The week the market stopped behaving like itself
There are weeks where price does what price does — it ranges, it trends, it retests levels with something resembling logic. And then there are weeks where the underlying rhythm just breaks.
This past week was the second kind.
Without getting into specifics that would require a disclaimer longer than this piece, the macro backdrop shifted in a way that compressed volatility in some sessions and then exploded it in others — not in the clean, tradeable way that follows a known catalyst, but in the jagged, dislocated way that follows uncertainty about the uncertainty. Spreads widened on instruments that rarely see spread movement worth noting. Correlations that had been stable for weeks — the kind you start to take for granted — decoupled for stretches of hours. Sessions that typically carry directional intent opened sideways, then lurched, then went sideways again.
I watched my EAs execute into that environment for about six hours on Tuesday before I made the call: everything off.
Not because something broke. Not because a position blew through a stop in some catastrophic way. But because I looked at what the systems were doing — technically correct entries, technically valid setups — and realized the market they were built for had temporarily left the building.
Automation doesn't remove judgment. It relocates it.
This is the thing I keep coming back to, and it's something I try to be honest about when the subject of algorithmic trading comes up — whether that's in conversations here, or in the context of the work we do at Adestto, or in the Académie when students start asking whether EAs mean they can stop thinking.
They don't.
What automation does is compress the decision loop on execution. It removes hesitation at the moment of entry. It enforces consistency across dozens of trades in a way that a human operating manually, under fatigue, under the emotional weight of a losing streak, simply cannot replicate. That's real. That's valuable. I'm not walking that back.
But the judgment doesn't disappear — it moves upstream. Instead of asking "should I take this trade right now?" on every candle, you're asking a deeper, slower question: "Is the environment this system was designed for still present?" And that question requires a human. It requires someone who can look at the broader context, feel the weight of what's happening in the macro, notice that the correlations have gone strange, and decide that the edge the system was built on is temporarily suspended.
A well-built EA has no opinion on its own relevance. It executes when conditions match its rules. If the rules say enter, it enters. It doesn't know that the week is weird. It doesn't know that liquidity is thinner than usual, or that the news cycle is moving faster than the price structure can process. It just executes.
That's not a flaw in the system. That's the nature of the tool. The flaw would be in treating the tool as a substitute for judgment rather than an extension of it.
I've seen traders — and I've been guilty of a version of this myself in earlier years — who automate precisely because they don't trust their own judgment. The bot becomes a way to outsource the discomfort of decision-making. And that works fine until a week like this one arrives, at which point the discomfort doesn't disappear; it just arrives later, bigger, and with worse timing.
The 48 hours I spent with everything paused weren't passive. I was watching. I was reading structure. I was asking myself what kind of week this actually was, and what it would need to look like before I was comfortable letting the systems back in. That's not a failure of automation. That's automation working exactly as it should within a framework that still includes a human at the top of the decision tree.
I turned the EAs back on Thursday afternoon. The environment hadn't fully normalized, but it had settled enough — spreads had compressed back to something reasonable, the sessions were carrying cleaner intent, the structural levels were holding with more conviction. Not perfect. But recognizable.
That recognition is the job.
What I'm watching going into next week
The question I'm sitting with isn't directional — it's about regime. Specifically: whether what happened mid-week was a temporary dislocation that has now resolved, or whether it was the beginning of a sustained shift in how this market is behaving.
There are a few things I'll be watching to get a read on that. The first is whether the correlations that broke down are re-establishing themselves — not just returning to historical levels, but actually moving together again with something like the consistency they had before. Correlation that comes back quickly and holds tends to suggest the dislocation was event-driven and contained. Correlation that comes back weakly, or keeps breaking, tends to suggest something structural has changed.
The second is session behavior — specifically whether the London and New York opens start carrying cleaner directional intent again, or whether we stay in this pattern of aggressive moves that don't follow through. That follow-through is what most of my systems depend on. Without it, the entries may look right but the trades don't develop.
And the third — honestly the most important — is my own state of attention. One of the things a week like this does is recalibrate the eye. After 48 hours of watching without trading, I'm looking at structure differently than I was before. That freshness is an asset. I want to use it carefully, not burn it in the first two sessions by rushing back to full exposure.
The market will tell me what it is. My job next week is to listen before I act.
— Mahugnon, depuis Montréal
⚠️ Personal observations for educational purposes only. Does not constitute financial advice. Trading derivatives carries a high risk of capital loss.
