The week ahead — August 30, 2026: A stretched tape into a payrolls week

Four macro prints and one earnings report land before Friday: ISM manufacturing Tuesday, ADP and JOLTS Wednesday, ISM services Thursday, then August payrolls Friday 9/4 with consensus +45k after July's -23k. Broadcom reports Wednesday after the close.

Revised September 1, 2026. Edited for length: repeated facts and levels were trimmed and duplicated template blocks removed. The analysis, levels, scenarios, sources and the view as of August 30 are unchanged.

The setup

Transition. Index trend intact, internals splitting: small caps under their weekly line while megacap tech holds, into a two-sided policy week.

What is actually going on

Going into September the question this market spends its time on has changed. For two years the debate was when the committee starts cutting. Now it is whether it hikes. At the July meeting rates were held at 3.50-3.75% on a 9-3 vote, and all three dissenters wanted a hike. The last time three members dissented the same way was September 2016. The committee's centre of gravity has moved, whatever any one official says in public. The Jackson Hole keynote on August 28 confirmed the tone, and the market repriced September from a near-certain hold to something close to a coin flip in a single session.

That matters well beyond one meeting, because it inverts the reflex that has governed every dip since 2024. For two years weakness was good news: soft data meant easing was coming, and the put under the market was the committee's willingness to respond. If the committee now sits on the tightening side of neutral, that put is weaker. A bad labour number stops being dovish relief; it becomes a growth problem with no offset. This is the largest change in the market's operating environment right now, and it is not yet reflected in how people talk about risk.

Underneath sits an inflation problem that monetary policy answers badly. Core PCE runs 3.3% year over year and headline 3.7%, and a good part of that pressure is energy rather than demand. Roughly 0.6 million barrels a day of Middle East supply is expected to stay offline through the end of next year on Strait of Hormuz constraints, Red Sea shipping is still disrupted, and refining capacity is tight. Supply-driven inflation only responds to rate rises through demand destruction. It explains why the hawkish argument survives a reasonable monthly core print. It also explains why the disagreement among strategists is unusually sharp. There is a serious published case that housing and labour give the committee no reason to move at all.

The second story is crowding, and this is not a nervously positioned market. Bank of America's August fund manager survey puts cash at 3.5% of assets, the sixth-lowest reading since 1998. US equity allocation sits at its highest overweight since December 2024, and sentiment overall is the third-most bullish since 2022. Managers spent the month rotating further into tech, banks and energy, and out of industrials and healthcare. The crowd and the leadership are now the same trade. Low cash predicts nothing, but it says something about capacity: the marginal buyer has already bought, so there is less money left to absorb a surprise and more people holding the same few names when one arrives.

The third story is that the AI debate has changed shape. This is the part most worth understanding, because it is no longer the argument people think it is. Almost nobody serious claims the demand is fake any more; the doubt has moved to the accounting underneath the earnings. Hardware depreciated over five to six years may only last two or three. Capex on some estimates will consume around 94% of hyperscaler operating cash flow. Hundreds of billions in signed-but-not-commenced data-centre leases sit off balance sheet. Interlocking vendor stakes and take-or-pay compute deals can make end demand look more independent than it is. That is a more dangerous kind of doubt than a demand question. It goes to whether reported profits are what they appear to be, and reported profits are what the constructive case rests on. None of it has to resolve badly, but it is the reason the thing to track is earnings revisions and not price.

Finally, the market keeps trying to broaden and keeps failing. Small caps lead on the year and lag worst on the quarter. The rotation happened, then un-happened, and they now sit under every intermediate average while the megacap complex holds the index up. Watch whether the next benign data point gets bought in small caps or in the same four names. That is what separates a market healing from one narrowing while the index looks fine.

The week ahead

• 2026-09-01ISM Manufacturing PMI (August)
• 2026-09-02ADP employment (August), JOLTS job openings (July), Atlanta Fed GDPNow (Q3)
• 2026-09-02Broadcom fiscal Q3 results after the close. Consensus near 3.24 EPS on 29.24bn revenue, with AI semiconductor revenue seen near 15.23bn against 5.2bn a year ago. Our local earnings cache lists 09-03; earnings dates move, so treat as scheduled.
• 2026-09-03ISM Services PMI, Challenger layoffs (August)
• 2026-09-04August employment report, 08:30 ET. Consensus near +45k after July's -23k, unemployment seen 4.1% to 4.2%, average hourly earnings +0.2% m/m. Also on our own audited local econ list.
• 2026-09-16FOMC decision, 2:00pm ET (meeting Sept 15-16), with the Summary of Economic Projections and dot plot. Hold versus a 25bp hike is priced near a coin flip; a cut is priced near 1%.
• 2026-11-03US midterm elections

The next five sessions

This week

Four macro prints and one earnings report arrive before Friday, and they arrive at a tape carrying almost no insurance. ISM manufacturing opens Tuesday, ADP and JOLTS follow Wednesday, and ISM services lands Thursday. The August employment report lands Friday at 08:30 ET, with consensus near +45k after July printed -23k, unemployment seen ticking from 4.1% to 4.2%, and average hourly earnings at +0.2% on the month. Broadcom reports Wednesday after the close, where the number that moves the complex is AI semiconductor revenue, seen near 15.23bn against 5.2bn a year ago.

What makes this different from an ordinary payrolls week is that the September decision is two-sided for the first time this cycle. That inverts the usual reaction function: normally a soft labour print reads as easing coming. Here a committee that has publicly put price stability first cannot obviously respond to weakness, so a cold headline stops being dovish relief and becomes a growth problem with no offset. A hot wage number feeds the hike directly, so both tails point the same way for equities. That is unusual, and it is why this horizon reads cautious while the trend itself stays intact.

Hold the second point alongside it: very little cushion is priced. Vol futures sit 13.90% below where they began the month and 13.21% under their own 50-day, and the BofA August survey has cash at 3.5% of assets. Low cash and low volatility are not safety; they mean the marginal buyer has already bought, with less left to absorb a surprise in either direction.

For the tape, the cleanest tell is not the index. SPY sits on its 20-day at 769, inside a box it has not left. Small caps are already moving, and IWM is below its 10-week, 8-week, 10-day, 20-day and 50-day lines at once, so the 297 cluster is a single reclaim level that either goes or does not. Watch where a benign print gets bought: in IWM the broadening is real, and in the same four megacaps it is another week of the index being carried.

The next four weeks

This month

September's shape is dominated by one scheduled event with a two-sided outcome. The FOMC meets September 15-16, with the decision at 2:00pm ET on the 16th. Futures and prediction markets treat it as close to a coin flip between a hold and a quarter-point hike. Kalshi had roughly 48% on the hike as of August 28. A cut sits near 1%, and that distribution is the month.

The detail that gets under-weighted is that this is a projection meeting. The Summary of Economic Projections and the dot plot come out with the decision, so even a hold is not neutral. The committee can leave rates alone and still reprice the path by moving the dots. A hold with a hawkish dot plot and a hike with a reassuring one are very different outcomes for long-duration equities. The binary framing of hold-versus-hike hides that.

July is the context here. The vote and the direction of the dissents are covered above; what matters for September is that a committee whose centre of gravity has already moved does not need new data to move again.

The data supports both sides, which is why this horizon does not deserve false precision. July core PCE at 3.3% year over year with headline at 3.7% is soft enough on the month at +0.2% to justify patience. It is high enough on the level to justify the dissents. Energy is the live input underneath. Crude is roughly flat on the month but sits 22.63% above its own 200-day on our proxy, with about 0.6 million barrels a day of Middle East supply expected offline through the end of next year on Hormuz constraints. An inflation impulse that comes from supply is the hardest kind for a central bank to answer. It is why the hawkish case has not gone away despite a benign monthly core print.

Structurally, September starts from a better place than July did. SPY is above an intact 50-day, not beneath a broken one, and August finished positive across all three indices. The month's real signal is whether price resolves outside the August range of 748.80 to 779.37. Inside it, nothing has been settled.

The next quarter

This quarter

Into Q4, an improving earnings line runs against the most stretched positioning of this cycle, and an election lands on top of both.

The constructive half is not in dispute. Index earnings estimates for 2026 have been revised higher across the market-cap spectrum, and AI-infrastructure names account for roughly half of this year's index earnings growth. Earnings revisions moving up is the most reliable support a market can have. It is currently present.

The other half is positioning, and the numbers are specific. The BofA August Global Fund Manager Survey shows US equity allocation at a net 27% overweight, the highest since December 2024. Global equity allocation sits at a net 56% overweight, the highest since November 2021 and a fourteenth consecutive month overweight. Cash is 3.5% of assets, the sixth-lowest reading since 1998. Goldman's risk-appetite indicator sits in the 99th percentile of readings since 1991, a level whose historical analogues show below-average forward returns out to twelve months. BofA's own reading of its survey is to reduce risk. Not because the economic view is wrong, but because positioning offers little protection if that view disappoints.

That is the tension: the direction is not the uncertain part, the range is. A quarter that begins near 21x forward earnings with record-low cash and a two-sided rate path has a wider distribution of outcomes than one beginning with cash on the sidelines, because the marginal buyer has already committed.

Breadth decides which tail it lands in. Leadership is narrow and concentrated in AI and energy, and the survey shows managers rotating further into tech, banks and energy while pulling back from industrials and healthcare. That concentrates the crowding instead of relieving it, and the observable test is simple. IWM reclaiming and holding above 305.18 while SPY holds 779.37 would say the earnings story has broadened beyond the names that have carried it. Until then, a single guidance disappointment anywhere in the capex chain re-rates the whole complex at once, because the same investors own all of it.

The next twelve months

This year

The year ahead turns on one question. Do the profits this market is priced for show up?

The case for them rests on spending that is contracted and visible. Published 2027 index earnings estimates cluster around double-digit growth, with one major house at 385 for 2027, roughly 13%. Hyperscaler capex is budgeted near 800bn this year and above 900bn next. Most published frameworks assume the multiple stays flat at roughly 21x. Notice what that means: the constructive case does not need re-rating, it needs the profits.

The serious bear case is no longer that AI demand is fake. By mid-2026 the argument had moved to the financial mechanics underneath it, and four strands deserve weight. First, depreciation. Hyperscalers write AI hardware down over five to six years while its useful economic life may be two to three, which critics estimate understates true depreciation by roughly 176bn across 2026-2028 and flatters reported earnings. Second, cash-flow coverage. PIMCO estimates combined hyperscaler capex will consume about 94% of operating cash flow across 2026-2027, leaving little room to fund the spending internally if revenue disappoints. Third, off-balance-sheet exposure: Moody's flags roughly 662bn of signed-but-not-commenced data-centre leases. Fourth, circularity. Interlocking vendor equity stakes, take-or-pay compute commitments and debt-funded GPU purchases among the largest AI vendors and their largest customers can make end demand look more independent than it is.

None of those four is a prediction, and none of them has to resolve badly. They matter because they are the mechanisms by which reported earnings could prove flattered, and that is the variable the constructive case depends on. So the way to track this horizon is not price and not sentiment. It is forward earnings revisions. If 2027 estimates keep rising, the mechanics are being absorbed. If they roll over, they were not.

Note the disagreement across this note: a cautious week sits inside a constructive year, and that is a coherent shape, not a contradiction. The near-term caution is about a dated, two-sided policy event landing on stretched positioning with no volatility cushion. The twelve-month view is about contracted spending and rising estimates. Different horizons carry different binding constraints, and a note that pretended they agreed would be less useful.

The levels

The levels that decide, if you only track three: SPY 779 above and 754 below, with 769 the pivot it keeps closing against. QQQ 718 is the line it sits under, 712 the 50-day beneath. IWM 297 is what small caps have to reclaim, and it is the weakest structure of the three.

• SPYsupport 753.96resistance 779.37Sitting essentially ON its 20-day (769.22), which makes 769 the pivot the week opens against, not a level it has to travel to. 779.37 is the ceiling: the 20-day and 60-day high, and where the most recent session was rejected after tagging 775.30 and closing below its open with a 51%-of-range upper tail. Below, 753.96 (50d) is the first structural shelf and 707.41 (200d) the floor. ATR14 is 4.37, about 0.57% of price, so a normal session covers roughly a third of the distance from here to 775. Open interest for the 08-31 expiry (OCC as-of 2026-08-27, not the latest book) is heaviest at 785 calls and 760 puts, with a second put shelf at 730. That is open interest only. We carry no greeks and this is not a gamma read.
• QQQsupport 712.05resistance 717.94The only one of the three below its 20-day (717.94), which makes that line the immediate hurdle instead of support, with just 0.62% of room down to the 50-day at 712.05. The weekly MACD histogram is still negative (-2.76), so the weekly frame has not repaired from July even though August was the strongest of the three on the month. ATR14 is 6.86, or 0.96%, the widest daily range of the three. It needs the least help to resolve either way.
• IWMsupport 292.4resistance 297.31The weakest structure and therefore the cleanest tell. It sits below the 10W SMA (297.22), 8W EMA (296.85), 10-day (299.53), 20-day (300.26) AND 50-day (297.31) after two consecutive red weeks, so they form a single cluster that has to be reclaimed for any breadth-repair argument to be made at all. Daily stochastic K at 0.8 is washed out, which cuts both ways: it is where bounces start and it is also what a persistent downtrend looks like early. It is still 9.31% above its 200-day at 270.56, so this is an intermediate break, not a structural one.

What would change it

Three things to watch, in checkable terms: SPY holding above 779 with IWM back above 297 would be the breadth repair that has been missing. SPY closing below 754, or IWM losing 292, points the other way. A September hike into this positioning is the third path.

Leadership

• Leading: QQQ and the AI-infrastructure complex (+4.13% in August, +16.90% YTD), Energy (USO +22.63% versus its 200-day), Gold on the month (GLD +10.05% MTD) though it broke hard in the final session
• Lagging: IWM and small caps (two red weeks, -1.4% on the week, -1.56% QTD, below every intermediate average), Long-duration Treasuries (TLT -2.51% versus its 200-day), QQQ on the QUARTER (-2.71% QTD) — leading the month, still repairing the quarter

August dispersion ran 2.57 percentage points between QQQ (+4.13%) and IWM (+1.56%), so the index a reader owned WAS their month; averaging them hides the story. The deeper split is that the year-to-date and quarter-to-date rankings invert. IWM leads YTD at +20.65% while lagging worst on the quarter, and QQQ leads the month while being the only one of the three red for the quarter. Leadership is not stable right now, and that instability is the main evidence for calling this a transition and not a trend. It also matters for the week ahead: BofA's August survey shows managers rotating INTO tech, banks and energy and out of industrials and healthcare, which concentrates the same crowding the tape is already showing.

Cross-asset

• GLD (gold proxy, NOT spot gold): week -3.42% · MTD +10.05% · vs 200-day -1.42%. Still up 10.05% on the month after giving back 3.42% on the week, with essentially all of that give-back in one session. Gold sold and the dollar bid together is a rates tell, not a growth tell. Track it through the payrolls print as the cleanest read on whether the hike is being priced further in.
• USO (WTI proxy, NOT spot crude): week -3.67% · MTD +0.41% · vs 200-day +22.63%. Flat on the month but 22.63% above its 200-day: the energy shock has plateaued at a high level instead of reversing, and that keeps headline inflation an active input to the September decision. WTI opened near 83.54 and Brent near 89.31 on Aug 28, with roughly 0.6 million barrels a day of Middle East supply expected to stay offline through the end of next year on Strait of Hormuz constraints.
• UUP (DXY proxy, NOT the dollar index): week +1.00% · MTD +0.04% · vs 200-day +2.11%. Bid on both the day and the week. A firmer dollar into a week of US data is the mechanism by which a hawkish repricing reaches earnings translation for the multinationals in the index.
• TLT (20y+ Treasuries): week +1.01% · MTD +1.17% · vs 200-day -2.51%. Positive on the WEEK but negative on the final session. The hawkish repricing is concentrated in that last session, so the weekly cross-asset numbers understate where the bond market finished. Do not read the +1.01% as the current state going into the payrolls print.
• IEF (7-10y Treasuries): week +0.03% · MTD +0.24% · vs 200-day -1.05%. The belly sold harder than the long end on the final session, which is the signature of a policy-path repricing, not a term-premium move. Watch the belly, not the long end, for the market's read on Sept 16.
• VIXY (VIX FUTURES ETF, NOT spot VIX; we carry no spot vol): week -3.02% · MTD -13.90% · vs 200-day -33.17%. The single most important cross-asset number going into this week: -13.90% month-to-date and -33.17% versus its 200-day means almost no volatility cushion is carried into a payrolls week with a live FOMC decision two weeks behind it.

Sources

• CNBC: Fed rate decision July 2026: divided Fed holds rates steady at 3.50-3.75% (9-3, three dissents preferring a hike)
• Federal Reserve Board: FOMC statement, 2026-07-29
• Federal Reserve Board: keynote remarks at the 2026 Jackson Hole Economic Policy Symposium, 2026-08-28
• CNBC: September Fed decision now a coin flip as hike odds increase after the Jackson Hole keynote (2026-08-28)
• FedRateCalc: 2026 FOMC meeting schedule: September 15-16, decision 2:00pm ET on the 16th with the dot plot
• CNBC: core PCE rose 3.3% annually in July (headline 3.7%, +0.2% m/m), released 2026-08-26
• Newsquawk: weekly economic calendar, 31 August – 4 September 2026
• Schaeffer's Investment Research: the week ahead: August jobs report takes center stage (payrolls consensus +45k, unemployment 4.2%)
• CNBC: three things we're watching in the week ahead, including Broadcom fiscal Q3 (2026-08-30)
• Bank of America August 2026 Global Fund Manager Survey: US equity allocation net 27% overweight (highest since Dec 2024), global 56% overweight, cash 3.5% (sixth-lowest since 1998)
• Investing.com: investors getting extremely bullish, BofA's August survey shows
• Goldman Sachs: S&P 500 forecast to climb as earnings growth powers stocks (2027 EPS, hyperscaler capex, 99th-percentile risk appetite, flat ~21x multiple)
• Silicon Analysts: hyperscaler AI capex and the depreciation wall (5-6 year schedules versus 2-3 year economic life)
• Epoch AI: hyperscaler capex versus cash flow
• State Street Global Advisors: two arguments against 2026 Fed rate hikes (the other side of the policy debate)
• Charles Schwab: 2026 mid-year outlook: narrow leadership, higher-for-longer rates, September seasonality
• Fortune: price of oil as of August 28, 2026 (WTI ~83.54, Brent ~89.31 open)
• US EIA: Short-Term Energy Outlook: Middle East shut-in crude, ~0.6 million b/d disruption through end-2027 on Hormuz constraints
• Prices, ranges, moving averages, ATR and open interest: our own Alpaca-SIP bars via friday\_wrap\_gather.py, bars through 2026-08-28; SPY option OI as-of 2026-08-27 (OCC lag).

Prices, ranges, moving averages and open interest in this note come from our own market-data bars through 2026-08-28. Every external claim names its source above. This note is forward-looking market research: it describes scenarios and the conditions that would confirm or invalidate them. It is not investment advice, not a recommendation, and states no forecast, price target or trade.