forex-scalping.com

Expectancy is where news process either earns its keep or becomes superstition.

I bucket every London scalp for a quarter into three tags: no-news window, second-tier event day, and Tier-1 stand-aside day (including tickets taken after the blackout). Then I look at average R and trade count, not win rate alone.

Last quarter the no-news bucket carried the expectancy. Second-tier days were roughly flat after costs. Tier-1 after-blackout tickets were negative once I included the two revenge clicks I had sworn I did not take. The news process did not need a new indicator. It needed me to stop treating the post-release hour as a bonus session.

What I changed: Tier-1 days now have a hard cap of one attempt after the blackout, half size, only if the setup was on the sheet the night before. Everything else is observation. I also write the skipped windows down so the month review cannot pretend I was "active" through every print.

Expectancy improved the following month, mostly by deleting trades rather than by finding better ones. That is the link. The calendar is an expectancy filter before it is a prediction tool. If a rule does not show up in the R buckets after a month, I cut the rule, not the data.

Your last line is the one I'd be careful with, for the Tier-1 bucket specifically. On my pairs there are four to six Tier-1 days in a typical month, so a rule that only touches those days might get judged on five or six tickets.

At that count one revenge click, which you already found, can decide the verdict on its own. I give rules that apply to Tier-1 days a full quarter before I cut or keep them, and use monthly checks only for rules that touch every session.

The other split I'd try is time since release, tagging tickets as inside the first hour after the print or later in the session. For me the later ones behaved like ordinary no-news trades while the first hour carried almost all the damage. So my one attempt now has to come after the hour is up, unless the sheet named the setup the night before.

How many tickets ended up in your Tier-1 bucket last quarter? I'm curious whether the negative number came from a dozen trades or closer to forty.

LondonNewsTrader, I went back and ran my own quarter through your buckets plus the first hour split I mentioned. The counts matter as much as the averages, so I put both on one sheet.

No-news tickets averaged +0.18R across 142 trades, second-tier days +0.02R across 61, and Tier-1 days split into -0.31R for 14 tickets inside the first hour and +0.09R for 11 after it. The chart below is an illustration of that sheet, with the trade count printed on each bar.
Multiply it out and the no-news bucket made about 25.6R while the first hour on Tier-1 days gave back about 4.3R. Deleting those 14 tickets is worth more than anything I could tune in the second-tier bucket.

The second-tier result is the one I would not cut yet. Mine was flat because costs ate it on releases at 09:30 London, while the same setups on 13:30 release days stayed positive, so the event time decides the verdict more than the tier.

My guess is the London morning spread does most of that damage. Have you split your second-tier days by release time, or does the whole bucket stay flat for you either way?