forex-scalping.com

My news filter failed for years because it lived in my head. Calibration meant writing thresholds I would not negotiate.

Current filter, no extra indicators:

- Tier-1 on the pair: flat from T-15 until spread snap-back and my re-entry clock.
- Second-tier: only pre-written setups, half size, no new ideas in the spike.
- Everything else: normal London rules.

Calibration is reviewing once a month whether I broke the filter. If I find two breaches, I do not add a widget. I shorten permission - for example, second-tier becomes stand-aside for two weeks. The filter has to be followable on a busy Thursday. If it needs a dashboard of indicators to interpret, I will ignore it under stress.

The phrase "actually followed" is doing the work. A perfect filter you override is decorative. I keep the written version next to the platform clock, not buried in a long playbook I never open mid-session. When I calibrate, I change one threshold at a time and keep the rest frozen for two weeks. Changing the whole filter after one bad CPI day is how the sheet becomes a mood diary.

The part I'd steal is shortening permission instead of adding a widget. Every time I bolted something onto my filter after a bad day, I ended up with a rule I needed to interpret, and interpreting is exactly what I can't do at 13:29.

One thing I'd want pinned down is "spread snap-back". Mine is written as spread back under 1.5 times its normal London level and holding there for 60 seconds, then a 3 minute re-entry clock on top. Without a number, snap-back quietly becomes "it looks calmer now".

I also count breaches per event rather than per ticket. On one NFP I broke the filter twice inside four minutes, and counting that as two breaches would have put me on stand-aside for what was really one lapse.

The flip side is a slow leak, like one second-tier ticket at full size every week. That's a single breach each time but a pattern over the month, so I tag size breaches separately from timing breaches.

Do you log a near miss, where you had the order typed in the window but didn't send it? I started counting those as half a breach and it told me more than the actual breaches did.

LondonNewsTrader, near misses taught me more than breaches, so I made the log stupidly concrete. A near miss is any order I typed, sized, and had my finger on send during a blackout window, even if I cancelled it.

Those half breaches clustered on second tier US prints where I had already given myself half size permission. The permission was the leak. Full stand aside for two weeks fixed it faster than any new widget.

I also keep a monthly count of three tags: timing breach, size breach, and near miss. When near misses outnumber real breaches two to one, the filter is clear and my hands are the problem.

The written sheet next to the clock now has one extra line for the current permission state. Second tier allowed, second tier half only, or second tier flat. Changing that line is a Sunday job, not a 13:20 job.

Have you tried a cool down after a near miss, like no new tickets for fifteen minutes? That small pause stopped me turning the near miss into a real one on the next spike.