Everyone Owned the Same Trade: The Psychology of the June 2026 AI Selloff

Market Cycle Commentary

Today is Saturday, June 27, 2026. The Nasdaq Composite just closed down 4.6 percent on the week, a fifth straight losing session that ended at 25,297. The S&P 500 ($SPX) gave back about 2 percent and now sits roughly three and a half percent below the record it set on June 2, the first day it ever closed above 7,600. The Dow finished the week green, the only one of the three to do it.

Read those three lines again. The distance between them is the entire story. If you only watched the S&P, you had a soft week. If you only watched the Dow, you barely had a week at all. If you were long the trade the whole market has been long all year, the AI complex, you got taken to the woodshed. The index stayed calm. Your account didn't. That gap is where the lesson lives.

For four months I've been writing one warning in different costumes. Back in May, in Ripping Through the War, I argued the melt-up was a training program teaching you that bad news didn't matter, and that when it finally broke it would break suddenly, on a second-order effect, not on the headline everyone was watching. This week it broke. Not on Iran. On an accounting question about how much money the AI build-out is actually going to make, and when.

What Actually Happened This Week

The crack opened on Monday, June 22, when Alphabet ($GOOGL) fell about 6 percent, its worst day in over a year, on worries about the sheer scale of its AI infrastructure spending and a run of senior talent leaving for rivals. The doubt didn't stay in one name. By Friday it had spread across the whole semiconductor complex, erasing a large slice of chip-sector market value in a matter of days. For the first time in a long while, chips ($NVDA and the names around it) led the tape down instead of up.

Then the second shoe dropped. A New York Times report said OpenAI is now leaning toward delaying its IPO until 2027. The company had filed confidentially with the SEC on June 8, and it's weighing whether to go public this year at a valuation below a trillion dollars or hold out for a cleaner shot at that number later. Part of what reportedly gave it pause was watching SpaceX ($SPCX) struggle after its own debut, the listing I wrote about two weeks ago in A Lizard, a Dog, and a Professor Walk Into the SpaceX IPO. When the most anticipated private company on the planet flinches at the public market, every holder of the AI theme reads it as a tell.

The selling wasn't an American event. It was heavy enough across Asia overnight to trip a circuit breaker on South Korea's Kospi, and Europe's technology shares fell hard alongside it. On June 23, NPR ran a segment asking the question out loud, in plain language: is AI one big bubble. Underneath all of it sat the Fed, which on June 17 held its policy rate at 3.50 to 3.75 percent for a fourth straight meeting and nudged its own projections higher, signaling that rescue cuts aren't on the way. Higher for longer is a direct tax on the most expensively priced, longest-duration assets in the market, which is exactly what the AI leaders are. We walked through that machinery in trading through a Fed pivot without losing your discipline, and it was on full display this week.

None of that is a collapse. It's a question the market had refused to ask for months, asked all at once. A tape priced for perfection doesn't field questions gracefully.

The Catalyst Is Never the Headline You're Watching

Keep this part long after the week is forgotten. For months the entire trading world had its eyes fixed on one risk. The war. Carriers in the Gulf, a faltering ceasefire, the price of oil. It was vivid and it was named and it was impossible to look away from, and your nervous system couldn't stop staring at it. And the market, to everyone's confusion, kept shrugging it off.

Then it broke on something almost boring by comparison. Not a missile. A spreadsheet. The dawning suspicion that the AI build-out might cost more, return less, and arrive later than the price assumed. That's not a coincidence. That's how it almost always goes.

Daniel Kahneman gave the mechanism a name: the availability heuristic. Your brain rates a risk by how easily an example comes to mind, not by how likely it is. A war writes itself across every screen, so it feels enormous, and you price it, hedge it, lose sleep over it. A slow erosion in the math of a capital cycle produces no images at all, so it feels like nothing, and you leave it unpriced. The threat that gets you is rarely the one on the front page. It's the one too dull to make the front page, sitting in plain sight inside everyone's largest position.

In May I wrote that the catalyst that finally cracks a melt-up is almost never the headline you're watching, that it's a second-order effect. I didn't know it would be the AI capital cycle, and nobody did. Naming the day or the trigger in advance was never possible and never the point. The point is structural. The danger lived in the trade nobody was stress-testing, precisely because everyone already owned it and it had only ever paid.

A Crowded Trade Has No Exit

Here's the lesson this market is teaching that the selloff posts before it did not, because the shape of this week was different. June 5 was a broad, correlated deleveraging, everything down at once. This week was a rotation. Money didn't flee the market. It left one room and walked into another. Out of tech, into the defensive corners that pushed the Dow green while the Nasdaq bled. The damage wasn't spread across your whole book. It concentrated in the one theme the entire crowd had agreed on.

That concentration is the quiet risk almost nobody sizes for. When a trade works for long enough, it stops feeling like a position and starts feeling like a fact. Everyone you follow is in it. Every dip in it has been bought. The consensus itself becomes a source of comfort, and comfort is the opposite of the alertness that keeps you sized correctly. You're not diversified because you hold ten names. You're diversified when those names don't all fall for the same reason on the same morning. This week, they did.

We've seen the most expensive version of this before. When Long-Term Capital Management came apart in 1998, the positions that were supposed to be independent all moved against the fund together, because in a real unwind correlations go to one and the diversification you thought you had isn't there. I wrote the full story in the LTCM autopsy. The mechanism that took down two Nobel laureates running a hundred billion dollars is the same one that hits a retail account overweight five AI names. Same software. Fewer zeros.

The body keeps its own version of the score. A crowded trade that turns doesn't feel like a normal loss. It feels like betrayal, because the position had become part of how you saw yourself, the smart one who got AI early. Antonio Damasio's work on somatic markers describes what happens next: the chest tightens, the breath goes shallow, the urge to act arrives before any clean reasoning does, and the story you tell yourself about the tape gets written to match the state your body is already in. A crowd of traders feeling that at the same moment, reaching for the same exit, is why the door is never as wide as it looked on the way in.

Why the Index Is Lying to You Right Now

If your dashboard is the S&P, this week looked survivable, down a couple of percent off a record, nothing dramatic. That headline number is hiding what happened underneath it. A market can fall hard in its leaders and finish nearly flat at the index level, because something else caught the money on the way out. The index is an average, and an average is built to conceal exactly the dispersion that's hurting you. The trader who reassures himself with the S&P print while his actual positions sit in the names that led the decline is comforting himself with a number that doesn't describe his account. Watch your book, not the benchmark. They're telling two different stories this week, and only one of them is yours.

What Discipline Looks Like This Weekend

This is the part the recap won't hand you, and it's deliberately simple, because simple is what survives contact with a red screen. Notice up front that none of it is a call on Monday. I don't know whether the AI names bounce or bleed from here, and neither does anyone saying it with confidence. Mark Douglas built a career on the one truth under all of this: any single trade has a random distribution of outcomes around your edge, and the market owes you nothing. You can't control the next print. You can control these.

• Inventory your concentration before you do anything else. Add up honestly what share of your risk sits in a single theme. If five positions all rise and fall on the same AI sentence, that's not five trades. It's one trade in five disguises. Knowing the real number is the whole point of treating position sizing as a psychology problem rather than a spreadsheet one.
• Don't try to make it back in the names that took it. The pull to win it back in the exact trade that hurt you is the revenge trade wearing a thesis. The hole gets deeper there, not shallower. Sit the re-entry pause from the two-minute pre-trade pause before you touch the position again.
• Size to the regime you're in, not the one you got used to. If your sizing drifted up through a year of the trade only paying, the reset starts at your smallest size until your execution is clean, not with a bigger swing to catch up.
• Re-read your March journal, if you kept one. The notes you took at the spring lows are the most honest training data you own. They tell you what your nervous system feels like when it's wrong. This week is trying to overwrite that file. Refresh it on purpose.
• Run the size-down test. Cut your size in half for one week and watch your decisions. If your selection sharpens and your stress drops, you've just proven your size was too large for this regime, regardless of what last month's statement says.

The traders who came out of the June 5 selloff intact weren't the ones who predicted it. They were the ones whose size was small enough that being wrong for a week was survivable. That's the only edge available against a catalyst nobody can name in advance.

The 2026 Cycle, One Nervous System and Six Tests

Step back and look at what this year has asked of a single trading brain.

• January demanded patience through headline whiplash.
• March demanded the discipline to do nothing while in real pain at the lows, in War, Tariffs, and Your Trading Brain.
• April demanded the humility to buy back higher than you sold, in From Panic to Euphoria.
• May demanded the restraint to size down at all-time highs, in All-Time Highs and the Trader's Dilemma, and then the harder discipline to keep treating risk as real while the tape taught you it wasn't.
• June is teaching the last one in the set: a trade everyone owns isn't safe because everyone owns it. It's fragile for exactly that reason.

One brain, six tests, and the same skill under every one of them. Not a forecast. A nervous system that stays regulated long enough to follow a plan it wrote while it was calm.

So here's the question to sit with this weekend, before the open finds you. Look at your single largest source of risk right now, honestly: the theme, not the ticker. Did you choose that size on purpose, with a written rule, deciding in advance how much you could lose if it broke? Or did the position grow into its size quietly, one winning month at a time, while the crowd around you kept saying it could only go one way? One of those is a decision. The other is a default you haven't noticed yet, and this is the week it sends the invoice.

A crowded trade is not a position. It's a fire exit with one door.

If you don't know where your own discipline and your relationship with risk actually stand right now, this is the week to measure it rather than guess. The TQ Assessment scores your behavioral discipline and your psychological relationship with risk directly, the two muscles a crowded unwind tests hardest. The work of sizing to the regime, staying regulated through a drawdown, and detaching your decisions from the consensus is taught in depth in the Complete Calm Trading Method and in Trade Calm. The selloff already happened. What you own into Monday is still yours to decide.