A news filter is not automatically conservative. It can remove the worst fills from a prop backtest, but it can also remove the trades that paid for the system.
For a prop trader, the useful question is not whether news feels dangerous. The useful question is whether the strategy survives the rulebook better after the news filter changes entry timing, exposure, spread, slippage, and missed opportunity.
What a news filter does in a prop backtest
A news filter blocks or changes trading around scheduled economic events. In a backtest, it might prevent new entries before high-impact releases, close open risk before the event, widen stop logic, or pause the system until spreads normalise.
That sounds simple. It is not. Each version changes a different part of the strategy.
| Filter type | What it changes | What the backtest must prove |
|---|---|---|
| Entry pause | No new trades around selected events | The system does not lose its edge by skipping those trades |
| Forced flat | Positions close before the event | The exit does not create worse drawdown or worse expectancy |
| Post-news delay | Trading resumes after a waiting period | The delay avoids noise without missing the real move |
| Spread guard | Entries stop when spread widens | The cost saving is larger than the lost opportunity |
| Event whitelist | Only specific events are filtered | The event list was defined before the test, not chosen after seeing the curve |
Why news filters can flatter the curve
News filters can overfit because the calendar is visible after the fact. It is easy to look at losing trades, notice that some happened near a data release, and then design a filter that removes exactly those trades.
That is curve fitting with a macro costume.
The honest test defines the event list first. It also defines the time window first: for example, whether the system stops before the release, after the release, or both. Changing the window until the equity curve looks cleaner is no different from tuning a moving-average length until the past agrees with you.
The filter can also create hidden selection bias. If a strategy made most of its money from post-news continuation, filtering those sessions may reduce drawdown and reduce profit at the same time. The final curve may look calmer but less useful. For a prop account, that trade-off matters because the account is constrained by both the loss floor and the target path.
This is close to time-of-day filters in prop backtesting, but sharper. A session filter changes the clock. A news filter changes the event risk attached to the clock.
The rulebook matters before the filter
Prop-firm news rules are part of the test specification. If the account type restricts trading around selected events, the backtest must model that restriction before the result is worth reading.
The mistake is testing a strategy as if it can trade through every event, then running it on a rule set that blocks some of those events. That creates a different strategy. The entries, exits, holding time, and exposure path all change.
For FTMO-style workflows, account type also matters. Some systems are designed for accounts that can hold through news and weekends; others must be force-flat around restricted windows. The difference is explained in FTMO Swing vs Standard backtesting, and the broader prop account framing is on the funding page. The current rule details belong to the prop firm's own pages, because those rules can change.
How to test a news filter without fooling yourself
A clean news-filter test treats the calendar like any other parameter: define it before the run, then punish it with out-of-sample data.
Use this checklist:
- Define the event source and event categories before testing.
- Define the no-trade window before testing.
- Decide whether open positions are held, closed, or only blocked from adding.
- Charge spread, slippage, commission, and swap on every fill that remains.
- Compare filtered and unfiltered results on the same symbols, period, and sizing.
- Check daily loss pressure, max drawdown, time-to-target, and expectancy, not just net return.
- Retest the filter on out-of-sample or walk-forward segments.
- Keep the filter only if it improves rule survival without making the edge too fragile.
The comparison should be path-aware. A filter that improves final return but creates deeper open-equity dips may be worse for a prop account. A filter that reduces drawdown but leaves too few trades may be statistically thin. The backtest has to show both.
The cost model gets harder around news
Scheduled releases often arrive with wider spreads, faster price movement, and worse fills than quiet tape. A news filter is usually trying to avoid that execution problem, so a zero-cost or fixed-cost test is the wrong measuring stick.
realbacktesting's public methodology uses intrabar M1 execution, cTrader broker M1 bars + tick-measured spread from 2021-2026, real per-symbol spread, real commission, swap, 1 bps slippage, and an 80,000 EUR model base. The drawdown path is checked against a 30% out-of-sample hold-out. The method is explained on the methodology page.
Those numbers do not prove a news filter works. They show the kind of execution model a filter has to survive before it deserves attention. If the filter only works because the backtest ignores event spreads or exit slippage, it is not a risk control. It is a prettier assumption.
What this does not prove
A passing news-filter backtest does not prove that the next release will be safe. It does not forecast the data print, model every broker's spread, or guarantee a prop account will pass. It only shows whether the defined filter improved the historical route under the costs and rules you actually tested.
The counter-argument is real too: some strategies are built to trade event volatility. For those systems, a news filter can remove the edge. The answer is not "always filter news" or "never filter news." The answer is to test the rule you will actually run.
Frequently asked
Should a prop backtest always avoid news?
No. A prop backtest should avoid news only if the tested rulebook requires it or if the filter improves rule survival after costs, missed trades, and out-of-sample checks.
Is a news filter the same as a spread filter?
No. A news filter uses scheduled events to block or alter trades. A spread filter reacts to current execution cost. They can overlap, but they test different risks.
Can a news filter be overfit?
Yes. A news filter is overfit when the event list or time window is chosen after seeing which historical trades lost money. Define the rule first, then test it out-of-sample.
What metric matters most for a news filter?
Rule survival matters more than final return. Check daily loss pressure, open-equity drawdown, expectancy, time-to-target, and the number of trades left after filtering.
The stubborn takeaway: a news filter is only conservative after the backtest proves the trades it removed were risk, not the edge.