Gap trading acts on the jump between one session's close and the next open, usually on stock indices. The most common rule fades the gap, expecting it to fill; the other trades in its direction, expecting continuation.
The claim is that gaps fill more often than not, or that large gaps signal a trend, depending on who is selling the strategy.
We could not test gap strategies honestly. CFD data around the open does not reliably record the prices a trader could have dealt at, spreads at the open are far wider than the data shows, and fills on a gap are the most uncertain in trading. Any backtest of a gap strategy on retail data is a backtest of prices that were not available.
The reason gap strategies look good in backtests is exactly this: the tester fills the trade at the open print, and the open print is not a tradeable price.
If you trade it anyway, forward test it on a live or cent account, measure the actual fills, and do not believe any historical result.