News trading enters around scheduled releases, either before, to catch the spike, or after, to follow the direction the market chose. News EAs place pending orders on both sides seconds before a release.
The claim is that releases produce large, fast moves that can be captured mechanically.
We could not test news trading honestly, for the reasons on the gap page multiplied. Historical data does not contain the spread at the moment of release, the requotes, the slippage on pending orders, or the seconds during which the platform is unresponsive. Backtests of news strategies fill both pending orders at their levels and show a profit that never existed.
The reason news EAs are sold with impressive backtests is that the backtest is the one environment in which the strategy works.
If you trade it anyway, run it live at small size through a month of releases and read the actual fills. Then compare them with the backtest.