‘It is not what you do for your children, but what you have taught them to do for themselves, that will make them successful human beings.’ Ann Landers
TL;DR
The index rose 0.49% and eight of eleven sectors fell. Russell 2000 down 1.51%, and almost the entire gap opened in Friday’s session.
Warsh’s Jackson Hole debut did the whole week in a morning. Two-year up ten basis points to 4.34%, 2s10s from +50bp to +39bp, September hike odds from roughly one in three to about sixty percent.
The long end went the other way. Thirty-year down five to 5.22%, in the same week speculators put through the largest weekly net sale of ultra-long futures the contract has recorded, with the thirty-year swap spread at its tightest since February. Record shorts, and the price went up.
Everything real got sold. Gold down 3.42%, crude down 3.67%, copper down 0.80%. Credit tightened through all of it, high yield up 0.16% and investment grade up 0.41%. Nat gas up 3.40% on Qatari LNG.
Financials led at +1.08% and printed a 52-week high on the most hawkish session of the cycle. Nothing new opened on the book. Two lines came off by rule.
**The front end is the whole trade now and the long end has a buyer.**Bear flattening continues into payrolls, 2s10s toward +25bp. Leadership stays with fee-and-float and balance sheet.**Read Friday as curve management, not tightening.**Record short interest in ultra-longs and the thirty-year still rallied five basis points. Someone wanted the back end pinned.**Gold down 3.4% is the loudest thing nobody owns.**Either the debasement trade is finished or it just got handed its best entry since June. Both cannot be true.**The book has no long-end hedge and no vol hedge.**Two lines came off by rule this week. I carry that exposure open-eyed into a payroll print.
Every wire wrote a week about a hawkish Fed chair. Fine, as far as it goes. But hold the whole curve up to the light: the two-year rose ten basis points on the week, the ten-year fell one, and the thirty-year fell five. A hawkish repricing does that to the front end. It does not do that to the back end. And it certainly does not do it while speculators are dumping ultra-long futures at the fastest weekly pace ever recorded in the contract (Bloomberg), with the thirty-year swap spread grinding to its tightest since February. Record selling, and the price went up. Somebody with a very large balance sheet and no P&L constraint was on the bid, and they have been telling you so on television for a fortnight. Robin Brooks called it before the speech and again after: the keynote was about anchoring the long end, and the market is pricing hikes that the exercise was never designed to deliver. If he is right, the street is positioned for the wrong thing, and the trade is not to short the belly but to fade anyone who still believes the thirty-year is free to go where the arithmetic says it should. A market that cannot set its own long rate is a different animal from the one everybody modeled in July.
The honest objection, and it is a good one, comes from TS Lombard: you cannot fight the market forever. The buyback operations run at roughly three a month with the maximum doubled to four billion dollars, which is nothing against the size of the Treasury market, and the original Operation Twist moved yields for a few months before fundamentals reasserted themselves and the effect washed out. Their conclusion is that any cap here is temporary and the real story is a macro regime still not fully priced. I accept the mechanism and disagree on the horizon. A cap that lasts one quarter is still a cap you can trade, and my sheet is written in weeks.
The five sessions were nearly a non-event until they were not. Monday through Wednesday traded a narrow band while the index ground up on negative breadth. Wednesday’s core PCE at 3.3% on the year landed inside consensus and removed the September cut from the price. Thursday belonged to a chipmaker. Friday belonged to the Chair.
RATES. Two-year 4.34%, up ten basis points. Ten-year 4.73%, down one. Thirty-year 5.22%, down five. 2s10s from +50bp to +39bp, 2s30s from +103bp to +88bp. Bear flattening led from the front, and all of it arrived in a single morning. The published band for this framework sits at +45 to +55bp. We are through the bottom of it, and the front-end move was the largest single repricing across any asset class on the day.
EQUITIES. S&P 7,711.76, up 0.49%. Nasdaq 100 up 0.42%. Russell 2000 at 2,972.37, DOWN 1.51%. Read that pair again. The large-cap index made money in a week the average stock lost it, and the divergence opened almost entirely on Friday, when small caps dropped 1.4% in a session. Abroad, the dispersion was wider still: Japan up nearly two percent, Germany up more than one and a half, France down one. A single Fed speech sorted the entire developed world by duration sensitivity.
CREDIT. High yield up 0.16%, investment grade up 0.41%. Spreads tightened through a hawkish repricing, with high-yield OAS around 269bp and IG in the low 80s. Credit did not flinch. Whatever Friday was, the market did not read it as a growth accident.
FX. The dollar bid across the majors, its best session of the month. Euro at 1.1587, down 0.62% on the day. Dollar-yen at 160.06, up 0.43%, and the yen has now given back most of what a reported ninety-eight billion dollars of joint intervention bought it. When that much official money buys that little durable move, the price is telling you the flow is structural.
COMMODITY. Gold down 3.42% on the week, and down 3.24% on Friday alone, with a seventeen-dollar range in the ETF inside one session. Crude down 3.67% with Brent near 88 at its last settle, giving back the entire prior-week war spike while physical transits through the strait stayed severely impaired. Copper down 0.80%. Natural gas UP 3.40%, the only real asset that worked, and it worked because Qatari LNG exports are running down about ninety-six percent six months into the conflict.
VOL. VIX weekly high 15.85 Monday, low 14.13 intraday Friday, close 14.43. Down 4.6% on the week. And yet volatility ETFs took in another sixty-seven million dollars, a fifth consecutive week of inflows. Somebody is hedging while the index says nobody is.
FLIGHT CHECK: FAILED, BOTH WAYS. Nothing behaved like a haven. Gold sold, credit tightened, bitcoin fell nearly thirteen percent and again failed to be anything other than a levered risk asset, and vol went down into an event. There was no flight. There was a repricing of the cost of holding dollars at the front end, and everything that trades off a discount rate paid for it. A positioning event dressed as a policy event, and positioning events run until the position is gone.
THE REGIME CALL, WEEK AHEAD. Bear flattening continues into the 4 September payroll print. 2s10s toward +25bp. Equity leadership stays where money is made from spreads and fees rather than from duration and capex. Real assets stay heavy while the dollar is bid.
I’M WRONG IF 2s10s reclaims +50bp on a close, or the thirty-year takes out 5.31% with the two-year unchanged. Either says the long end is back in charge and this whole frame is upside down.
The prior edition called bear flattening led from the front end, with the two-year going from 4.24 toward 4.40 and the thirty-year pinned in a 5.15 to 5.35 range the Treasury had told you it would defend. The two-year closed 4.34. The thirty-year closed 5.22, inside the range, having touched 5.17 Tuesday. Sector of the week was financials, up 1.08% against an index up 0.49%. All three worked, and they worked for the stated reason, which happens rarely enough to be worth saying out loud. Verdict: correct, and the range call on the long end is the part I would press.
NVIDIA, AND THE ROUND TRIP IN A DAY. Up 8.74% Thursday on a print that doubled revenue year on year. Down 4.57% Friday, from an intraday 229.26 to a close of 217.55. Net for the week, +1.32%. The market bought the quarter and sold it before the ink dried, which tells you the marginal holder is a renter, not a believer. Nothing in the numbers changed between Thursday’s close and Friday’s. Only the discount rate did. A story stock that trades on the two-year is not a story stock anymore, it is a long-duration bond with a fab attached.
PAYPAL, DOWN FOURTEEN PERCENT as a fifty-billion-dollar take-private collapsed. One deal dying is a story. A fifty-billion take-private dying in the same week announced bank M&A ran at a seven-year high tells you the financing market is wide open for balance sheets and shut for platforms. Worth remembering when the next AI credit tries to syndicate, and there is a great deal of AI credit queuing.
CITADEL SECURITIES, RECORD QUARTERLY TRADING REVENUE OF $7.3 BILLION, more than triple a year ago, with net income up over 250%. Put that next to the fee complex lagging on price for a third straight week and you have my financials sleeve in two numbers. Somebody is being paid enormously well for dispersion. The listed vehicles that do the same thing have not been paid at all.
Eight of eleven sectors fell. The index rose. Everything about the week’s internals sits in that sentence, and dispersion inside the sectors was wider than dispersion between them.
FINANCIALS, +1.08%. Best of the eleven and a fresh 52-week high, printing 58.41 intraday on the most hawkish session of the cycle. Money centers led, insurance held, regional banks fell 0.75%. Higher-for-longer is being paid as a deposit franchise subsidy rather than charged as a funding cost, at least while deposit betas stay lagged. Positioning: own the sort, not the sector. Long the fee-and-float complex against the spread lenders.
ENERGY, -1.51%. Crude gave back the whole war spike, and here is the part worth sitting with: it did that while five commodity vessels transited Hormuz on 25 August against a ten-day average of fifteen, and while August crude exports through the strait ran near 2.3 million barrels a day against 4.49 million in July and 15.82 million before the conflict. Physical flow is still severely impaired and the price fell four percent anyway. A discount-rate week, not a supply week. Then Friday night brought an announced deal for majority US control of 65 billion barrels of Venezuelan reserves (New York Times). Positioning: no new length in crude beta until that supply story has a price.
TECHNOLOGY, +1.30%. Second best, almost entirely semis, with software and hardware both lagging. Rich Ross at Evercore made the call that explains the whole board: a flatter curve catalyzes momentum and semis, exactly as the July steepening drove their unwind. Leadership this week was a curve trade wearing a growth costume. Against it, roughly a third of second-quarter index EPS growth came from AI-infrastructure names, and aggregate earnings looked far broader than they were once investment gains are stripped out. Positioning: hold, do not add, and watch the credit market rather than the earnings calls.
COMM SERVICES, +1.43%. Top of the board. Mega-cap platforms carried it while telco stayed inert and media did nothing anyone will remember. A sector that is now three companies and a rounding error. Positioning: the cleanest expression of large-cap concentration, and concentration is exactly what is being questioned.
CONSUMER DISCRETIONARY, -0.69%. Retail split violently on results, with one apparel name up forty-two percent on guidance and a short base. Underneath the beats: July retail sales fell 0.6%, the first decline in nine months, confidence printed a seven-month low at 89.4 and sentiment sat near 51. Autos flat, travel soft, homebuilders hurt by a front end that just repriced. Positioning: the subsector story is housing, and housing needs a two-year below 4%, which is not on the menu.
CONSUMER STAPLES, -0.63%. No defensive bid, in a week with plenty to be defensive about. Pricing power is exhausted, which is my ag-against-packaged-food trade in one line. Positioning: stay short the packaged names against input cost.
HEALTH CARE, -1.98%. Worst of the eleven, giving back a huge prior week. A failed colorectal trial took one large biotech down eleven percent while a competitor collected its fifth upgrade on melanoma data. Add the largest patent cliff in decades arriving now. Positioning: no sector view, and I would rather say that than manufacture one.
INDUSTRIALS, -1.72%. Second worst. Capital goods soft, transports softer, defense held on a two-hundred-fifty-billion-dollar procurement authorization moving through Congress. A sector needing cheap financing lost a week where financing got dearer at the front. Positioning: underweight until the front end stops moving.
MATERIALS, -0.67%. Copper down 0.80%, chemicals mixed, packaging flat. Quiet week, genuinely, and the number is the whole story. Nothing to see.
UTILITIES, -0.09%. Effectively unchanged while the front end moved ten basis points, which is remarkable. The AI-load story is now doing real work as an offset to rate sensitivity, though a Californian wildfire-cost headline took one large name down twelve percent intraday. Positioning: I am short at half size and this week made the case for keeping it at half.
REAL ESTATE, -1.33%. Office still broken, residential hurt by the front end, data centers the only thing anyone wants, retail REITs ignored. Cap-rate math needs a lower discount rate and did not get one. Positioning: the cleanest short against a pinned long end that refuses to fall.
FINANCIALS. Long the fee complex, short the spread lenders.
The sector printed a 52-week high in the week the September cut died and a first-time Chair called inflation concerning. Consider how odd that is. The consensus read of a hawkish surprise is that it is bad for banks: funding costs up, credit risk up, multiples down. Friday delivered the exact opposite cross-section, and the mechanism is the deposit franchise. If the policy rate stays high and deposit betas stay lagged, spread income holds at levels the curve says should be eroding.
Look one layer down and the sector is arguing with itself. Insurance up 0.79%. Brokers, exchanges and IB up 0.31%. Regional banks DOWN 0.75%. Three weeks running, the fee-and-float complex has lagged the balance-sheet complex, and three weeks running the fundamental data has gone the other way: first-half advisory volume through a trillion dollars, August investment-grade supply at a record, capital-markets revision baskets up thirty percent, and the largest private market maker printing its best quarter ever. Fundamentals improving while price lags is either an opportunity or a warning. The honest answer is that a three-week cross-section in this sector is noise, and I am sizing as though I believe that.
THESIS Higher-for-longer is a deposit subsidy, not a funding threat, while the fee pool re-rates on volume a no-guidance Fed manufactures for free.
I’M WRONG IF Regional banks outperform the sector two consecutive weeks with 2s10s below +40bp, which would mean the market is paying for spread rather than for fees and the entire sort is inverted.
Look at the divergence chart, then ask who is on the wrong side of it. Active managers sit at their longest since July 2024 with fund cash under one percent, into a month that has punished every recent midterm year. Small caps just lost 1.5% in a week the index gained. Concentration has not broadened, it has narrowed: nearly half the S&P remains AI-related, the mega-cap complex is above thirty percent, and the major indices have scattered their all-time highs across four separate dates. PauloMacro has tracked that scatter since February and reads the current version as unhealthy tape rather than healthy rotation (** his note is here**). Hard to argue with the arithmetic.
The institutional flow is already moving. BlackRock and Aviva have b…
