“Committed to a Discipline, Not to a Decision” — Warsh Sends September Hike Odds to 57%, and Chicago PMI Collapses to 47.1
Data:
Main Theme: “A Hawkish Chair and a Broken Regional Survey” — Warsh said financial conditions do not look restrictive and that better inflation readings have not convinced him the trend is improving, sending September hike odds from 35% to 57% and the two-year yield up 12 basis points. Buried underneath, Chicago PMI collapsed to 47.1 against a 57.9 consensus — the weakest reading of 2026.
Friday delivered the Fed communication the market had waited a week for, and it was hawkish without being explicit. Chair Kevin Warsh used his first Jackson Hole keynote to avoid both forward guidance and a reaction function, telling the audience: “I stand here today committed to a discipline, not to a decision.” But the substance was unambiguous — he described an economy showing few signs of restraint from the Fed’s current rate of around 3.6%, said financing conditions did not look restrictive to him, and stated that “while this summer’s readings were better than expected, they do not tell me that underlying trends have meaningfully improved.”
Markets repriced hard at the front end. Fed funds futures moved to 57% odds of a hike at the 15–16 September meeting, up from 35% the day before. The two-year yield jumped roughly 12 basis points to 4.35% and the ten-year rose 5.6bp to 4.728%, with the 30-year back at 2008 levels. The dollar index gained 0.5% to 99.65.
Equities gave back an early gain: the Dow finished roughly flat at −0.02%, the S&P 500 fell 0.25% and the Nasdaq Composite dropped 0.52% after being up as much as 0.5% earlier. All three still closed the week higher, though by less than 1%.
🟥 U.S. Equities | An Early Gain Given Back
| Index | Friday | Week | Session Stance |
| Dow Jones Industrials | −0.02% | Higher | Essentially flat |
| S&P 500 | −0.25% | Higher | Turned lower after the speech |
| Nasdaq Composite | −0.52% | Higher | Had been up 0.5% earlier |
| All three indices | — | Gained less than 1% on the week | Thanks to the midweek Nvidia rally |
Marvell Technology was the session’s notable single-stock decline, falling 8%. The company offered limited detail on its fiscal 2028 outlook, dampening sentiment after hopes that a Google partnership worth up to $12.2 billion in shares would further boost earnings. The stock remains up 184% in 2026, buoyed by demand for its products used in AI infrastructure.
The Marvell reaction fits the pattern that has governed this entire season. It entered the print with a 179% year-to-date gain and a 58x forward price-to-earnings multiple, and prior earnings reactions had swung from an 18% gain to a 20% loss. A solid quarter with vague forward detail was not enough for a stock priced that way.
🏦 Jackson Hole | What Warsh Actually Said
The speech was a statement of governance philosophy rather than policy signalling, and that was deliberate.
| Topic | Warsh’s words |
| On guidance | “I stand here today committed to a discipline, not to a decision.” |
| On market dependence | “We should not indulge a regime in which market participants are looking primarily to the Fed for their next trade.” |
| On recent inflation data | “While this summer’s readings were better than expected, they do not tell me that underlying trends have meaningfully improved.” |
| On the target | The 2% PCE target is “firm and fixed” — and if underlying inflation is not moving toward it fast enough, “we have work to do.” |
| On policy stance | Financing conditions “didn’t look restrictive”; an economy showing few signs of restraint from the current rate of around 3.6% |
He stopped short of saying whether he would support raising rates in September and offered no path for policy, consistent with his stated approach that the central bank should communicate less about coming moves. The initial market reaction was muted; the repricing came as traders digested the substance.
The critical line for policy purposes is that financing conditions do not look restrictive. That is a direct statement that the current setting is not doing enough work — and it echoes Kansas City’s Jeffrey Schmid, who said the same thing on Thursday, and Cleveland’s Beth Hammack, who spoke at 9:00am and had said “now is the time to act.” Chicago’s Austan Goolsbee, a 2027 voter, subsequently agreed with Warsh on the details of the economy and agreed inflation is the main issue right now.
The rates response was concentrated at the front end, which is the correct read of a hawkish-but-unspecified message:
- Two-year yield: up around 12 basis points to 4.35%.
- Ten-year yield: up 5.6bp to 4.728%.
- Thirty-year yield: back up to 2008 levels — Investrade noted this came “just as long-term yields were beginning to cool down.”
- Dollar index: +0.5% to 99.65, back near its 100-day moving average of 99.60 and above its 200-day at 99.17.
📰 Macro | The Number Nobody Talked About
Chicago PMI collapsed to 47.1 in August, against a 57.9 consensus and a 57.6 prior reading — the weakest print of 2026 and a miss of more than ten points.
This is a genuinely startling number and it got buried by the Warsh speech, which landed fifteen minutes later. It also sits in direct opposition to every other manufacturing indicator of the past fortnight:
| Indicator | Reading | Signal |
| Empire State Manufacturing (Aug) | 20.6 vs 11.0 expected | Highest since 2022 |
| Philadelphia Fed Manufacturing (Aug) | 47.4 vs ~25 expected | Strongest since April 2021 |
| S&P Global composite PMI (Aug flash) | 56.0 | 52-month high |
| ISM Manufacturing (Jul) | 55.6 | Highest since May 2022 |
| Chicago PMI (Aug) | 47.1 vs 57.9 expected | Weakest of 2026 — in contraction |
A reading below 50 indicates contraction. Chicago PMI is a regional survey with a heavy manufacturing and auto-supply-chain weighting, and it is more volatile than the national measures — but a ten-point miss into contraction, in the same month two other regional surveys hit multi-year highs, is not noise that can be waved away. Tuesday’s national ISM Manufacturing print is now considerably more important than it looked a week ago.
😟 Michigan | Revised Up, With the Inflation Line Revised Down
| Measure | Aug final | Aug prelim | Jul final |
| Sentiment | 51.7 (cons 51.0) | 51.0 | 55.2 |
| Current conditions | 51.9 (cons 51.8) | 51.8 | 54.8 |
| Expectations | 51.5 | 50.6 | 55.4 |
| 1-year inflation outlook | 4.0% | 4.3% | 4.2% |
| 5–10 year inflation outlook | 3.3% | 3.3% | 3.3% |
The revision matters and it cuts against the hawks. One-year inflation expectations were revised down to 4.0% from a preliminary 4.3%, and now sit below July’s 4.2%.
Newsquawk’s framing is the correct one: the final Michigan reading carries less weight than the flash, since the preliminary print does the repricing. What stands out is the step down from the prior month — 51.7 against 55.2 in July — rather than the modest beat against consensus. The index bottomed at a record low of 44.8 in May 2026, recovered to 49.5 in June and 55.2 in July, and has now given back most of that improvement.
The 3.3% five-to-ten year reading has not budged, but it remains above the levels that would signal expectations are truly anchored at the 2% target — which is precisely the concern Warsh described.
📊 The Payroll Benchmark Revision
The preliminary annual revision to the government’s payroll data was released at 10:00am ET — the same minute as Warsh’s keynote and the final Michigan reading.
I have not been able to confirm the 2026 figure from a primary source at the time of writing, and I am not publishing a number I cannot verify. For context, last year’s version came in at −911,000 jobs, meaning the labour market had been materially weaker than originally reported.
This matters because the July employment report already revised May and June down by a combined 103,000, taking the trailing twelve-month average of job creation to roughly 34,000 a month. A large benchmark revision on top of that would materially change the labour picture the FOMC works from — and it lands two weeks before the September meeting.
📌 Reading the Week
- Warsh gave the market a framework and a bias without a commitment, and the front end did the repricing. Hike odds moved from 35% to 57% and the two-year rose 12 basis points on a speech that contained no forward guidance at all. “Financing conditions didn’t look restrictive” is the operative sentence.
- Michigan removed the hawks’ best data point on the day they most needed it. One-year inflation expectations were revised down to 4.0% from 4.3%, and now sit below July. Warsh went hawkish anyway, which tells you he is weighting realised core PCE at 3.3% over survey expectations.
- Chicago PMI at 47.1 against 57.9 expected is the most under-covered number of the month. It contradicts Empire State at 20.6, the Philadelphia Fed at 47.4 and a 52-month-high composite PMI. Tuesday’s ISM Manufacturing print now carries the burden of resolving that.
The week ahead: ISM Manufacturing, JOLTS and construction spending Tuesday with Dell reporting; ADP, factory orders and the Beige Book Wednesday with Broadcom and Snowflake; and August payrolls Friday — into 57% hike odds.
Companies
Theme: “184% Up, 8% Down” — Marvell fell 8% on vague fiscal 2028 detail despite a Google partnership worth up to $12.2 billion in shares. It is still up 184% this year. The pattern that has run since late July held to the last session of the month: the higher the year-to-date gain, the lower the tolerance for ambiguity.
Friday was a light corporate session dominated by a single decline, and that decline is a clean summary of the whole reporting period. Marvell delivered a solid quarter and a very large partnership announcement, and fell 8% because it would not put numbers on the following fiscal year.
🔌 1. Marvell: The Google Deal Was Not Enough
Marvell Technology fell 8% on Friday. The company makes networking, connectivity and custom chips used in AI data centres, and it offered limited detail on its fiscal 2028 outlook, dampening investor sentiment after hopes that a Google partnership worth up to $12.2 billion in shares would further boost earnings.
The Google arrangement deserves attention on its own terms. A partnership valued at up to $12.2 billion in shares is a substantial commitment from a hyperscaler to a custom-silicon supplier, and it is exactly the kind of contracted relationship that has been rewarded elsewhere this season — Microsoft’s $678 billion remaining performance obligations, Amazon’s $496 billion AWS backlog, CoreWeave’s $104 billion, Supermicro’s $60 billion.
It was not enough, and the reason is positioning. Marvell entered the print with:
- A 179% year-to-date gain as of 24 August, rising to 184% for the year even after Friday’s 8% decline.
- A 58x forward price-to-earnings multiple.
- A 52-week range of $61.44 to $329.88 — a stock that has moved by a factor of more than five within a year.
- Prior earnings reactions ranging from an 18% gain to a 20% loss.
Analysts had set the bar high: consensus EPS of $0.65, representing 30% growth from $0.50 a year earlier, with full-year fiscal 2027 EPS of $3.07 implying 42% growth against $2.16 in fiscal 2026. When expectations are rebuilt that aggressively on the way up, ambiguity about the following year is itself a negative.
📉 2. The Season’s Governing Pattern, Restated
Marvell is the final entry in a list that has been consistent since 29 July.
| Company | YTD entering print | Outcome |
| AMD | +140% | −8% on 107% Data Center growth |
| SanDisk | +400% | −10% on a 12% EPS beat |
| Western Digital | +220% | −10% on a beat-and-raise |
| Datadog | Nearly doubled | −17% on a beat-and-raise |
| Cisco | — | −8.77% on record revenue, margin guide |
| Applied Materials | +110% | −5%+ on an 80% FCF collapse |
| Fabrinet | — | −11.3% on record revenue, weak FCF |
| Marvell | +179% | −8% on vague FY2028 detail |
| Analog Devices | — | Rose — on a 52% adjusted operating margin |
| Nvidia | +13% | +8.4% — on a $108bn guide and 70% FY2028 growth |
The two exceptions prove the rule. Analog Devices rose because it delivered an exceptional margin without a stretched valuation. Nvidia rose because it paired a cash-flow deterioration with a forward guide large enough to justify it — $108 billion for the current quarter and roughly 70% revenue growth projected for fiscal 2028.
Marvell had the growth but not the forward numbers. The lesson for the September reporting calendar is that the market is now paying for multi-year visibility, not for quarterly performance — the same conclusion SanDisk’s Investor Day rally demonstrated on 13 August.
🎯 3. What Broadcom Faces on Wednesday
Broadcom reports on 2 September and enters with a specific set of overhangs.
- It fell below $400 on 18 August on AI expansion financing concerns — the same borrowing question that Bank of America and Bloomberg quantified on Thursday at more than $210 billion of hyperscaler issuance and roughly $600 billion of total AI buildout borrowing since last year.
- It fell around 7% during the week of 17 August.
- It rose 3.79% on Thursday in the Nvidia read-through, then traded with the sector on Friday.
- Unlike Marvell, it has stayed closer to flat in 2026 — which by this season’s pattern is a favourable setup rather than an unfavourable one.
The question for Broadcom is the same one Marvell failed: will it put numbers on the forward period, particularly on custom-silicon commitments from hyperscalers? Nvidia set the template by guiding to $108 billion and projecting 70% growth two fiscal years out. Marvell declined to and lost 8%.
📋 4. Other Threads Into September
- Dell reports Tuesday 1 September alongside Medtronic and Palo Alto Networks — a server and infrastructure read that intersects directly with the memory cost inflation Nvidia documented at $279 billion of supply commitments.
- Snowflake, HPE, NetApp and Five Below report Wednesday 2 September with Broadcom, giving reads across data platforms, enterprise hardware, storage and the low-income consumer.
- Dollar General’s Thursday beat — EPS up 33% to $2.48 against $2.00 expected, with raised guidance and resumed buybacks — remains the strongest consumer datapoint of the month, against Dollar Tree selling off and Dick’s Sporting Goods cutting full-year profit guidance in its worst session on record.
📌 Analyst Take
The reporting season that ends this week produced one consistent and tradeable rule: the market is pricing forward visibility, and it is charging a premium for its absence in exactly proportion to how much a stock has already risen.
Marvell is the cleanest demonstration. A $12.2 billion Google partnership is unambiguously good news. A 58x forward multiple on a 179% year-to-date gain means good news is already in the price, and vague fiscal 2028 commentary is a reason to sell. The same arithmetic explains Applied Materials falling on a $700 million guidance beat and Datadog falling on a beat-and-raise.
For September positioning this suggests a specific screen: favour AI-exposed names that have underperformed the theme and that report with contracted forward numbers. Broadcom fits the first criterion; whether it fits the second is Wednesday’s question.
And the macro overlay has changed. With September hike odds at 57% and Warsh describing financing conditions as insufficiently restrictive, the discount rate applied to these long-duration names is now more likely to rise than fall into the FOMC. That argues for the same conclusion from the other direction: multi-year contracted revenue is worth more than a strong quarter, because it survives a higher discount rate.
General
Friday 28th – Sunday 30th August 2026: The Chair Chose Discipline Over Guidance
Kevin Warsh used his first Jackson Hole keynote to do something unusual: he declined to guide, declined to describe a reaction function, and instead set out a philosophy of central bank communication — while making clear that he does not believe policy is currently restrictive. The market resolved the ambiguity in one direction. September hike odds moved from 35% to 57%, and the two-year yield rose 12 basis points.
Beneath that, two data points pointed in opposite directions and neither received the attention it deserved. Chicago PMI collapsed to 47.1 against a 57.9 consensus — the weakest of 2026. And Michigan one-year inflation expectations were revised down to 4.0% from a preliminary 4.3%.
- What “Committed to a Discipline, Not to a Decision” Means in Practice
Warsh’s framing is a deliberate institutional choice and it will shape how this market trades for the rest of his term.
His stated objection: “We should not indulge a regime in which market participants are looking primarily to the Fed for their next trade.” That is a direct repudiation of the forward-guidance era, and it explains why the July FOMC statement contained no guidance at all — not an oversight, a policy.
The practical consequence is that every meeting is now genuinely live, and the market must price policy from data rather than from signalling. Friday demonstrated how that works: a speech with no commitment moved hike odds 22 percentage points, because the market had to infer the reaction function from a description of conditions.
The description was the hawkish part:
- “Financing conditions didn’t look restrictive to him.”
- An economy showing “few signs of restraint” from the current rate of around 3.6%.
- “While this summer’s readings were better than expected, they do not tell me that underlying trends have meaningfully improved.”
- The 2% PCE target is “firm and fixed”; if underlying inflation is not moving toward it fast enough, “we have work to do.”
Note the specific reference to PCE rather than CPI. Core CPI is at 2.5%; core PCE held at 3.3% on Wednesday. By naming the PCE target as firm and fixed, Warsh has effectively told the market which of the two measures the committee weights — and it is the higher one. That resolves an ambiguity this publication has flagged repeatedly as the source of the 9–3 July vote.
The chorus behind him is now substantial. Hammack said “now is the time to act” on Thursday and spoke again at 9:00am Friday. Schmid said rates are not providing restraint. And Chicago’s Austan Goolsbee — a 2027 voter — agreed with Warsh on the details of the economy and agreed inflation is the main issue right now.
- The Michigan Revision Undercut the Hawks and They Went Hawkish Anyway
This is the most analytically interesting sequence of the day.
In the previous edition I identified the preliminary Michigan one-year inflation expectation of 4.3% as the number that would “hand the hawks a direct argument” hours before Warsh spoke. The final reading revised it down to 4.0% — below July’s 4.2% — and the five-to-ten year measure held at 3.3% for a third consecutive month.
Warsh delivered a hawkish message regardless. The inference is straightforward and useful: he is weighting realised inflation over survey-based expectations. Core PCE at 3.3% is the binding constraint in his framework; a three-tenths improvement in a consumer survey is not.
That has a direct implication for how to trade the September meeting. If expectations surveys do not move him, then the only data that matters between now and 15–16 September is realised inflation and the labour market — which means Friday 4 September’s payroll report and the annual benchmark revision carry disproportionate weight.
The sentiment level itself remains poor and deteriorating. The final August reading of 51.7 compares with 55.2 in July, and the index bottomed at a record low of 44.8 in May 2026 before recovering to 49.5 in June and 55.2 in July. Most of that recovery has now been given back, with declines that the survey noted were not concentrated in any single group — falling across demographic and political lines, with particularly sharp reductions among Republicans, older consumers, lower-income households and those without college degrees.
- Chicago PMI at 47.1 Is the Month’s Most Important Overlooked Number
A collapse from 57.6 to 47.1, against a 57.9 consensus, is a miss of more than ten points into contraction territory — and it is the weakest reading of 2026.
It contradicts everything else in the manufacturing data:
| Survey | Latest | Direction |
| Empire State (Aug) | 20.6 vs 11.0 exp | Highest since 2022 |
| Philadelphia Fed (Aug) | 47.4 vs ~25 exp | Strongest since April 2021 |
| S&P Global composite (Aug flash) | 56.0 | 52-month high |
| S&P Global manufacturing (Aug flash) | 53.2, output 51.9 | 13-month low on output |
| ISM Manufacturing (Jul) | 55.6 | Highest since May 2022 |
| Chicago PMI (Aug) | 47.1 vs 57.9 exp | Weakest of 2026 |
Two readings are available and both deserve consideration.
The dismissive one: Chicago PMI is a single regional survey with heavy auto and industrial-supply-chain weighting, and it is among the most volatile business surveys published. A ten-point swing in either direction is not unprecedented.
The serious one: the S&P Global flash already showed manufacturing output at a 13-month low of 51.9 with input purchases falling for the first time since February and delivery times at four-year worsts. Chicago may be picking up the same de-stocking dynamic earlier and more sharply than the national measures. If the safety-stock building that supported goods production through the Hormuz disruption is now unwinding, the regional surveys measuring order books would stay strong while the surveys measuring current production would break — which is precisely the divergence observed.
Tuesday’s ISM Manufacturing print is now the arbiter, and it has become the most important release of the coming week alongside payrolls.
- The Curve, the Dollar, and a Warning From Citi
The rate move on Friday was concentrated at the front end, which is the textbook response to a hawkish shift without a growth signal.
- Two-year: +12 basis points to 4.35%.
- Ten-year: +5.6bp to 4.728%.
- Thirty-year: back to 2008 levels, having been “just beginning to cool down.”
- Dollar index: +0.5% to 99.65, reclaiming its 200-day moving average at 99.17 and approaching its 100-day at 99.60.
A cross-asset observation worth recording came from Citi’s Dirk Willer, who noted that keeping the 30-year Treasury yield below 5.30% could create downside pressure on the US dollar.
That inverts the usual framing and is worth thinking through. The conventional reading is that higher yields support a currency. Willer’s point is the opposite: if the authorities suppress the long end — through Treasury buybacks, or by deploying the general account, or by any other means — the dollar loses the yield support that has held it up. This month the Treasury has done exactly that, doubling buybacks and signalling it could use a $1 trillion cash balance.
So there is a genuine policy trade-off now visible: the government can defend the long end or it can defend the currency, but the tools that address one weaken the other. With federal debt past $40 trillion and roughly $600 billion of AI-related corporate borrowing competing for the same capital, that trade-off is not going away.
- Two Corrections and What They Change
Accuracy first: two items from Thursday’s edition require correction, and the second changes an interpretation.
On semiconductors. I reported the iShares Semiconductor ETF gaining 1.17% and built a case that Thursday represented a rotation “from silicon to software.” CNBC reported the same ETF up 3%, the VanEck Semiconductor ETF up 3.5%, Nvidia +7.4%, Marvell +5.8% and Micron +4.5%. The sources conflict and I cannot reconcile them from available data. The rotation thesis may still hold — Salesforce at +11.2% and CrowdStrike at +9% did outpace Nvidia — but I stated it with more confidence than the underlying figures support.
On oil, the error is directional and more consequential. I reported WTI rising toward $83 on reports Iran and Oman would share revenue from overseeing Hormuz traffic, framing it as tolls being imposed without US assent. CNBC reported oil extending declines on Thursday — Brent October down 0.63% to $87.29 and WTI October down 0.56% to $81.77 — because the market read the Iran–Oman talks as raising hopes of easing supply.
The correct characterisation of the underlying news: Iran and Oman are working on finalising details of an agreement to control the Strait, according to a senior Iranian source cited by Reuters, following comments from Iran’s Revolutionary Guards that the two countries had agreed how to share the waterway.
Why this changes the interpretation. I framed a toll regime as a diplomatic defeat for Washington that would keep the risk premium elevated. The market took the opposite view: any functioning arrangement, on whatever terms, is a mechanism for restoring flows. That is much closer to Commonwealth Bank of Australia’s threshold — Brent toward $70 within a $70–100 second-half range if flows recover to just 50–60% of pre-war quantities. The bearish case for crude is stronger than I represented it on Thursday.
📊 Global Macro Sentiment Summary — 28–30 August 2026
| Narrative Channel | Core Fundamental Trigger | Net Portfolio Posture |
| Index Structure | Dow −0.02%, S&P −0.25%, Nasdaq −0.52%; all three higher on the week by less than 1% | 🟨 Gave back an early gain |
| Fed — Warsh | “Committed to a discipline, not to a decision”; financing conditions “didn’t look restrictive”; 2% PCE target “firm and fixed” | 🟥 Hawkish without guidance |
| Rate pricing | September hike odds 57%, up from 35% the prior day | 🟥 Major repricing |
| Rates | 2-year +12bp to 4.35%; 10-year +5.6bp to 4.728%; 30-year back to 2008 levels | 🟥 Front-end led |
| Dollar | DXY +0.5% to 99.65, above the 200-day at 99.17 | 🟩 Yield support |
| Chicago PMI | 47.1 vs 57.9 expected, from 57.6 — weakest of 2026, in contraction | 🟥 Contradicts all other surveys |
| Michigan | Sentiment 51.7 vs 55.2 in July; 1-year inflation expectations revised DOWN to 4.0% from 4.3% | 🟨 Undercut the hawks |
| Fed chorus | Hammack, Schmid and Goolsbee all aligned with Warsh on inflation being the main issue | 🟥 Broad |
| Marvell | −8% on limited fiscal 2028 detail, despite a $12.2bn Google partnership; still +184% YTD | 🟥 Visibility premium |
| Cross-asset | Citi: keeping the 30-year below 5.30% could pressure the dollar lower | ⚠️ Policy trade-off |
| Payroll benchmark | Preliminary annual revision released — 2026 figure unconfirmed at time of writing; last year’s was −911,000 | ⚠️ Watch this |
Compliance and framing notes. Warsh explicitly declined to say whether he would support a September hike — do not present his remarks as a commitment. The preliminary payroll benchmark revision was released Friday but the 2026 figure was not confirmed at the time of writing; cite last year’s −911,000 only as historical context. And note the two corrections above regarding Thursday’s semiconductor ETF move and the direction of the oil price.
Upcoming News
Monday, August 31st, 2026 — Theme: “A Quiet Close to a Loud Month” — No major earnings or economic releases are scheduled, giving the market a month-end session to position for a week that runs ISM Manufacturing, ADP, the Beige Book and August payrolls into 57% hike odds.
Monday is month-end with an empty calendar, which makes it a rebalancing session rather than an information one. The market enters it having repriced the September FOMC from 35% to 57% odds of a hike in a single hour on Friday, with the two-year at 4.35% and the 30-year back at 2008 levels. Nothing scheduled on Monday will change that; everything from Tuesday onward could.
🔴 Calendar — Monday, August 31st, 2026
Times in ICT (Hanoi). ET is ICT minus 11 hours.
| Time (ICT) | Currency | Event / Indicator | Consensus | Impact |
| — | USD | No major earnings announcements or events | — | 🟢 Low |
| During session | — | Month-end rebalancing flows | — | 🟠 Med |
| 22:30 | USD | 3-Month and 6-Month Bill Auctions | — | 🟢 Low |
| Morning | CNY | China official PMI (August) (typically month-end) | — | 🟠 Med |
Charles Schwab’s investor calendar lists 31 August as having no major earnings announcements or events. Secondary releases may still appear — verify against your own terminal.
- Why Month-End Matters More This Month
August produced unusually large dispersion between asset classes, which mechanically generates rebalancing flows.
- Equities: all three major indices finished the final week higher, but by less than 1%, after a month containing both record highs and a synchronised global bond selloff.
- Bonds: the 30-year touched a multi-decade high above 5.34% on 18 August, retreated on the Treasury buyback announcement, and is back at 2008 levels after Warsh.
- Gold: five consecutive weekly gains through 21 August, reaching multi-month highs above $4,700.
- Bitcoin: a 22% weekly advance in the week to 21 August, its best in two years.
Portfolios that started the month at target weights are unlikely to have ended it there. Expect mechanical selling of the outperformers and buying of the underperformers into the close, which can produce moves unrelated to news.
- The Week That Follows Is the Most Consequential Since July
| Date | Data | Earnings |
| Tue 1 Sept | August ISM Manufacturing PMI; July JOLTS; July construction spending | Medtronic, Palo Alto Networks, Dell |
| Wed 2 Sept | August ADP nonfarm employment; July factory orders; Federal Reserve Beige Book | Broadcom, Snowflake, HPE, NetApp, Five Below |
| Thu 3 Sept | Initial jobless claims; ISM Services (typically) | — |
| Fri 4 Sept | August payrolls; annual nonfarm payroll benchmark revision | — |
| 15–16 Sept | FOMC decision and dot plot | 57% probability of a hike priced |
Three of these now carry elevated weight because of Friday.
ISM Manufacturing on Tuesday is the arbiter of a genuine data conflict. Chicago PMI collapsed to 47.1 against a 57.9 consensus on Friday — the weakest of 2026 and in contraction — while Empire State printed 20.6, the Philadelphia Fed 47.4 and the S&P Global composite a 52-month high of 56.0. The national ISM reading will indicate which of these is signal. July’s reading was 55.6, the highest since May 2022.
The Beige Book on Wednesday is more useful than usual. With Warsh having abandoned forward guidance and told the market to look at data rather than at the Fed, the Beige Book is now one of the few windows into how the committee is actually characterising conditions across districts — particularly on whether financing conditions look restrictive, which is the specific claim he made on Friday.
Payrolls on Friday is the decisive release, and it comes with the annual benchmark revision. The preliminary revision was published on Friday 28 August; the 2026 figure was not confirmed at the time of writing, and last year’s came in at −911,000 jobs. The July report already cut May and June by a combined 103,000, taking the trailing twelve-month average to roughly 34,000 a month. If the benchmark revision is large, the labour market the FOMC believed it was managing may not exist.
- Carry-Over Into the Week
- Warsh: “I stand here today committed to a discipline, not to a decision.” Financing conditions “didn’t look restrictive”; the 2% PCE target is “firm and fixed”; if inflation is not moving toward it fast enough, “we have work to do.” September hike odds moved from 35% to 57%.
- Two-year yield +12bp to 4.35%; ten-year +5.6bp to 4.728%; 30-year back at 2008 levels; dollar index +0.5% to 99.65.
- Chicago PMI 47.1 vs 57.9 expected — weakest of 2026. Michigan final sentiment 51.7 (from 55.2 in July) with one-year inflation expectations revised down to 4.0% from a preliminary 4.3%.
- Marvell fell 8% on limited fiscal 2028 detail despite a Google partnership worth up to $12.2 billion in shares; it remains up 184% for the year.
- On Thursday, Iran and Oman were reported to be finalising details of an agreement to control the Strait of Hormuz, with the Revolutionary Guards saying the two countries had agreed how to share the waterway. Oil fell on the news: Brent October to $87.29, WTI October to $81.77.
- The Positioning Question Into September
Two things changed on Friday that should shape allocation decisions for the coming fortnight.
First, the direction of the discount rate. For most of August the debate was whether the long end would ever rally. Warsh has now made the front end the live variable — 57% odds of a hike, with a Chair who says policy is not restrictive. That is a different risk from a term-premium problem: it hits short-duration credit, floating-rate borrowers and rate-sensitive consumers directly, and it compresses equity multiples across the board rather than only in long-duration growth.
Second, the reporting season has closed with a clear rule. The market is paying for multi-year contracted visibility, not for quarterly beats. Nvidia rose on a $108 billion guide and a 70% fiscal 2028 projection; Marvell fell 8% for declining to put numbers on the same period, despite a $12.2 billion Google partnership. Broadcom on Wednesday faces exactly that test.
Historically, September is the weakest month on Wall Street. Entering it with 57% hike odds, a Chair who has removed forward guidance, an unresolved conflict in the manufacturing data, and a payroll benchmark revision two weeks before the FOMC is an unusually crowded set of unknowns.
Compliance note: Charles Schwab’s calendar lists 31 August as having no major earnings or events; secondary releases may still appear and should be verified. The 2026 payroll benchmark revision figure was not confirmed at the time of writing — do not cite a number. And note that September is historically the weakest month for US equities, which is a seasonal observation rather than a forecast.
Snapshot
Friday 28th – Sunday 30th August 2026 — Theme: “Discipline, Not a Decision” — Warsh declined to guide and moved September hike odds from 35% to 57% anyway, by saying financing conditions do not look restrictive. The two-year rose 12 basis points, the dollar gained 0.5%, and Chicago PMI collapsed to 47.1 — the weakest of 2026.
The month ended with the Federal Reserve’s new Chair rewriting how the market has to read policy. Warsh used his first Jackson Hole keynote to reject forward guidance as an institution — “we should not indulge a regime in which market participants are looking primarily to the Fed for their next trade” — while describing an economy that shows few signs of restraint at a policy rate of around 3.6%. Equities gave back an early gain, the front end repriced sharply, and two data releases that arrived in the same minute pointed in opposite directions.
🏛️ The Bottom Line
The Dow Jones Industrial Average finished roughly flat at −0.02%, the S&P 500 fell 0.25%, and the Nasdaq Composite declined 0.52% after having been up as much as 0.5% earlier in the session. All three indices still closed the week with gains, though by less than 1%, thanks to the midweek Nvidia rally.
Chair Kevin Warsh delivered his first Jackson Hole keynote at 10:00am ET, avoiding both forward guidance and a reaction function. Key remarks: “I stand here today committed to a discipline, not to a decision”; “We should not indulge a regime in which market participants are looking primarily to the Fed for their next trade”; “While this summer’s readings were better than expected, they do not tell me that underlying trends have meaningfully improved.” He said financing conditions “didn’t look restrictive” to him, described an economy showing few signs of restraint from the current rate of around 3.6%, and stated that the 2% PCE target is “firm and fixed” — adding that if underlying inflation is not moving toward it fast enough, “we have work to do.”
Fed funds futures moved to 57% odds of a hike at the 15–16 September meeting, up from 35% the prior day. The two-year yield rose around 12 basis points to 4.35%, the ten-year gained 5.6bp to 4.728%, and the 30-year returned to 2008 levels. The dollar index rose 0.5% to 99.65, back near its 100-day moving average of 99.60 and above its 200-day at 99.17.
Chicago PMI collapsed to 47.1 in August against a 57.9 consensus and a 57.6 prior reading — the weakest print of 2026 and a move into contraction territory.
The final August University of Michigan survey came in at 51.7 against a 51.0 consensus and preliminary reading, versus 55.2 in July. Current conditions were 51.9 and expectations 51.5, revised up from a preliminary 50.6. One-year inflation expectations were revised down to 4.0% from a preliminary 4.3% and July’s 4.2%, while the five-to-ten year measure held at 3.3% for a third consecutive month. The index had bottomed at a record low of 44.8 in May 2026.
The preliminary annual revision to payroll data was released at 10:00am ET, in the same minute as the keynote and the Michigan reading. The 2026 figure was not confirmed at the time of writing; last year’s revision came in at −911,000 jobs.
Marvell Technology fell 8% after offering limited detail on its fiscal 2028 outlook, dampening sentiment after hopes that a Google partnership worth up to $12.2 billion in shares would further boost earnings. The stock remains up 184% in 2026. It had entered the print with a 179% year-to-date gain and a 58x forward multiple.
Cleveland’s Beth Hammack spoke at 9:00am, having said on Thursday that “now is the time to act.” Chicago’s Austan Goolsbee, a 2027 voter, agreed with Warsh on the details of the economy and agreed inflation is the main issue right now. Citi’s Dirk Willer noted that keeping the 30-year Treasury yield below 5.30% could create downside pressure on the US dollar.
📉 Reference Levels for the Monday Open (August 31st)
| Asset | Support | Resistance | Operational Bias |
| S&P 500 | 7,674 → 7,652 | 7,730 → 7,798.99 (record) | 🟨 Week higher, Friday lower |
| Nasdaq Composite | 26,180 → 25,980 | 26,541 → 26,803 | 🟥 Gave back an early gain |
| Dow Jones | 53,417 → 53,277 | 53,569 → 54,349 | 🟨 Flat on the day |
| US 2Y Yield | 4.20% | 4.35% → 4.50% | 🟥 57% hike odds |
| US 10Y Yield | 4.66% | 4.728% → 4.80% | 🟥 Rising |
| US 30Y Yield | 5.20% | 2008 levels | 🟥 Cooling reversed |
| US Dollar Index | 99.17 (200-day) | 99.65 → 100 | 🟩 Reclaimed the average |
| Brent Crude | $85 → $70 (CBA downside) | $87.29 → $94 | 🔄 Iran–Oman talks bearish |
| WTI Crude | $78 → $75 | $81.77 → $85 | 🔄 Extended declines |
| Gold | $4,569 | $4,731 → $4,850 | 🟨 Dollar strength a headwind |
📊 Market Sentiment & Bias
Fed: 🟥 Hawkish, and the framework has changed. No forward guidance by design. “Financing conditions didn’t look restrictive” is the operative claim, and hike odds moved 22 percentage points on it. Naming the 2% PCE target as “firm and fixed” tells the market the committee weights core PCE at 3.3% over core CPI at 2.5%.
Rates: 🟥 The front end is now the live variable. Two-year +12bp; 30-year back to 2008 levels just as it was beginning to cool. This is a different risk from the term-premium problem that dominated mid-August.
Manufacturing: ⚠️ A genuine conflict. Chicago PMI at 47.1 versus 57.9 expected, against Empire State at 20.6, the Philadelphia Fed at 47.4 and a 52-month-high composite. Tuesday’s ISM is the arbiter.
Consumer: 🟥 Deteriorating, but expectations improved at the margin. Sentiment at 51.7 against 55.2 in July, with declines across demographic and political lines. One-year inflation expectations revised down to 4.0% — and Warsh went hawkish anyway.
Earnings: 🟥 Visibility is everything. Marvell fell 8% on vague fiscal 2028 detail despite a $12.2 billion Google partnership, while up 184% for the year.
Cross-asset: ⚠️ A policy trade-off is now explicit. Per Citi, suppressing the 30-year below 5.30% could pressure the dollar. The Treasury has spent August doing precisely that.
💡 Top Trade Takeaway: “The Live Variable Moved to the Front End”
Focus: Reassess floating-rate and short-duration credit exposure against 57% September hike odds. Favour AI names that report with contracted multi-year numbers over those with strong quarters and vague forward detail. Treat Tuesday’s ISM as the arbiter of a real manufacturing conflict. Keep gross exposure moderate into a week containing ISM, ADP, the Beige Book, Broadcom and payrolls with a benchmark revision.
Logic. Warsh changed the mechanism by which this market prices policy. By rejecting forward guidance as an institution — “we should not indulge a regime in which market participants are looking primarily to the Fed for their next trade” — he has made every meeting live and forced the market to infer the reaction function from descriptions of conditions. Friday demonstrated the consequence: a speech containing no commitment moved September hike odds 22 percentage points, because the description was hawkish. “Financing conditions didn’t look restrictive” and “few signs of restraint” at a 3.6% policy rate is a statement that the current setting is insufficient.
The most useful single detail for forecasting is that he named the 2% PCE target as “firm and fixed.” Core CPI is at 2.5% and core PCE at 3.3% — a gap that has driven the committee’s 9–3 split all year. By anchoring to PCE, Warsh has effectively told the market which measure binds. And he did so on the day Michigan revised one-year inflation expectations down to 4.0% from 4.3%, which means survey expectations do not move him. Only realised inflation and the labour market will.
That places disproportionate weight on Friday 4 September. August payrolls arrive alongside the annual benchmark revision — the preliminary version of which was published on 28 August, with last year’s equivalent cutting 911,000 jobs. The July report already revised May and June down by a combined 103,000, leaving a trailing twelve-month average near 34,000 a month. A large benchmark revision two weeks before the FOMC would change the labour picture the committee believes it is managing.
The manufacturing conflict is the week’s other unresolved question, and it is unusually stark. Chicago PMI collapsed to 47.1 against a 57.9 consensus — the weakest of 2026, in contraction — on the same day the S&P Global composite sits at a 52-month high and two regional Fed surveys are at multi-year peaks. The reconciling hypothesis is that the safety-stock building which supported goods production through the Hormuz disruption is now unwinding, consistent with S&P Global already showing manufacturing output at a 13-
month low and input purchases falling for the first time since February. Tuesday’s ISM print resolves it.
Calendar discipline: Monday 31 August — month-end, no scheduled events, expect rebalancing flows. Tuesday 1 September — ISM Manufacturing, JOLTS, construction spending; Dell reports. Wednesday 2 September — ADP, factory orders, the Beige Book; Broadcom and Snowflake report. Friday 4 September — August payrolls and the annual benchmark revision. 15–16 September — FOMC, with 57% odds of a hike priced.
The report belongs to The Concept Trading and Van Hung Nguyen
