Monthly Report • September 4th 2026

The tape is at a record, the bond market is at a decade high, and the Fed is about to pick one of them.

Bottom Line Up Front

Equities enter September within about 1% of an all-time high (S&P 500 at 7,748 on Thursday's close, record just above 7,800 two weeks ago), yet the 10-year Treasury touched 4.82% this week, its highest since November 2023, the 30-year sits at 5.24%, and after this morning's blowout August jobs report (+162K vs. ~55K expected, unemployment 4.1%) futures price better-than-even odds that Kevin Warsh's Fed raises rates on September 16. Layer on a re-escalating U.S.–Iran conflict with Brent back near $95, the worst-seasonality month of the year, and a midterm election eight weeks out, and the setup is a market that has priced a lot of good news into an environment that is getting more expensive to finance. Base case for the month: a choppy, range-bound September that resolves lower into the FOMC and then trades on whether the Fed's dot plot confirms a hiking bias, with the AI capex trade continuing to carry the index while the rest of the market stalls.

Market Bias Dial — FEAR ←→ GREED

Composite: 4.5 / 7 — Neutral-Greed. The reading is split in a way that matters. Price-based inputs (SPY above its moving averages, VIX at 14.3, high-yield spreads in the richest decile of their history) all say greed. Positioning-based inputs say the opposite: SPY put/call ratios on the asleepace.com chain run 1.4–1.9 on near expiries and 1.9 on open interest for October, breadth has more new lows than new highs, and a third-party Fear & Greed composite reads 41 (Fear). That divergence — index near highs, hedges bought, breadth soft — is the classic late-rally profile: a narrow group of AI winners is holding the index up while institutions quietly buy protection under it. The full scoring is in the appendix.

Where SPY Stands and Where It's Likely Headed

SPY is trading at $769.90 midday Friday, down about 0.4% after the jobs print, having closed at $773.17 Thursday. The asleepace.com options chain shows a negative gamma regime on today's expiry with net GEX of roughly −$3.3B and the gamma flip at 771, essentially on top of spot. Gamma exposure (GEX) measures how much market-makers must buy or sell to stay hedged as the index moves; when net GEX is negative and spot is below the flip, dealers are short gamma, meaning they sell into declines and buy into rallies, which amplifies moves rather than dampens them. The system's own regime-stability flag reads "unstable," and the expected-move model puts an 85% probability on a break rather than a pin today.

Nearer-term levels: max pain 768, put wall 765, call wall 778 on the 0DTE chain. Today's most notable flow is directional put buying at 769 (284K contracts against 5.8K open interest, a 49x volume/OI ratio) — that is not a hedge, that is a bet. On the September 18 monthly expiry, the biggest single print of the day is 43,881 contracts of the 760 put; the largest call open interest sits at 790 (61K contracts). Read together, dealers are positioned for a 760–790 box through quad witching, with the downside leg being actively topped up.

The forward term structure is in normal contango: at-the-money implied volatility runs about 6.5% for Tuesday's expiry, 10% for Sept 16 (the FOMC day), 11.1% for Sept 18, 11.4% for Oct 2, and 12.1% for Oct 16. That is low. The Sept 18 at-the-money straddle (770 call $6 + 770 put $7) prices roughly ±$13, or ±1.7%, through an FOMC where a hike is a coin flip plus quad witching. The Oct 16 straddle (770 call $13 + 770 put $12) prices ±$25, or ±3.2%, a one-standard-deviation range of roughly 745–795 over the next six weeks. For an event set that includes CPI, a live Fed meeting with new dot plots, quad witching, quarter-end, the September jobs report, and an open shooting war in the Gulf, that is a cheap distribution — which is the core of the hedging idea further down.

Implied volatility "skew" (how much more expensive puts are than equivalent calls) is modest at 2–3 points on SPY, while the IWM chain is far more stressed: net GEX −$978M, put/call 1.55, skew of 10.6 points, gamma flip at 297 versus spot 295. Small caps are where the hike fear is concentrated, which makes sense — they are the most floating-rate-sensitive corner of the market.

Macro Backdrop & Quarter-Ahead Outlook

The jobs report changed the September debate

The August employment report (BLS, Sept 4) showed payrolls up 162,000 against a consensus near 55,000, with June and July revised up a combined 55,000 — July's originally reported loss of 23,000 is now a gain of 21,000. Unemployment held at 4.1%, average hourly earnings rose 0.3% on the month and 3.1% year over year, and the average workweek ticked up to 34.4 hours. The one soft spot: information-sector employment fell 23,000, led by computing infrastructure, data processing and web hosting (−8,000) — a notable wrinkle given that sector is supposedly the beneficiary of the AI buildout. The market reaction was immediate: the two-year yield jumped 7 basis points to around 4.40% and hike odds for the Sept 15–16 meeting rose to roughly 52–55% on futures and prediction markets (CME FedWatch via Bloomberg; Kalshi via CNBC).

The context matters. The Fed has been on hold at 3.50–3.75% since December 2025. Warsh replaced Jerome Powell in late May. The July meeting produced a 9–3 vote with three dissents in favor of hiking, and the June dot plot showed nine of seventeen participants projecting at least one hike in 2026 (Schwab). The August 7 jobs miss had pushed hike odds down to ~40%; Warsh's Jackson Hole speech on August 28 pushed them back above 55% by saying that better summer inflation prints "do not tell me that underlying trends have meaningfully improved" and that financial conditions were "not broadly restrictive" (Federal Reserve; CNBC). Governor Waller then pulled the other way on Thursday, saying he'd be "inclined to support" a hold absent an inflation surprise. This morning's data is the surprise Waller was hedging against.

Inflation: better on paper, not at the source

Headline CPI printed 3.4% year over year in July (down from a 4.2% May peak that was the highest since 2023), core CPI 2.5%, and the three-month annualized core rate slipped to 1.6% (TD Economics; BLS via CNBC). Core PCE — the Fed's preferred gauge — was 3.3% year over year in July (Nuveen). The problem is that the disinflation has come from the energy reversal after the spring ceasefire, and energy is now re-accelerating. Gasoline is still up ~25% year over year and fuel oil ~39% (Trading Economics). Warsh's own framing at Jackson Hole was that 54% of PCE components ran above 3% annualized over the past twelve months. August CPI lands Friday, September 11, four days before the meeting.

The war is not over, and oil knows it

The U.S.–Israel war with Iran began February 28; the Strait of Hormuz was effectively closed, Brent peaked above $118 on March 31, a two-week ceasefire in April collapsed, a June 17 memorandum to reopen the strait collapsed within days over routing disputes, and the U.S. and Iran traded strikes again this week for the first time since July. Brent is around $95, up more than 7% on the week; Iran struck at Kuwait, Jordan and Bahrain; only six commodity vessels transited Hormuz Wednesday versus a ten-day average near thirteen; and the world's largest tanker operator, Mitsui O.S.K., now expects no normalization by year-end (CNBC, Sept 3; Trading Economics). RBC's Helima Croft estimates the world is still losing roughly 8 million barrels a day versus pre-war flows. The energy sector (XLE) is up 10.4% over one month and 12.1% over three, the best of any sector — that is the market's honest read on the ceasefire odds.

What the bond market is pricing

Yields at 4.77% (10y), 5.24% (30y) and 4.40% (2y) give a 2s10s curve of roughly +40 basis points and a 3m10y of nearly +90 — a curve that has steepened on the long end even as the front end prices a hike. That combination — rising term premium plus a hawkish Fed — is unusual and uncomfortable for equities: it means the long end is not buying the idea that hikes will slow the economy enough to bring rates back down, and it is happening globally (U.K., German, French and Japanese yields all rose this week per CNBC). Nuveen still forecasts a 4.25–4.50% 10-year at year-end, which implies a rally from here; the bond market is not there yet. A 10-year at 4.8% against an S&P 500 forward P/E near 21 (inferred from Nvidia's 23.9x being cited as a 14% premium to the index) leaves an equity risk premium close to zero — a valuation observation, not a timing signal, but one that makes the index more sensitive to any further rise in real yields.

Dollar, gold, credit

The dollar (UUP $28.03) is flat to slightly firmer; gold (GLD $407.50, roughly $4,400/oz) has stalled a few percent below its highs as real yields rose, though the Dutch central bank quietly repatriating 86 tons of gold from New York and Ottawa to London "in view of increasing geopolitical unrest" is the kind of detail that tends to matter later. Credit is the most complacent asset class: high-yield OAS at ~260–275 basis points sits in the richest decile of its history against a long-run median near 450, and investment grade at ~78–81 (Nuveen; ICE BofA via Convex). Spreads have tightened through the rate selloff. That is either confirmation that the economy is fine or the setup for the classic equity-credit divergence; for now it is the former.

The AI cycle: carrying the index, narrowing the market

Nvidia's late-August quarter (guidance for roughly 70% revenue growth next fiscal year, an ~9% one-day pop adding $442B of market value, a $12.9B acquisition of Hugging Face this week) and Micron's 208% year-to-date run are the reason the index is where it is. Schwab notes Nvidia and Micron alone account for roughly one-third of aggregate 2026 S&P earnings growth and the top ten earners for two-thirds; FactSet had Q2 earnings growth at 50.4%, the highest since 2021 (interactive investor). The Magnificent Seven are ~34% of the index. The "great rotation" narrative from earlier this year — capital moving from silicon to steel, energy, staples and small caps — has partly reversed: industrials (XLI) are down 6.4% over one month, the worst sector, while tech closed at +1.3% Thursday. Today's tape is the pattern in miniature: NVDA +1.4%, MU +4%, AMD +3%, while Apple −2.4%, Microsoft −1.9%, Alphabet −1.4%, Amazon −1.1% and Tesla −6% (NHTSA opened an audit of ~1,000 Cybercabs). The AI trade is no longer "mega-cap tech" — it is a semiconductor and infrastructure trade, and the rest of the Seven are being used as a funding source.

Midterms: eight weeks out

The November 3 midterms are the first under a non-consecutive second term since 1894. Republicans hold roughly 218 House seats against Democrats' low 210s; Democrats need a net gain of about three seats. Polymarket prices roughly a coin flip on a Democratic sweep with Democrats favored for the House and the Senate a tossup (Polymarket, Sept 4); Decision Desk and Inside Elections have been shifting individual districts in both directions this week. For markets, the relevant historical patterns are well known: midterm years tend to be the weakest of the four-year cycle into the election and among the strongest after it, and the S&P 500 has risen in the twelve months following every midterm since 1950. The mechanism is not partisan — it is the removal of uncertainty plus the fact that the incumbent party tends to lean on fiscal and rhetorical stimulus into a vote. This cycle that pressure is already visible in the public friction between the White House (which wants lower rates) and a Fed chair signaling hikes. A divided Congress after November would freeze the legislative agenda — tariffs, tax, spending — which markets have historically treated as benign.

Three months out

The base case through Thanksgiving: the Fed hikes once (September or October) or holds with an explicitly hawkish dot plot; the 10-year holds a 4.5–5.0% range; Hormuz stays contested with Brent $85–105; AI capex guidance stays strong through the October earnings season; the election passes and the year-end seasonal bid arrives. The bear case is a hike plus an oil spike plus a rates-driven de-rating of the index at 21x — that is the 2022 template, and the long bond is already whispering it. The bull case is an August CPI print soft enough to keep Warsh on hold, a durable Hormuz deal (Brent $75), and a broadening rally into the election. Probabilities are not assigned because the September 11 CPI print alone can swing the central scenario; anyone who claims a defensible distribution before that number is guessing.

Key Dates to Watch

Week of Sept 7 (Labor Day Monday, markets closed)

• Wed Sept 9 / Thu Sept 10 — Oracle FQ1 earnings (date reported as Sept 9 or Sept 10 by different calendars; confirm before trading). A backlog read on AI infrastructure demand.
• Thu Sept 10 — PPI; 10-year Treasury auction typically this week.
• ⚠ Fri Sept 11 — August CPI. The last inflation print before the Fed decides. 30-year auction typically this week.

Week of Sept 14

• ⚠ Tue–Wed Sept 15–16 — FOMC decision, Summary of Economic Projections, dot plot, Warsh press conference. A hike is roughly a coin flip; the dot plot is the bigger event.
• Thu Sept 17 — Retail sales (typical slot).
• ⚠ Fri Sept 18 — Quad witching / September OpEx. 760 puts and 790 calls are the loaded strikes.

Week of Sept 21

• Tue Sept 22 — Micron earnings (estimated date). The single most-leveraged AI beneficiary of the year reports into a stock that has tripled.
• Late week — Q2 GDP third estimate; August PCE (typical slot, final Fed inflation read before October).

Week of Sept 28

• Wed Sept 30 — Quarter-end rebalancing; heavy 800-call open interest on the Sept 30 expiry.
• Fri Oct 2 — September jobs report.

Beyond the window

• Oct 16 — October OpEx (put/call OI 1.9 on the asleepace.com chain; 730 puts are the largest position).
• Oct 27–28 — FOMC.
• Tue Nov 3 — Midterm elections.

Upcoming Catalysts — What Could Move the Tape

The September 11 CPI print is the fulcrum for everything else. A core print at or below 0.2% month over month keeps the "Waller hold" alive and probably produces a relief rally into the meeting that tests the 778–790 call walls; a 0.3% or hotter print makes the hike close to certain and puts the 760 put wall in play before the Fed even speaks. Because the FOMC day is four sessions after CPI and quad witching is two sessions after that, the entire September distribution is compressed into eight trading days from the 11th to the 18th — which is exactly where dealer positioning (negative gamma, unstable regime) is most likely to exaggerate whatever direction the data picks.

The Fed meeting itself is a two-part event. The decision matters less than the dot plot: if the median 2026 dot moves to show a hike (or two), the two-year reprices toward 4.6% and the small-cap/rate-sensitive complex (IWM, XLRE, XLU, regional banks) takes the hit regardless of what happens on the day. Warsh's stated distaste for forward guidance ("we should not indulge a regime in which market participants are looking primarily to the Fed for their next trade") means the press conference is likely to be less informative than the projections — plan around the document, not the Q&A.

Oil is the wildcard that does not follow a calendar. This week's re-escalation ended a month of relative calm; a strike on Gulf energy infrastructure would push Brent through $100 and hand the Fed a hike it can justify without reference to the labor market. A genuine reopening deal would do the opposite and is the single most bullish event available to equities this month.

Names on the Watchlist

• SPY — The index is the trade this month; the 760–790 dealer box and the 745–795 one-sigma range frame everything. Next catalyst: CPI Sept 11.
• IWM — Most rate-sensitive index, most stressed options chain (net GEX −$978M, skew 10.6 pts). Lags on hike days, leads on hold days. Next catalyst: FOMC Sept 16.
• NVDA ($231.67) — Still the index's engine; post-earnings gap higher has held. A Hugging Face deal and the Cybercab/Tesla noise show the AI complex is now doing M&A and product launches at scale, which is what late-cycle looks like. Next catalyst: ORCL backlog read Sept 9/10 as a demand proxy; none of its own in window.
• MU ($997) — Up ~208% year to date, reports around Sept 22. The stock has become a leveraged bet on HBM pricing; the setup is asymmetric in both directions. Next catalyst: earnings Sept 22 (est.).
• ORCL ($157.62) — The first AI-infrastructure print of the month and the cleanest read on whether hyperscaler capex guidance is still rising. Next catalyst: earnings Sept 9/10.
• XLE ($64.22) — Best sector over one and three months; the market's Hormuz hedge. Reports on tanker traffic through the strait are the daily catalyst.
• XLI ($174.56) — Worst sector over one month (−6.4%, −6.6% relative to SPY). The "silicon to steel" rotation unwound fast; a Fed hold would likely bring it back. Next catalyst: FOMC Sept 16.
• TLT ($82.32) — 30-year at 5.24%; the long end is the cleanest expression of the term-premium debate. Next catalyst: 30-year auction week of Sept 7; FOMC.
• TSLA ($352) — Down ~6% on the NHTSA Cybercab audit the day after rallying 5% into the launch event. Headline-driven; not a positioning trade.

Outsized Risk/Return Ideas

1. Buy the cheap distribution: SPY Oct 16 760/730 put spread

Instrument: Buy SPY 760 put / sell SPY 730 put, October 16 expiry. Thursday closes: 760P ≈ $8, 730P ≈ $3, net debit ≈ $5 per spread (verify live; prices move).

Thesis: Six-week implied volatility of ~12% is pricing a ±3.2% range through CPI, a live FOMC with new dots, quad witching, quarter-end, a jobs report and a hot war. The put wall at 760 and the largest October put position at 730 are the exact strikes dealers will be forced to defend, which means a break through 760 in a negative-gamma regime is likely to travel. Max profit $25 on $5 at risk (5:1) if SPY is at or below 730 (−5.2%) on Oct 16.

What has to be true: A hike or hawkish dot plot on Sept 16, or a hot Aug CPI, or a Gulf escalation — any one of the three.

What invalidates it: Soft CPI (core ≤0.2%) plus a Fed hold with an unchanged dot plot; SPY reclaiming and holding 778–790. Cut at half the debit if SPY closes above 790.

Timeframe: Through Oct 16; reassess after the Sept 16 decision.

2. Pair: long SPY / short IWM into the FOMC

Instrument: Dollar-neutral long SPY, short IWM (or long SPY calls / long IWM puts for defined risk).

Thesis: Small caps are the rate-sensitive end of the equity market, and their options chain already shows it — IWM net GEX is deeply negative and skew is five times SPY's. If the Fed hikes or shows a hiking dot plot, IWM underperforms; if the Fed holds, the AI-heavy S&P still has the earnings tailwind. The pair wins in the hawkish scenario and roughly breaks even in the dovish one.

What has to be true: The two-year stays above 4.30% through the meeting.

What invalidates it: A dovish surprise (hold + dots shift toward cuts) that triggers a small-cap short-covering rally; the two-year through 4.15%. Line in the sand: IWM outperforms SPY by 3% on a closing basis.

Timeframe: Sept 8 through Sept 18.

3. Fade the industrials washout — but only after the Fed

Instrument: XLI shares (or Oct/Nov 175 calls for defined risk).

Thesis: Industrials are down 6.4% in a month and 6.6% against the index, the sharpest sector reversal in the tape, on a combination of rate fear and rotation back into AI. Manufacturing employment is up 58,000 since December and the manufacturing diffusion index hit 61 in August (BLS) — the fundamentals did not fall 6%. This is a "thrown out with the bathwater" candidate, but the trigger is a rates ceiling, not a price.

What has to be true: The Sept 16 dot plot does not add hikes beyond one; the 10-year fails to hold above 4.85%.

What invalidates it: A two-hike dot plot or the 10-year through 5%. Do not front-run the meeting.

Timeframe: Entry after Sept 16 if conditions are met; hold into the post-election seasonal window.

4. Cheap tail hedge on the long end: TLT puts

Instrument: TLT November puts, ~5% out of the money (verify live pricing).

Thesis: The 30-year is at a multi-year high and the curve is bear-steepening while the Fed leans hawkish — the market is not convinced hikes will cap long rates. A Fed that hikes into a war-driven oil shock while the Treasury runs large deficits into an election is the scenario in which the long end breaks. This is not a base-case trade; it is a tail hedge that pays in the one scenario where equities, credit and the dollar all lose at once.

What has to be true: 10-year breaks 4.85% and holds; a weak 30-year auction the week of Sept 7.

What invalidates it: Nuveen's year-end 4.25–4.50% forecast starts to look right — a soft CPI and a Fed hold. Small size, accept the premium loss.

Timeframe: Through November.

The $1,000 Allocation

The book is 55% equity beta (with a third of that in the sector that benefits from the war), 13% in defined-risk options that all pay in the same hawkish/escalation scenario, and 25% in bills — a deliberately Neutral-Greed posture that respects the record-high index while refusing to be unhedged through Sept 11–18. Rebalance triggers: (a) soft CPI + Fed hold + unchanged dots → close the put spread and IWM puts for whatever they're worth, deploy the XLI tranche and half the cash into SPY; (b) hike or two-hike dot plot → let the hedges run, keep the XLI tranche in cash, add to XLE on any dip; (c) Hormuz reopening deal → cut XLE to 5%, add to SPY and XLI. This is how a systematic desk might express the view; it is illustrative, not personal advice.

Anomalies and Things That Don't Add Up

• Payrolls versus the sector that pays for them. Headline jobs beat by 100K, but information employment — including computing infrastructure and data processing — fell 23K in a month when the AI buildout is supposedly the growth story. Either the AI capex boom is capital-intensive and labor-light (plausible, and bad for the "AI lifts everyone" narrative), or the establishment survey is lagging. Worth watching in the October print.
• Options say fear, price says greed. SPY put/call ratios of 1.4–1.95 near-dated and 1.9 on October open interest, breadth with 180 new lows against 55 new highs, and a 41 Fear reading on a composite index — while the index sits 1% from a record and VIX is 14. The extreme 49x volume/OI directional put buying at 769 today is not hedging behavior. Someone is paying for downside.
• Credit is asleep. HY spreads at 260–275 bps tightened through a week in which the 10-year hit a three-year high and the Fed signaled hikes. Credit and rates are telling incompatible stories; historically rates win and credit catches up.
• A cheap FOMC straddle. ±1.7% through a coin-flip hike decision and quad witching is unusually cheap for the event risk. Either the options market is confident the outcome is known, or it is mispricing — the report leans mispricing.
• Internal single-name gamma feed is stale. The asleepace.com multi-ticker context endpoint returned NVDA at $152, AMD at $295, AAPL at $301 and AMZN at $229 — all far from live quotes ($231, $470, $320, $256). Max-pain and flip levels for those names were discarded. QQQ, IWM and META matched live prices and were used. Worth a look at the feed.
• Bear-steepening into a hiking Fed. The 30-year at 5.24% and a 2s10s of +40 bps while the front end prices a hike is the term-premium story, not the growth story. It is the least equity-friendly curve shape available.

Other Threads Worth Pulling

• The AI trade has narrowed to semis and infrastructure. Nvidia, Micron, AMD and Oracle are up on days when Apple, Microsoft, Alphabet and Amazon are down. That is a different, more cyclical trade than "mega-cap quality," and it makes the index more dependent on a handful of memory and accelerator prices. The next test is Micron on Sept 22.
• Central-bank gold repatriation. The Dutch central bank moved a quarter of its New York/Ottawa reserves to London between March and August. This is the second European central bank to do so in a year. It has no immediate market impact; it is the kind of thing that gets cited in hindsight.
• Wage growth versus inflation. Real average hourly earnings have been negative for roughly four months (CPI 3.4% vs. wages 3.1%). That is the number that decides midterms, and it is the reason a Fed hike into an election is politically radioactive regardless of what the data say.
• Diesel. Ukrainian strikes on Russian refineries plus Gulf outages have made the global diesel market unusually tight even with crude below its peak. Diesel feeds directly into freight, food and PPI — the inflation channel that does not show up in the crude price.

Appendix — How the Bias Score Was Built

Data gaps noted honestly: AAII and CNN readings were not retrievable at generation time and were substituted with a transparent third-party composite; the internal single-name gamma feed was stale for NVDA, AMD, AAPL and AMZN and was excluded. All SPY structure data is from the asleepace.com chain as of 15:45 UTC, Sept 4, 2026; live quotes via Robinhood at 15:47 UTC.