge. Cut it, and the equity curve will often smooth out without touching the signal logic.
Sample size is the catch. Under 20 trades in an hour bucket, treat the number as a hint, not a result.
Which hour of the day is your worst for fills, and did removing it change results?