The first spike is mostly machines
Within the first fraction of a second after a headline number, algorithms parse it and trade. Human reaction time is a quarter of a second at best, and the typical retail click takes much longer once the platform delay is added.
The first spike therefore reflects the algorithms' read of the number versus forecast. Because many use similar logic, the spike tends to be one-directional and large. Then humans arrive, and the market begins to think.
The implication for scalpers is clear. Any attempt to trade the first spike is a race against machines, which win.
The more interesting question is what comes after. Algorithms are often wrong about nuance. A headline that looks bullish may come with a downward revision to the previous figure. The market corrects within seconds or minutes.
Fading the spike works in some cases and fails in others. What makes it a poor strategy is the cost: the spread is at its widest, and fills are unpredictable.
A more reliable approach waits for the spike to complete, watches for a pause, and takes a measured entry in whichever direction the pause suggests.
What does the market do in the 30 seconds after the first spike, and has that been recorded across enough releases to see a pattern?
The first spike is mostly machines
The first spike is mostly machines
It’s Fairman 
-
FXS HongKong Scalper
- Posts: 543
- Joined: Sat Oct 24, 2026 10:15 pm
Re: The first spike is mostly machines
For human traders, the useful part comes after: the second and third minute, when the market reconsiders the number with the details. A rule like "no entry in the first 60 seconds" keeps most retail scalpers out of the algorithm-only zone.
FXS Official Team | Hong Kong Session Desk
-
Amanhasnoname
- Posts: 2060
- Joined: Fri Oct 30, 2026 9:45 am
Re: The first spike is mostly machines
I actually kept a spreadsheet of this for about forty releases on EURUSD, mostly US retail sales and jobless claims, because I wanted to know whether fading was worth it. What I found was that the half minute after the spike split roughly into two camps. When the headline and the revisions agreed, price usually paused for ten or fifteen seconds and then pushed a little further in the same direction. When the revisions disagreed, the pause was shorter and the retrace was sharp, often giving back half the spike before the first minute was out.
The trouble was that I could not tell which camp I was in fast enough by reading the release myself. By the time I had found the revision line, the move was mostly done.
So my rule now is simple. I do nothing until the spread has dropped back to something close to normal, which on my broker takes anywhere from twenty to forty seconds. Only then do I look at where price has paused and trade the break of that small range with a reduced size.
The trouble was that I could not tell which camp I was in fast enough by reading the release myself. By the time I had found the revision line, the move was mostly done.
So my rule now is simple. I do nothing until the spread has dropped back to something close to normal, which on my broker takes anywhere from twenty to forty seconds. Only then do I look at where price has paused and trade the break of that small range with a reduced size.