Week 45 — Nov 3 – Nov 6, 2026

Start with the rule, because the rule is the story. The model has three alarm rules, called the fast cut, that sit on top of the regime gauge. Each watches one thing: the plumbing of the money markets, the price of credit, and the stock market's own trend. If any one of them trips, the model cuts the risky sleeves to a fixed defensive cap no matter what the gauge says. On Thursday the credit rule tripped. The gap between what Baa-rated companies pay to borrow and what the US government pays rose to 1.57 percentage points, against a six-month average of 1.55. Two hundredths over the line. The rule does not grade on margin, and nothing in it is done by hand, so the cut is in this edition's targets and in the three books.

What a cut actually means here. For the three Macro Tide products, the alarm rule caps the risky sleeves at 35% of the book, down from close to 100%, and the rest goes to the defensive side: cash and managed futures, a fund that follows trends across markets and tends to earn its money when stocks fall for a while. One more rule then acts on top. The defensive side would normally hold long Treasury bonds too, but the model only owns them when the ten-year is cheap to its fair value, and it is not; that bond share is split three-quarters to cash and a quarter to US equity. So the final mix is about 40% in the risky sleeves, 51% in cash and 9% in managed futures. Everything we count as risk is cut in proportion: US equity goes from 48% to 21%, international and emerging-market stocks from 23% together to 8%, commodities from 16% to 5%, gold from 11% to 4%, bitcoin from 3% to 1%. Gold is cut too, which surprises people; in this model gold is a risk sleeve that happens to hedge, and the version of the rule that kept gold whole through a cut did not earn its keep in testing. The books trade at Friday's close, in the ledgers, and the dated rebalance is recorded there. A cut is a position, not a prediction. It reverses when the rule clears, and the rule clears when the Baa gap drops back under its six-month average or the average catches up to it.

How thin the trip is, honestly. The series the rule reads is published daily, and the rule takes the month's latest print as the month's value, so a Friday print back under about 1.55 would un-trip it. Friday's print posts on Monday. The model does not wait for that, by design: a rule that waits for confirmation is a rule that fires late. We would rather take a cut that reverses in a week than skip one that should have been taken. That is the trade this rule makes, and this is the first time in the live record it has been asked to make it.

The week that set it off. Monday and Tuesday were quiet. Wednesday was the best day for the Dow since April 2025, a broad rally in banks, retailers and health care while chip stocks fell. Thursday gave all of it back and more: the S&P fell 4.36%, its worst day since 4 April 2025, and the Nasdaq entered a correction, more than 10% off its October record. Chip stocks had their worst day since March 2020, with the semiconductor fund down 11% on the day and the memory makers down 17% or more. Friday was a small bounce. On the week the S&P lost 2.68% and the Nasdaq 3.37%; the S&P is now 5.7% below its 6 October record.

Why it happened, in a sentence. Democrats won both chambers in Tuesday's midterms, and the market read two things into that: newly elected state governors are likely to slow data-centre permitting, which Goldman put at about 45% of planned US capacity, and a divided government pushes toward an end to the Iran war, which pulled oil down. The first hit chips and the AI trade. The second hit energy stocks and helped everything that suffers from expensive oil. Neither is a gauge input. The gauge reads data, not politics, and the data this week was strong.

The economy is not the problem. Payrolls rose 165,000 in October against about 60,000 expected, with the two prior months revised up by a combined 84,000. Unemployment fell to 4.1%. Jobless claims are under 200,000 and have been since mid-September. Factory orders, job openings and both ISM surveys came in firm. Productivity grew 3.6% annualised in the third quarter while labour costs per unit of output fell, which is the combination that lets wages rise without pushing up prices. Consumer sentiment rose and households' one-year inflation expectations fell to 4.3% from 4.7%. If the stock market were reading the economy it would have risen this week. It was reading politics and the AI trade.

Oil fell 13% in a week, and that is the inflation story. The December WTI contract went from 104.12 to 90.34, most of it on Wednesday, on the Iran-war read above. Energy stocks fell 12% and the commodities sleeve we hold fell 6%. Inflation expectations priced into bonds dropped from 2.34% to 2.23%. Our inflation cycle is driven mostly by oil this year, and the oil series the gauge reads is the EIA's, which lags a few days and still shows about 100; the fall will show up in the gauge over the coming prints. Apollo's economist wrote on Wednesday that a return to conflict is the risk the other way, since Iranian exports are near zero and the loop runs as long as they stay shut in. We are not forecasting it either way.

Gold did its job this week. It rose 4.7%, silver 5.1%, both on the week the stock market fell hardest. That is what the gold sleeve is for, and the first week in three that it has done it. Bitcoin fell 1%, bonds were flat, and the long-bond sleeve we do not hold would not have helped: the ten-year yield fell only 8 basis points to 4.86%.

The Fed said nothing and the market moved anyway. No official moved markets this week. The jobs report pushed a December rate increase further into the price; the prediction market has it at about 76%, up from 72% a week ago. The two-year yield fell to 4.51% even so, and that is one reason conviction rose: the gauge reads a lower two-year as policy getting a little less tight over six months. The plumbing was calmer: the Treasury's cash balance fell 43 billion dollars, so bank reserves rose 36 billion to 2.93 trillion, and the overnight secured rate fixed two basis points over what the Fed pays on reserves. The funding alarm rule is four basis points from its trigger, a little closer than last week on the 21-day average but not moving fast. The Treasury's quarterly refunding on Wednesday kept coupon sizes unchanged and kept its guidance that they stay unchanged for "the next several quarters", with up to 70 billion dollars of buybacks for liquidity. Nothing there for the long end to worry about.

Credit, the leg that fired. High-yield spreads went from 3.26% a week ago to 3.48% on Thursday, which is high-yield spreads widened 22 basis points. The weakest tier, companies rated CCC, now pays about 13.2% over Treasuries, nine and a half points more than single-B companies; Apollo's economist asked on Thursday whether that stress is spreading to single-B, and set out three conditions for it to climb a notch: time at high rates, a refinancing window closing, and forced selling of CCC loans pulling single-B with it. Last week we recorded Jeffrey Gundlach saying our instrument does not look where the stress is, and that credit was the leg with the least room. A week later the instrument fired. Both things are true: the Baa gap is a slow, high-quality measure, and it still moved.

Breadth. The good, the bad, and the ugly. The good is remarkable for a week like this. The equal-weight index, which treats every stock in the S&P the same, fell 1.0% on the week against the cap-weighted index's 2.7%, and on Wednesday it closed at a record high while the index was still 2% off its own. On Thursday it fell 2.9% against the index's 4.4%. Health care, financials, staples, materials, Apple, Microsoft and Lilly all set 2026 highs on Wednesday. This is a market where the average company is doing fine and the AI complex is being repriced. The bad is that we do not hold the average company. Our US sleeve is the S&P fund, the mega-cap fund and a momentum fund, and the momentum fund fell 8.7% on the week, six points worse than the index; it is now 23% below its June peak. The mega-cap fund held up, down 1.4%, because Amazon, Microsoft and Alphabet rose on the week. The ugly: the chip complex. The semiconductor fund is down 31% from its June high and the memory names down more than 60%. We hold no chip sleeve; our exposure is through the index and momentum funds, which is where this week's cost came from.

Does the breadth move us? A little, and in the right direction. Breadth enters the growth cycle as the year-on-year change in the equal-weight-to-cap-weight ratio, and it strengthened again. It is a tailwind for the growth score. It does not and should not change the cut, which reads credit, not stocks. A broadening market and a widening credit spread can both be true, and this week they were.

What all of this did to the gauge. Same regime, Goldilocks. The growth score barely moved, at about +0.49. The monetary score eased from −0.79 to −0.69 on the lower two-year. The positioning score moved because the VIX jumped from 19 to 26; in this model a scared options market is a contrarian positive, so it added to conviction. Conviction rose from +0.05 to +0.11, still inside the band the model calls neutral. Had there been no cut, the targets would have moved by a point or so. There was a cut, so the gauge's reading is recorded and overridden.

What it did to the books. The books are marked to Friday's close and then rebalanced to the cut targets, so the ledgers below show both the week's loss and the new, defensive mix. The week's cost came from the US sleeve, where the momentum fund's 8.7% fall is about a third of a 48% sleeve, and from commodities, down 6% on oil. Gold offset part of it. The ledgers below carry the exact figures, dated, and the "since last rebalance" line in each shows the week against the S&P. Momentum has now cost the book in three of the last four weeks; the gate that holds it in the sleeve runs on a published register and has not flipped. That is a question for research, not a hand adjustment.