← Paulo Macro hub  ·  Research hub  ·  Research library

Ramblings & Ruminations: Instability, Soft Patch Bonds, New Trade Idea

The long note promised in the morning chat, in three parts. Instability: implied correlation has "never been this low" and the implied-vs-realized compression is tighter than pre-Yenmaggeddon July 2024, protection is cheap because "there are more sellers than buyers," and non-dealer equity futures length is back near pre-Liberation-Day — "I am strapped in with downside protection." Bonds: the Citi Surprise index confirms the 3Q soft patch, and "the 5Y and 10Y future have both flushed the July low, but the 2Y and long bond have not (yet)" — "the long bond is not confirming the War trade," so he keeps his bond calls. New trade idea: palladium, on positioning alone — speculative length is "rinsed back to ~$1,100/oz Aug2025 levels, but price is ~$200 higher" — explicitly at idea stage, with no vehicle named ("please don't ask me how to play it").
2026-SEP-01 · PauloMacro (Substack, PAID) · written post — chart-heavy, no timestamps · ↗ Read original · transcript · actionable insights
One-line take: the long-form companion to the same-day morning chat note — it takes the chat's two teasers (correlation/dispersion, the bond divergence) and adds the evidence, then bolts on a brand-new idea at the front of the funnel. The first section is an instability inventory: Warren Pies at 3Fourteen showing "implied correlations have never been this low and therefore vulnerable to a macro shock"; near-dated implied correlation having "flipped above 3mth… something we see as market stress starts to rupture"; a chart of his own (inspired by Kevin Muir) measuring 3mth implied minus realized correlation, whose compression "was even tighter than pre-Yenmaggeddon in July 2024"; VIX-3m vol "starting to tick up… inversion coming in the volatility surface?"; and non-dealer equity futures positioning "approaching pre-Liberation Day length again." The conclusion is a position, stated flatly: "I am strapped in with downside protection." His own counter-argument is honest — "I would normally say that the sudden visibility by several observers that we may be in a window for weakness means it won't happen" — answered by the price of protection: "nobody seems to own protection (otherwise why would it be so cheap…)." Section two revisits the soft patch: "The Citi Economic Surprise Index seems to confirm this as the data has been progressively missing since June (note Japan rolling over now as well — Europe next?)," and the curve non-confirmation gets its exact form: "the 5Y and 10Y future have both flushed the July low, but the 2Y and long bond have not (yet)." He maps two war scenarios rather than picking one — a March redux ("Stocks down, Bonds down/Yields up, Gold down, USD up, Oil up") versus a different Part Deux ("Stocks down, Bonds up/Yields down, USD down, Oil up") — and lets the tape referee: "despite oil rallying $5 on the latest Iranian 'love taps' this afternoon, the long bond is up a mere 2bps." With CTAs short bonds — "a bug in search of a windshield" — and payrolls Friday, he keeps the calls: "Hard trade." On gold he is unusually candid: "I have no strong view, but feet to the fire and I am biased lower," because managed money is "more net long on COMEX than at any point in history on an outright notional $ basis." And the new idea, palladium, is deliberately unfinished: "I'm not doing anything here yet and have barely begun the work… please don't ask me how to play it."

Key points

Row-scope note. This is a pure macro / positioning note and Paulo names no company, fund or ETF as an investment — the load-bearing objects are implied vs realized correlation, dispersion, options skew, the VIX term structure, equity-futures positioning, Treasury note and bond futures (2Y, 5Y, 10Y, long bond), the Citi Economic Surprise index, gold, oil and palladium. Per this hub's convention, rates, FX and commodities are not tabled as tickers unless he names a vehicle, so there is no stock table on this page — nothing is invented. Palladium is deliberately not tabled here, even though the next day's chat note carries it as PALL: that page tabled it because "get long sooner than I thought" was an actionable position change, whereas this post is the origination and explicitly refuses an instrument — "I'm not doing anything here yet and have barely begun the work… please don't ask me how to play it because… I haven't done all the work." The palladium view therefore stays on the index through the Sep-02 PALL row, not through a second row invented here. Firms and writers cited as chart sources — 3Fourteen (Warren Pies), The Macro Tourist (Kevin Muir), Citadel (Rubner), Goldman Sachs, Barclays, "my pal JH" — are attributions, not securities, and are not tabled. The roughly a dozen embedded charts are images and are not reproduced in transcript.txt; the text carries the argument. Note on dating: this is the longer "note later today" promised in the 2026-09-01 morning chat note — two distinct pieces published the same day, which is why this folder carries the -ramblings suffix.

The obsession that hasn't paid yet — criticality, and the Dornbusch excuse for it

His own best counter-argument — "everyone suddenly sees it"

Implied correlation — "never been this low," and the shock vulnerability that implies

The term structure of correlation is already flipping

His own gauge — implied minus realized correlation, tighter than pre-Yenmaggeddon

Why protection is cheap — more sellers than buyers, and flat skew

Pent-up energy — VIX/correlation instability and positioning back at pre-Liberation-Day

The soft patch, confirmed — and now spreading

The bond divergence, stated precisely — and it is both ends that refuse

War Part Deux — two cross-asset maps, and a refusal to pick

Gold — "tricky," and honestly unresolved

CTAs, payrolls and the Warsh constraint — "a bug in search of a windshield"

New trade idea — palladium, on positioning alone

Palladium vs platinum — the worse asset as the higher-beta expression

In plain English

Part 1 — why he is paying for insurance he has not needed

Two things can be measured about a stock market beyond whether it is going up. Correlation is how much shares move together; dispersion is how far apart the winners and losers end up. Right now correlation is at a record low — shares are wandering off in their own directions — and that is usually reported as a healthy, stock-picker's market.

Paulo reads it the other way round. If shares are not moving together, a portfolio full of separate bets looks safe on paper, so investors are allowed to take more of them; and insurance against everything falling at once looks unnecessary, so it gets cheap. The danger is that this is reversible in an instant. One big enough shock and everything moves together again — and then the portfolios that looked diversified turn out to be one bet, and the insurance nobody bought is the only thing that would have helped.

He offers three pieces of evidence that the market is not merely calm but unusually calm. Option prices imply a level of independence between shares that has never been recorded before. The gap between what options prices assume and what is actually happening is more compressed than it was just before the August 2024 "Yenmaggeddon" — a day when a crowded currency trade unwound and volatility exploded. And the shape of the option market has changed: normally, protection against a fall costs noticeably more than a bet on a rise, because people fear losses more than they crave gains. That premium has flattened out, which means today's buyers are more afraid of missing a rally than of a crash.

His conclusion is a purchase rather than a prediction: "I am strapped in with downside protection." He is not saying a crash is coming this month; he is saying insurance is on sale, and quoting Bernard Baruch — buy straw hats in winter. The honest weakness in the argument, which he states himself, is that the same worry has now been voiced publicly by several well-known strategists, and a worry that everyone shares usually stops paying. His answer is that talk is cheap and positions are not: if people really were afraid, protection would not still be this cheap.

Part 2 — the bond market that will not play along

The US government borrows over different lengths of time — 2, 5, 10 and 30 years — and each has its own futures contract. When something genuinely big changes in the economy, all of them tend to move together. What Paulo has spotted is that they have not: the 5- and 10-year contracts have broken below their July lows, while the 2-year and the 30-year have held. A break confined to the middle of the curve is the signature of traders being squeezed out of positions, not of the world's view of growth or inflation changing.

He gets a clean test of that the same afternoon. Oil jumps $5 on renewed Middle East attacks. If bonds were trading the war — war means higher oil, higher oil means inflation, inflation means higher long-term interest rates — the 30-year should have sold off hard. It barely moved. So the war is not what is driving bonds, and something else (he suspects a steady, unglamorous flow of money rotating into them) is quietly absorbing the selling.

That matters because Friday brings the August jobs report, and he has been arguing since June that the US economy would go through a weak patch this quarter — a claim the Citi Economic Surprise Index, which simply tracks whether data beats or misses forecasts, now supports. He notes Japan's version of that index has started rolling over too, and wonders whether Europe follows. Trend-following funds are positioned short bonds, which he thinks is the interesting part: they have little left to sell if bonds keep falling, but a great deal to buy if interest rates turn down. That lopsidedness is the trade — a friend calls the set-up ready for a catalyst; Paulo calls it "a bug in search of a windshield."

He then refuses to simplify it, which is the most useful thing in the section. A weak jobs number could push money into bonds. But it might also make the Fed less likely to cut in September for reasons he doesn't spell out, in which case short-dated bonds rally and long-dated ones fall — the opposite of a uniform rally. He is not resolving it. He is holding options on bonds, which lose only their premium if he is wrong, and waiting: "Hard trade."

Part 3 — palladium, at the idea stage

Palladium is a precious metal used mainly in the exhaust systems of petrol cars. Paulo's interest in it has nothing to do with cars, and — unusually — nothing to do with supply and demand either. It is entirely about who owns it.

In commodity futures markets you can see how speculators are positioned. Palladium ran from about $1,000 to $2,000 an ounce without speculators ever committing to the upside, and they have since swung back to a net short — betting on lower prices — at the same level of pessimism seen in August 2025, just before a big rally. But the price today is roughly $200 higher than it was then. That combination is what caught his eye: the same amount of disbelief, at a higher price. Markets that make higher lows while the crowd stays negative are, in his words, "what a bull market is traditionally supposed to look like."

He is scrupulous about the weaknesses. Palladium's fundamentals are worse than platinum's, which he already follows and owns. The market is so small that large institutions essentially cannot participate. And none of this stops the price falling further in the short run. His actual argument is a relative one: if precious metals turn and platinum leads, palladium is the smaller, more neglected, higher-octane version of the same move — "palladium is to platinum what silver is to gold."

The important caveat is that this is not yet a trade. He says outright that he has "barely begun the work," asks readers for help on how to express it, and refuses to name a vehicle: "please don't ask me how to play it." That is why no ticker is recorded against this note. The day after, in his Sep-02 chat post, the idea acquires a reason to hurry — Russia supplies roughly 40% of the world's palladium and he thinks that risk is unpriced — and it is there, not here, that the position first appears on this index.

Key points extracted from the PauloMacro Substack post of 2026-SEP-01 (saved in transcript.txt) for personal study. The post's embedded charts are not reproduced. No securities are named as investments in the note and none are inferred. Not investment advice. © PauloMacro for source material.