News Filters in EAs: When Your EA Should Stand Aside

A news filter stops an EA from opening trades around big scheduled releases like NFP, CPI and central bank decisions, when spreads widen and slippage spikes. Tight stop, short timeframe EAs need one most. Wide stop H4 systems often do not.

Around a major scheduled release - a central-bank decision, CPI, non-farm payrolls - the market microstructure changes for a few minutes. Liquidity providers pull quotes, the spread can widen to several times its normal size, and orders fill with slippage that never appears in calm conditions. A human trader feels the tension and waits. An EA, unless told otherwise, trades straight into it.

A news filter is the "told otherwise" part: a rule that keeps the system flat, or at least stops it opening new positions, during a window around scheduled high-impact events.

What actually happens to your costs at news time

• Spread: a pair that normally quotes 1 pip can quote 5-20 pips for a few seconds either side of the print.
• Slippage: stops and market orders fill where liquidity exists, not where the level was. A 2-pip stop can exit 15 pips away.
• Gaps: price can step over levels entirely, so pending orders trigger far from their price.

None of this shows up in a backtest built on fixed spreads - which is one of the ways a backtest lies to you. If your test assumes normal costs at 13:30 on NFP Friday, the live version of that trade will be worse, sometimes dramatically.

Which systems need one - and which don't

The tighter your stops and the shorter your timeframe, the more news matters:

• Scalpers and tight-stop M5/M15 systems: a news spike can cost more than a week of edge. These need a filter.
• Intraday systems with moderate stops: borderline - test both ways and compare.
• H4/daily swing systems with wide stops: news is usually just another candle. A filter often removes good trades along with bad fills.

That last point is the trade-off people miss: news moves are also where trends start. Filtering every event does not only remove bad fills - it removes entries. The question is never "is news dangerous?" but "for this system, does sitting out cost less than trading through?" That is an empirical question, answered the same way as any other design choice: test it on data the optimiser never saw, then walk it forward.

How news filters are built

• Calendar windows: block new entries from X minutes before to Y minutes after named high-impact events for the traded currency. Typical windows are 15-60 minutes each side.
• Volatility guards: instead of a calendar, refuse to trade when the current spread or short-term range exceeds a multiple of its recent average. This also catches unscheduled shocks the calendar cannot know about.
• Management-only modes: some systems keep managing open positions but simply do not open new ones inside the window - a middle ground worth testing.

Learning to read the calendar itself - which events are genuinely high-impact for which pairs - is covered in our economic calendar guide.

Testing a filter honestly

Two rules keep you honest. First, model costs realistically around events - if your platform supports variable spread in testing, use it; if not, at least stress the result with worse fills, as covered in spread modelling in backtests. Second, treat the filter as a parameter like any other: added after seeing results, it is another opportunity to overfit. Validate it out-of-sample like everything else.

A note on prop firms

Several funded programmes restrict trading around named events - some prohibit opening positions within a window, some prohibit holding through them. If you run EAs on funded accounts, the filter is not optional, it is compliance. Our guide to prop firms that allow EAs covers where the lines are.

FAQ

What is a news filter in an EA?

A rule that stops the EA opening new trades (and sometimes tightens management) in a window around scheduled high-impact events such as central-bank decisions and NFP, when spreads widen and slippage spikes.

Do all EAs need a news filter?

No. Tight-stop, short-timeframe systems benefit most because a spread spike can be many times their normal cost. Wide-stop swing systems on H4 charts often measure news as ordinary volatility.

Do prop firms restrict news trading?

Several do. Some funded accounts prohibit opening or holding positions in a window around named high-impact events - check the specific programme's rules before deploying an EA on one.

Nebula builds MetaTrader systems and proves them on data they have never seen before you risk anything - the free tier runs the full engine with one real deploy included.