Monday, March 9, 2026
TO: The buy-side consensus
FROM: Reality
RE: Your models
Tear them up.
Not partially. Not the Iran scenario footnote on page 47 of your January outlook. The whole thing — the assumptions, the base case, the "risks are balanced" boilerplate that your compliance team loves. It's garbage now. Beautiful, well-formatted, Bloomberg-terminal-ready garbage.
Here's what your playbook said: U.S. CPI would drift toward 2.2% by Q4. The Fed would cut twice in 2026 and once in 2027, with a new chair committed to independence arriving on schedule. The 10-year yield would hold around 4%. Tech would rotate back into favor on AI capex stories. The S&P 500 would return about 12%, easily outpacing the 30-year average. Soft landing. Controlled descent. Gradual re-acceleration.
A perfect little glide path. Very PDF. Very Q4 roadshow.
And then oil went to $92 a barrel and none of it matters anymore.
Let's go through what actually happened while your quant models were busy extrapolating trend lines.
U.S. oil prices hit their highest level since 2023 — sitting at $92.93, up 8.8% in a single session on Friday. The Dow had no green names in the first fifteen minutes of trading. Zero. Goldman Sachs dropped 3.68%. American Express fell 3.61%. Caterpillar lost 2.81%. The kind of morning that makes portfolio managers check whether their phones are working because surely something is broken.
Nothing was broken. Everything was working exactly as it should — which is the problem.
The Dow shed over 1,100 points at session lows. The S&P and Nasdaq slipped roughly 1.4% to 1.8% in early trading. By close, the Russell 2000 had fallen 2.39%, barely holding above 2,500 — nearly flat on the year as the wheels fly off the small cap trade. The "breadth recovery" thesis that was so fashionable in January is currently on life support.
And then came the February jobs report. Payrolls declined by 92,000 — the third time in five months the economy has shed jobs. Forecasters had penciled in -50,000. The miss wasn't catastrophic. It was worse: it was confirmation.
Now here is where the memo gets uncomfortable for the strategists still working the "transitory oil shock" angle.
The 10-year Treasury yield, which sat comfortably below 4.00% in late February, climbed to 4.14% by mid-session on March 5th. Not because growth is ripping. Not because the labor market is overheating. Because oil at $90-plus is doing what oil at $90-plus always does — it reprices inflation expectations faster than any central bank communication can contain them. Bonds sold off alongside equities. For the first time in recent history, the traditional 60/40 portfolio failed to provide protection during a conflict, as both asset classes declined simultaneously.
That's the trade that every institutional portfolio is built around. When the hedge stops hedging, the models are flying blind.
The Federal Reserve's "wait-and-see" approach may reach a breaking point by the June meeting. If energy prices remain at current levels, the Fed may be forced to abandon its projected rate-cutting cycle entirely. Not taper it. Not delay it. Abandon it. And then sit in a room with a softening labor market, sticky wages, and $90 oil, and explain to Congress why their dual mandate is currently a contradiction in terms.
Jerome Powell's term ends May 15th. Whoever steps into that chair inherits a machine with two accelerators and no brakes.
Meanwhile, across town in the semiconductor complex, the week managed to produce a second disaster that would have dominated headlines in any other universe.
Nvidia dropped 2.50% on March 5th. AMD fell 3.38%. The proximate cause: the Trump administration is reportedly working on licensing requirements that would effectively extend AI chip export controls from a China-focused policy to a near-global one. The proposed rules would expand prohibitions to a worldwide net, targeting state-of-the-art GB300 GPUs — chips with up to 700 petaFLOPS FP8 and 288GB of HBM3e memory. Export batches under 1,000 units would require individual scrutiny. Known consignments would need advance approval.
This is the licensing regime that the chip industry has spent three years trying to avoid. And here it is.
The selloff in software stocks this week marks a new market focus on perceived AI losers — the companies that spent the last two years writing "AI-powered" into every earnings call but whose actual AI revenues remain theoretical. The market is now asking a sharper question: who actually makes money from this, and who was just renting the narrative?
Nvidia's fiscal year 2026 revenue came in at $215.9 billion — a 65% year-on-year increase. The fundamentals are real. The question is whether the regulatory ceiling is real too. When export controls expand from China to 40+ nations, the addressable market doesn't shrink incrementally. It gets carved up by geopolitics with a dull knife.
You want a through-line? Here it is.
Every major macro assumption that institutional money was running on in January is now being stress-tested simultaneously. Prior to the outbreak of war, the global economy was showing signs of renewed life — the global PMI's output index hit one of its highest readings since the pandemic in February, signaling annualized GDP growth of around 3.0%. That was the setup. The momentum was real. The PMI data was clean.
And then the Strait closed.
Oil prices act like a tax increase — which may counterbalance the stimulus expected from expanded tax refunds arriving in March and April. The fiscal tailwind that was supposed to carry consumer spending through the spring just ran into a $90 energy headwind. Those two forces don't cancel cleanly. They create friction — for households, for margins, for the companies in between.
Gap fell 13.49% on Friday after blaming winter storms for the closure of 800 stores. Victoria's Secret dropped 8.67% for the second consecutive day. These aren't macro stories. They're consumer stories. But consumer stories are how macro stress eventually makes it to Main Street — through earnings guidance cuts, through inventory panic, through margins that can't absorb energy and wage pressure at the same time.
The strategists will revise their outlooks this week. The notes will arrive in your inbox with words like "recalibrating," "scenario weighting," and "near-term volatility." They will tell you the base case remains intact. They will tell you this is a temporary dislocation.
They said the same thing in March 1973.
The week ahead brings U.S. CPI on Wednesday. Forecast is 2.2% year-on-year, with core CPI month-on-month at 0.2%. If those numbers surprise to the upside — which oil at $90 has a way of eventually engineering — the last argument for a June Fed cut evaporates. Completely. And the 10-year yield will have an opinion about that before the ink on the release is dry.
Watch the bond market. It's the only participant in this conversation that doesn't have a PR department.