Missing candles in one-minute data are not nothing

[Shadow Trader]

Scroll through a one-minute chart of a quiet pair at night and gaps appear. A bar is missing here and there, with a jump between its neighbours. The chart simply skips.

A missing bar means no tick arrived in that minute. On a thin pair at a quiet hour, that's normal. It also means that indicators calculate on a series that has holes in it, treating the next bar as if it followed immediately.

The effect on indicators like moving averages or ATR is subtle. A twenty-bar average now covers more than twenty minutes in real time, and its meaning changes.

For backtests, gaps can matter more. Some testers fill missing minutes with a flat bar at the previous close, others leave them out. The two approaches generate different signals.

Checking is easy: count the bars per hour over a month and compare with sixty. Where the count falls short, the data is sparse.

Strategies built around time-based rules, such as "exit after ten bars," are the most affected, because ten bars stops meaning ten minutes.

How sparse is the data on the pairs traded outside the main session?