Mistakes I made with post-news reversion scalps

[LondonScalper]

Post-news reversion scalps, fading the first spike after a release, are part of my playbook. Before they became profitable, I made a lot of mistakes. The main ones, roughly in order of how much they cost me:

Fading too early. I used to enter against the spike within seconds, before spreads had normalised. The cost alone often made the trade a loser before it started.

Fading without a level. Not every spike reverses. The ones that do usually run into a level that already existed before the release. I was fading spikes in the middle of nowhere because they looked overextended.

Fading real surprises. When a figure is far from expectations, the first move is often the start of a trend, not an overreaction. I lost the most money fading moves that were justified.

Doubling down. When the fade went against me, I added to it, sure that the reversion was coming. Sometimes it did, often too late.

Ignoring the second release. A spike from one figure can be followed minutes later by another release that pushes price further.

Trading every release. Reversion works better on some events than others. I now only trade it on releases where my own log shows a positive record.

Most of these I fixed simply by writing rules and following them. The hard part was accepting that many releases should not be traded at all.

[Shadow Trader]

Fading too early and fading without a level are the two that still show up in my older logs. I now wait until spread sits near the pre release median and price has tagged a level that existed before the print.

Extreme surprises stay off the reversion list. When the figure is far from consensus the first spike is often the start of a trend, and my fades there paid for learning that the hard way.

Doubling down is banned in writing. One attempt at half size on a second tier release is the max; if it fails I flat and leave that event alone for the day.

The second release in a cluster still catches me if I skip the calendar check. I keep a sticky with the next print time beside the ticket so I do not invent a quiet window that does not exist.

Which releases stay on your reversion list from your own sample, and which ones did you drop after the surprise fades? I care more about your drop list than the keep list.

[Shadow Trader]

LondonScalper, fading real surprises is the costly one, and I needed a number to stop arguing with myself. I now measure each surprise against the typical forecast miss for that release, using the last 24 prints.

If the miss is under 0.75 of that typical miss, the fade is allowed at normal size. Between 0.75 and 1.5 it drops to half size, and beyond 1.5 I stay out no matter how stretched the spike looks.

The sneaky version is the revision. A headline right on consensus with a large revision to the previous month can push price as hard as a real miss, so I read the revision line before I treat the figure as boring.

One mistake missing from your list is fading into the London fix window. A spike just before 16:00 can extend on fix flows, and two of my worst fades came from ignoring the clock.

The number took the debate out of it. I still feel the pull to fade a big move, but the sheet already says no.

How do you define an extreme surprise in your own log, by points from consensus or by how unusual the miss is for that release? Points alone fooled me on quieter releases.