MEMO: RE: The Velocity of Confidence Collapse (A Pattern Recognition Exercise)

TO: Anyone still reading earnings calls instead of order flow
FROM: The desk
DATE: Feb 13, 2026
RE: Why the most dangerous market moves happen in plain sight

---

The Setup

Two weeks ago, the narrative held. It was airtight. AI infrastructure needs will drive capex for five years. Software companies are the future. Tech valuations are "justified." The Magnificent Seven would carry the market like they always do. Money flows where logic points.

Then something shifted. Not dramatically. Subtly at first. A 67.5% gross margin miss at Cisco (down 12%). Weak retail sales on Tuesday. Some chatter about wealth advisors being "threatened" by fintech AI tools. A 13.5% drop in CBRE. By Thursday morning, the MSCI Asia Pacific fell 1%—the first decline in six sessions—and suddenly every computer running the same quantitative model tried to exit simultaneously.

This wasn't a surprise. Surprises are discrete events. This was a revelation: the disruption thesis nobody wanted to fully price was always going to be real, and the market was simply pretending otherwise until the numbers made pretending impossible.

What Actually Happened

Let's catalog it plainly:

Thursday: The S&P 500 fell 1.57%. The Nasdaq fell 2.03%. This wasn't a correction. This was repricing. Tech stocks that had doubled on the thesis that "AI will drive margins" got knocked down 5-12% the moment the first real sign of margin compression appeared.

The Broader Move: In the same breath, Vertiv surged 24% on strong data center guidance. Rivian jumped 14% on delivery beats. Applied Materials rallied 13% on earnings. The market wasn't selling risk. It was rebalancing risk. The AI buildout trade is intact. The "AI will save all software margins" trade is dead.

Capital Flows: Into Broadridge (now at lows unseen since November 2023). Out of Morgan Stanley, Charles Schwab, LPL Financial. Out of real estate—both commercial and residential (home sales expected to drop 4.6% in January). Into energy. Into infrastructure plays that benefit from economic growth without margin compression risk.

The Debt Picture: Retail delinquencies at 4.8% of outstanding debt. The highest point in nearly a decade. Student loans. Credit cards. Mortgages. All deteriorating in lower-income cohorts. This is not noise. This is the canary singing, clearly, from its cage.

Jobs Report Twist: 130,000 jobs created in January. The unemployment rate at 4.3%. This should have been bullish. Instead, bond traders immediately repriced Fed cuts from June to July. Two-year Treasuries shot to 3.5%. The message from the market: strong growth + sticky inflation = rates stay elevated. And rates staying elevated = multiple compression. And multiple compression = the only trade that worked (growth at any valuation) stops working.

The Historical Parallel Nobody Wants to Acknowledge

Zoom out to 2017-2018.

Everyone believed in the narrative: tech stocks are the future. Faang is unstoppable. Valuations don't matter because growth is forever. The market ran on that story for years. Then, in a 48-hour window in February 2018, something mundane happened: a jobs report came in hot, bond yields spiked, and suddenly investors realized that $5 trillion in passive capital was all trying to exit narrow doors at the same time.

The correction lasted weeks. The repricing lasted months. The recovery took years.

We're not at that inflection point yet. But we're close enough to see it. The difference is that this time, the disruption thesis is actually real. Software margins are getting compressed by AI. Wealth advisors are at risk from better algorithms. Real estate is getting bifurcated by capital flows.

The market isn't wrong about disruption. It's wrong about whose margins get disrupted and when. And that wrongness is now getting corrected, stock by stock, sector by sector.

What the Numbers Say

The Dow hit 50,188—a record. The S&P 500 is near all-time highs. But look beneath the surface:

• The iShares Software ETF (IGV) is down 30% from its highs and in bear market territory.
• Palantir is down 20% year-to-date despite being, theoretically, an AI enabler.
• Broadridge, Verisk, Salesforce, ServiceNow—all trading at 52-week lows or testing them.
• Gold got crushed (down 11% in one session), then bounced 6%. Bitcoin sank to $67,500, then edged back toward $74,500. Crypto now down 30% from October's $126,000 record.
• The Russell 2000 (small caps) is outperforming. Value is outperforming. Energy is outperforming.

This is the signature of a market that has lost consensus. The "one trade" is breaking apart. What follows is either (a) a new consensus forms quickly and the market reprices higher on different anchors, or (b) the repricing continues in waves as investors realize the breadth of margin compression is wider than first thought.

The Bet

Here's what I'm watching: Can the Fed stay on hold through June while the consumer quietly deteriorates and software margins get methodically re-rated lower?

History says no. The Fed cuts when things are breaking. We're not there yet. But retail delinquencies, credit card delinquencies, job postings all cooling simultaneously—that's the precursor.

The market is experiencing what I'd call "confidence velocity." Speed of repricing is accelerating. The window between "the narrative is still intact" and "oh god, the narrative might not be intact" is closing.

When that window closes completely, it won't feel like a crash. It'll feel like a gap. And gaps are hard to catch.

The question isn't whether disruption happens. It's whether the market can digest the speed at which it reprices between now and June.