What We Told Hedge Funds This Week
Semi-hot CPI.. and the USD WEAKENED once the dust settled. The pain trade is lower USD, lower oil, higher equities
A semi-hot CPI: core at 0.29%, above the street’s 0.22% (and above our own nowcast), hot enough to clear every fence-sitter’s bar and make a September hike the base case. And then the interesting part: once the dust settled, the USD WEAKENED and risk assets went HIGHER. On the one print everyone said would decide the month. That is not a market that wants to reward the consensus positioning, and the positioning data tells you exactly why: the market is very, very long dollars and very, very short bonds. The pain trade from here is lower USD, lower oil and higher equities.. and it will surprise everyone.
Let us start where the week ended: the August CPI. Headline 0.40% MoM (3.35% YoY), core 0.29% MoM (2.45% YoY). Semi-hot, in other words: above the 0.22% consensus, and above our own nowcast too, which had the headline right (0.36% versus 0.40%) but ran too low on core as some of the energy pass-through leaked into the core categories.
Look at the details: transportation services at +0.45% (jet fuel into airfares.. that is the crack spread showing up in core), used cars and trucks at +0.37%, shelter at +0.26%, and a rent-of-primary-residence at 0.17% that was actually on the soft side. Medical care services at -0.25% and apparel at 0.02% were soft as expected, and the energy complex did exactly what we said: energy commodities +4.24%, gasoline +3.90%, fuel oil +10.15%. So the composition was as expected, the level a touch hotter.. and the level is what the fence-sitters trade on.
And that level is decisive for September. We wrote it on Monday and said it on the State of the Union: if monthly core prints above 0.2%, we get a hike in September, and they do not much care whether it is 0.24 or 0.29. Williams gave the market 0.2% as his line, Waller said “another hot reading” and he considers tightening.. 0.29% clears both bars with room to spare.
So the fence-sitters have their number, and the September FOMC is a hike as the base case now, joining the RBNZ and the ECB (which hiked again yesterday) in the club. Remember the framing: for as long as the oil-products situation is unsolved, central banks with a HEADLINE mandate have a hiking bias.. they simply have to. What they do not have to do is commit to a path, and that is the template we expect from Warsh next week: the RBNZ-style dove-hike, a move with an unchanged track, because everyone in that room knows that one memorandum of understanding puts us straight back in disinflation territory.
Now to the part that actually matters for the book, because the market REACTION was the real news of the day, and I want this to be crystal clear. A semi-hot core print that clears every hawk’s threshold, a hike repricing into the base case.. and once the dust settled, the dollar was WEAKER on the day, equities were HIGHER, and oil was softer.
The knee-jerk went the textbook way for about half an hour maximum; the close went the other way. Bonds sold off at the front, sure, but the long end barely blinked. That is not a muted reaction, it is a BENIGN reaction to a hawkish print, and it signals one thing: POSITIONING. The market is already VERY, VERY long the dollar and VERY, VERY short bonds.
Look at the CFTC data in the 2yr: the short position is a monster, north of two million contracts at the peak, against a long book a third of the size. Look at the USD positioning as a share of open interest: crowded long, and the dollar has been UNDERPERFORMING its own positioning for weeks. And look at EURUSD against the reversed USD positioning: the euro has been remarkably resilient while the specs have piled into dollars against it.
When a market fails to move in the supposed direction on news that clearly supports that direction, it is typically because everyone who wanted to be in the trade is already in it.. and that is what turning points look like from the inside. So let me spell out the pain trade from here, because it is the mirror image of how the street is positioned: LOWER USD, LOWER oil and HIGHER equities. Everyone is long dollars, everyone is short bonds, everyone has been told that hot inflation plus a hiking Fed means a strong dollar and weak risk. Today the market took that consensus, priced the very print it was waiting for, and then went the other way. It will surprise everyone.. which is precisely why it is the pain trade.
And here is a thought that goes against the grain: a September hike might actually HELP the long end. The whole Bessent project of the past month.. the doubled buybacks, the TGA story, the eSLR reform, the coordinated JPY intervention.. plus Warsh’s “tough talk” in Wyoming has so far proven INSUFFICIENT to tame the stress further out the curve. The 30yr round-tripped after the buyback announcement, the first enhanced operation (6bn, 3× the old cap) was received as a disappointment, and the term premium refuses to compress on words alone. A credibility hike changes that calculus: it is the classic bear-flattener, where the front end takes the pain and the long end gets to price a central bank that is actually fighting the inflation the market fears. If Warsh hikes into a monster-short bond market with an unchanged track, the long end rallies on credibility and the front end squeezes on positioning.. and both of those are constructive for duration further out the curve. Which is why we are turning constructive on long-end bonds (more below).
Meanwhile, Bessent is not done.. he has barely started. Yesterday’s 6bn was the 10-20yr sector; the 20-30yr operation is still to come, and it is the more obvious candidate for a larger step-up given that is where the stress actually sits. The firepower argument has not moved an inch: the TGA sits around 937bn with a practical floor near 500bn, so even a conservative 250bn drawdown funds 16 months at 6bn per operation and more than nine months at the 10bn I have been arguing for.. three full quarters of QE-shaped flow from the Treasury building. Starting at 3× was, in hindsight, smart: it keeps the powder dry and lowers the bar for the next operations to surprise on the upside.
And remember why it is USD-negative: every TGA dollar that funds a bond purchase lands as a fresh bank reserve, every bank reserve eases the SOFR complex, and a Treasury that is explicitly trying to weaken its own currency to bring foreign buyers back (Bessent’s own words this week: do not bet against him, he has asymmetric information) does not stop after one operation. The doom loop of intervention-selling-USTs has been turned into a positive loop where a weaker dollar means Japan does not have to sell.. that is the most important macro development of the month, and a 6bn opener does not undo it.
In Focus: The regime shift.. and why it is a TIGHT call
Our regime engine flipped this week, in the US and in the World composite: from QE Like to Lame Duck QE.. weakish growth, sticky-to-rising inflation, still-easy liquidity. The US Lame Duck probability went from 6% a month ago to 28%, overtaking QE Like at 27%, while Gung Ho collapsed from 22% to 9%. The driver is exactly the scissors we have described for a fortnight: the US rising-inflation probability has crossed 50% (from 13% a month ago), growth has slipped to 25%, liquidity is unchanged at 72%. Today’s CPI, if anything, hardens the flip.
And the pattern recognition in that regime is uncomfortable for our macro legs, so we wrote it down honestly this week. Historically, Lame Duck QE is a USD-UP regime (USDSEK at 2.6× the base odds of a positive move, USDJPY 2.5×, EURUSD 0.74×), a duration-DOWN regime (US 10yr 0.58×, Bunds 0.39×), and a gold-and-oil-UP regime (2.1× and 2.0×). Cross that against the book and you get four conflicts on the macro side (short USD, received front end, no energy long, the Bitcoin leg) and five confirmations on the micro side (hardware and memory at 1.57×, utilities 1.6×, uranium at a screaming 3.45×, gold 2.06×, and the software underweight). The regime engine likes our stock-picking and dislikes our macro overlay.. useful to know about yourself.
But here is the caveat that matters, and it is why we have NOT flipped the book: a regime shift in our nowcasting means the probability of one regime has increased relative to the others.. it does not mean the regime is locked in. Lame Duck QE sits at 28%, QE Like at 27%, three other regimes in double digits.
That is a very TIGHT call between the USD-positive regime (Lame Duck) and the USD-negative one (QE Like), and it makes it hard to flip anything with conviction on the shift alone. Now layer on what we learned today: the hawkish print that should have been rocket fuel for the USD-positive regime produced.. nothing. The dollar ended the day WEAKER into the most crowded long positioning in years.
Add the de-escalation optionality (Mikkel’s read is that we could be within weeks of the US taking its foot off the pedal in the Gulf, even without a grand peace deal, and Trump has now explicitly pushed the war timeline past the midterms.. which in this sine-wave conflict has meant lower oil every single time), and the case for keeping a short USD LEAN survives the regime shift.
Portfolio update: A solid week.. the rotation is ON
A genuinely good week for the book: +9.4% YTD versus +6.9% a week ago, so roughly two and a half points added in a week where our biggest macro lean was under pressure and the CPI went against our nowcast. The resilience came from exactly where we said it would in the State of the Union: hardware. Micron at +253% on entry, Dell +133%, HP +61%, CrowdStrike +76%, SentinelOne +35%.. and SanDisk clawing back from -19% to -14% as the memory complex caught the bid.
This is the rotation we called: the value gap between semiconductor equities and the live Korean export data is not 50% anymore, it is north of 100% if you take the chart literally, and it is the exact same shape of gap we had in March and April when we said “now it is time for semiconductors”. NAND prices are up 40% in a month. A couple of the memory names have broken out of wedges. And the cleanest-shirt logic holds: it is not an issue for Micron or SanDisk whether diesel trades at 100 or 200, to put it bluntly.. their earnings cycle is SO solid short-term that you get bang for the buck here and now, which is the characteristic you want when the world turns uncertain. The options market agrees, in its own way: semis versus software implied vol sits at parity, the market paying NOTHING extra for hardware optionality, which is what a low in the hardware-versus-software cycle looks like.
And the broader valuation backdrop backs the cleanest-shirt call: only 27 stocks in the S&P 500 trade above 40× forward earnings, the same count as at the 2020 Covid low and the 2022 bear-market low. Earnings are HIGH, also in forward terms, and the market is treating them VERY carefully.. paying less for record earnings by the month.
We stay long hardware, underweight software, tilted US over Korea into the licence decision (where, as we said on the show, the incentive structure for a NON-renewal is stronger than the market thinks: Micron and SanDisk can walk into the White House and ask why their Korean competitors get to keep a Chinese supply chain when they do not).
First, the short USD lean: it has done remarkably well since we turned dollar-negative in July, and it did so despite higher energy prices that SHOULD have strengthened the dollar through the terms-of-trade channel.. we have had hiccups over the past two weeks and today’s print was one of them, but the fact that the dollar cannot strengthen with everything going its way is the strongest confirmation we could ask for that the market is over-positioned for dollar strength. The lean stays.
Second, ENERGY: we are not overly tempted to buy it here, even after a good July and August that we mostly missed (we abandoned the short energy lean with good timing in July but never got on the long train). The refining picture is genuinely dire.. global refining capacity online is the lowest in modern history at 89.6% of nameplate, with three fronts open at once.. but China is back in the refiner’s seat, escalation has been a SELL signal all year, and the de-escalation optionality is real. We hold sympathy for SOME energy as a hedge against the world burning, and the regime engine now underlines that in red, but it is a hedge, not a conviction.
Third, BEAM Therapeutics, which is down 30% on our entry and was the week’s ugly line.
What happened: on Tuesday Beam presented updated Phase 1/2 data for BEAM-302, its one-time base-editing therapy for alpha-1 antitrypsin deficiency, at the ERS Congress.. and the stock gapped from ~30 to ~26 and kept sliding. The trigger was a safety signal: some patients showed elevated liver enzymes, and data from certain dose cohorts were excluded from the analysis, which the market read as a yellow flag on a liver-targeted therapy. Add the broader rotation out of small-cap biotech (health care is the WORST sector in the Lame Duck QE playbook at 0.47×, so the regime is fighting this one too), some insider selling, and a classic sell-the-news reaction to data that, on the efficacy side, actually supports an ACCELERATED approval path.. and you get a 10% gap.
Now the bull case, which has not changed: BEAM-302 is a first-in-class one-time correction of the root cause of AATD in BOTH the liver and the lung, the 29-patient safety profile was otherwise described as acceptable (mild-to-moderate infusion reactions being the most common event), JPMorgan kept its Overweight calling the liver signal manageable, Wainwright reiterated Buy with an 80-dollar target, and the risto-cel BLA path in sickle cell remains the second shot on goal. Liver enzyme elevations in a liver-directed gene therapy are a known, monitorable class effect, not a novel toxicity.. and the accelerated-approval path is the prize. We hold. It is a 1.7% position sized for exactly this kind of binary noise, and we do not add biotech beta into a 0.47× regime reading.. but we are not selling a platform story on a transient enzyme signal.
Trades and themes across asset classes
In FX: short USD as a lean, unchanged and quietly REINFORCED by today’s non-reaction. The most hawkish print in months left the dollar WEAKER on the day against a crowded long positioning, Bessent is explicitly working against it, and the regime call between USD-positive and USD-negative is a coin toss. EURUSD towards 1.2050 stays the target; SEK and NOK legs off the table (the most Lame-Duck-exposed), JPY via the intervention channel stays.
In Rates: a genuine SHIFT this week. The received US front end is now a September-meeting trade only: a hike is the base case and the front end is where the pain lands, so we size it down to a lean and stop calling it a conviction. But we are turning CONSTRUCTIVE on long-end bonds. A credibility hike into a monster-short bond market, with Bessent ramping the 20-30yr buybacks from a TGA that funds 16 months at the current pace, is a bear-flattener that rallies the long end.. and today’s muted long-end reaction to a hot print tells you the sellers are exhausted. Own duration further out the curve; let the front end wear the hike.
In Commodities: gold over Bitcoin within the debasement sleeve (2.06× versus 0.76× in the regime engine, and the TGA flow is the tailwind for both but gold is the cleaner expression). Short crack spreads stays, with eyes open.. three refining fronts, but China is back buying and running its refiners harder, and escalation has SOLD the barrel all year. Some energy as a portfolio hedge is now warranted rather than optional. Not chasing flat oil.
In Equities: BUY HARDWARE, underweight software.. the value gap to Korea, the implied-vol parity, the 27-names-above-40× backdrop and the earnings cycle all point the same way. The memory complex on top (Micron, SanDisk, tilted US over Korea into the licence cliff and the summit), the electricity sleeve as the energy-squeeze winner (Bloom +193%), the End-Consumer Scarcity trio (Dell, HP, Apple) as dollar-priced hardware immune to the TTF curve, the AI-on-top software rail (CrowdStrike, SentinelOne) keeps its earned sleeve. Definium at 13.5% remains the biggest position into Panorama, Beam is held through the noise, and no added biotech beta while health care sits at the bottom of the regime table.
Have a great weekend, and good luck out there!
