Brandon Leon

Week ahead · Macro · the jobs crack · CPI week · August 10, 2026

The labor market blinked first.

Author Brandon Leon Posted 2026-08-09 (figures as of the August 7 close) Coverage Week ahead · the trip-wire ledger · the deployment gates · CPI week

TL;DR: Nine days after three FOMC members dissented in favor of a rate hike, July payrolls fell by 23,000 — the first negative monthly print of this expansion — while May and June were revised down by a combined 103,000 jobs. September hike odds fell to roughly 40%. Our wires responded as written: the sub-100K soft branch triggered a duration review; the power-stock gates split, with CEG moving above its entry ceiling and VST meeting its price condition only after a 9% decline. We bought neither. The cash remains intact ahead of Wednesday's CPI report and Thursday's $25 billion 30-year Treasury auction. Five charts below; the wires reset.

1. The door everyone was watching

All summer the risk had one name. Inflation ran five years above target. A new chair arrived promising “no soft inflation target — not on this Committee's watch.” Three fellow FOMC members dissented in favor of a hike eleven days ago, and the 30-year closed at levels last seen in 2007. Our primary macro wires assumed the next shock would come through the inflation door. So did everyone's.1

Friday, 8:30 a.m., the knock came from the other side of the house. The July employment report didn't miss the +83,000 consensus; it printed negative. Twenty-three thousand jobs lost, the first negative monthly print of this expansion. The revisions compounded the weakness: May and June, the months supporting the hawkish case, were revised down by a combined 103,000 jobs — to +129,000 and +57,000, respectively. In one release the three-month picture went from cooling to stalling. Unemployment held at 4.1% only because government payrolls shrank by 53,000 while private hiring was modest at +30,000.1

Chart 1 — The knock at the wrong door

Three months of payrolls: the story the hawks were telling, revised out from under them — and July falling clean through our wire.

Monthly payrolls (thousands) the wire: sub-100K = re-examine +129K +57K −23K May (rev. −66K) June (rev. −37K) July First negative monthly payroll print of this expansion. Unemployment 4.1%; private +30K, government −53K.

BLS July employment report (Aug 7) with prior-month revisions. The soft branch of our trip-wire #1 required a sub-100K print; July cleared it by 123 thousand jobs.1

The market's reaction is Chart 2, and it barely needs narration. A September hike went from near-certain to a coin-flip to fading, in the space of ten days. Morgan Stanley's desk put it plainly: the jobs print eases the pressure, “but next week's inflation data will still likely be the deciding factor.” Inflation didn't go away. It just stopped being the only problem.2

Chart 2 — A regime unwinding in three steps

September hike odds: near-certain, then a coin-flip, then fading — in ten days.

Market-implied odds of a September hike (%) ~76–82% ~57–60% ~40% pre-FOMC (Jul 28) after the 9–3 hold (Jul 29) after −23K jobs (Aug 7) Ranges reflect differing snapshots across futures/prediction markets; the direction is the story.

CME FedWatch and futures snapshots as reported at each event. The wire we owe on this: the soft branch fired, so the duration re-examination happens now, in writing — Section 4.2

Simply: Everyone spent the summer bracing for bad inflation news. The shock came from the other direction: America lost jobs in July, and the two prior months were quietly revised way down. Markets went from expecting a September rate hike to doubting one entirely. Wednesday's inflation number will go a long way toward telling us which problem the Fed actually has.

2. Meanwhile, at the gates

The jobs shock landed in the middle of our own busiest stretch of the summer. Two weeks ago we published entry rules for the power stocks, the first real buys of the deployment series, and both companies happened to report this week — back to back, like a controlled experiment someone ran for us.3

Thursday morning, Constellation did everything right. Earnings beat, full-year guide raised to $11.50–$12.50, the Calpine deal closed, 920 megawatts of new nuclear contracts. Its reward was a stock price that rose straight through our $263.56 ceiling to $269.89, which put it above our entry ceiling. Guidance ✓, price ✗. Friday morning, Vistra reaffirmed guidance, beat on EBITDA under ugly revenue optics, and fell 9.2% into the jobs-day tape to $140.59. It met its price gate — five dollars above its own published $133 stop, in the middle of one of the year's sharpest macro shocks.

We bought neither, and the cash is why this section is short. Buying Vistra on Friday would have been rule-compliant. It also would have meant deploying into a stock trading within 5% of its stop, hours after the labor market cracked, four sessions before the CPI print that will be the next major test of the rate backdrop for the utility trade. The plan's own half-cash rule holds everything through Wednesday anyway. The gates re-arm Thursday: Constellation below ~$264, Vistra per its standing gate — with position sizes that treat that nearby stop as the feature it is. Patience cost us nothing this week except the itch.

Simply: The two power companies we've been waiting to buy both reported. One did great but its stock rose past our price limit; the other passed our test only by falling 9% on the worst day, landing uncomfortably close to our sell line. We bought neither — Wednesday's inflation report will tell us whether this whole trade gets easier or harder, and waiting four days for that answer is cheap. The buy signals reset Thursday, after the number.

3. AMD did everything right, and it didn't matter

The week had actually opened well for the AI trade. Palantir reported Monday night: revenue up 93%, guidance raised again, up 15% after hours — the proof-gets-paid pattern we've charted since the hyperscalers, still working. Then Tuesday, AMD posted the best numbers in its history — record $11.5 billion in revenue, data-center sales up 107% — and fell 9% anyway. Nothing was wrong with the quarter. At 54 times forward earnings, the quarter was already in the price.4

So the axis gets its second amendment in two weeks. First: self-funded proof gets paid, externally-funded spend gets discounted. Now: proof gets paid at a price. Our own gates taught us the same thing at Constellation two days later. A great company is not a great trade at any number.

Chart 3 — Two weeks of exams

Print reactions across the AI stack: proof got paid — until the multiple got graded too.

Post-print reaction (%), July 22 – August 7 MSFT+16.9% PLTR+15% AH AMZN+9% AH GOOGL−7.1% META−8.6% AMD−9% AH VST−9.2% Green = paid; terracotta = sold. AH = after-hours reaction; others = print-day or next-day close. AMD's −9% came on record results — the market graded the stock's 54x forward multiple, not the quarter.

Reactions as reported per print: MSFT (Jul 30 close), PLTR/AMZN/AMD (after-hours), GOOGL (Jul 23), META (Jul 30), VST (Aug 7). Self-funded proof got paid; rich multiples and externally-funded spend got sold.4

Simply: Earnings season's pattern held, then sharpened. Companies proving their AI spending earns money keep getting rewarded — unless the stock was already priced for perfection. AMD posted spectacular numbers and still fell 9%, because at 54 times earnings, spectacular was already the assumption. Same lesson our own buying rules taught us this week. The next name on our exam list is Synopsys, August 26.

4. Oil walked it all back

The month's quietest move may matter most on Wednesday. WTI closed above $92 in late July, Brent went through $100, Hormuz was shut, and our escalation wire was armed. Three weeks later, WTI sits near $78 and Brent near $82 — below where they stood before the FOMC, most of the war premium gone. Our oil wire never fired in either direction. The deflation is now a live input: it's a big reason Wednesday's CPI consensus is only +0.1%, and if Brent closes below ~$80 the round trip completes just as the Fed needs it to.6

Chart 4 — The round trip

WTI's war premium: three weeks up the mountain, three weeks back down — just in time for CPI week.

WTI crude, settles ($/bbl) $82.49 — where the spike began (Jul 17) $92.19 (Jul 23) — through our $90 wire $79.26 (Jul 28) ~$78 (Aug 7) Jul 17 late July Aug 7 Key settles drawn; intermediate sessions omitted. Brent traced the same arc: $100.69 peak (Jul 23) to ~$82 (Aug 7).

WTI settles as previously footnoted (Jul 17–29) and Forbes/Fortune oil trackers (Aug 4–7). The premium's exit is a disinflation gift arriving the week the CPI prints.6

The book waited with us. The DRAM basket at $50.60 is the sore spot, 15% under the fill, held on Tuesday's underwrite with the China-spot tripwire quiet. ASML round-tripped to $1,740.99, within a percent of our fill; the add-zone visit came and went unexercised. The quantum sleeve sits at $153.46. Micron, watched from the bench, at $877.57. And the cash ended the week where it started: intact, earning its 4%, having said no twice.5

Chart 5 — The book, marked

What we hold, against what we paid — and the cash that refused two gates this week.

DRAM basket $50.60 · −15% vs fill held — underwritten Tuesday; tripwires quiet ASML $1,740.99 · −0.7% vs fill add-zone came and went — unexercised Quantum sleeve (QTUM) $153.46 venture-sized; QC-ADDS specs ~Sept 20 CASH — the whole tranche intact · ~4% bills held through Wednesday's CPI, by rule

Marks at the August 7 close vs. disclosed fills, per the bpleon price feed. Micron ($877.57) is watched from the bench — re-entry by published triggers only.5

Simply: Oil quietly gave back its entire war spike — good news for Wednesday's inflation number. Our holdings: the memory fund is down 15% (held, with published warning signs), ASML is back to breakeven, the small quantum position is fine, and the cash earned its 4% while saying no to two tempting buys. One thing to watch: if Brent slips under $80, Wednesday's inflation number gets even friendlier.

5. Wednesday, 8:30 a.m.

Everything now runs through one number. July CPI prints Wednesday morning, consensus just +0.1% because the oil spike round-tripped inside the month — and for the first time all summer, the same report reads differently depending on which door you think the trouble comes through. A cool print after negative payrolls says the hiking argument is finished, and maybe that something is genuinely slowing. A hot core print after negative payrolls is the trapped-Fed branch, where Warsh's “family fight” reconvenes in September with both sides holding evidence. Thursday at 1:00 p.m. ET, the Treasury auctions $25 billion of 30-year bonds into whichever answer printed the day before.2

The owed response, delivered: the jobs wire's soft branch fired, so here is the duration re-examination, in writing, as promised. We stay below-benchmark into Wednesday — not from inflation conviction anymore, but because a two-sided macro with a $25 billion supply test pending argues for optionality over heroics. If core comes in at or under 0.2% and Thursday's auction clears without a tail, the below-benchmark stance has lost its argument and we size duration back toward neutral — dated, in next week's ledger, graded like everything else we've ever promised here.

The summer's story was a Fed daring inflation to blink. The labor market blinked instead — and Wednesday morning decides whether that was the plot twist or the ending.
This week's wires · responses pre-committed

#1 · CPI, Wednesday. Core ≤0.2% = the hike cycle is likely done: duration sizes toward neutral, gates re-arm Thursday. Core ≥0.3% with jobs cracking = the trapped-Fed branch: nothing deploys, defense deepens. #2 · The 30-year auction, Thursday 1:00 p.m. ET. A clean auction confirms CPI's story; a tail of more than roughly 3 basis points on the hot branch is the credibility alarm. #3 · The re-armed gates. Cool branch only: CEG below ~$264, VST per its standing gate, sized so that a stop-out at $133 is a contained, pre-defined loss. #4 · Oil's floor. Brent closing below ~$80 completes the round trip and flatters the next CPI; the escalation clause (>$100, or a confirmed production strike) stands.

Simply: Wednesday's inflation report is the key near-term catalyst. Mild = the rate-hike scare likely ends, and we likely start buying the power stocks Thursday. Hot while jobs shrink = the Fed's nightmare scenario, and we stay defensive with our 4% cash. Thursday's big bond auction then grades the market's confidence either way. All the responses are written down before the number prints — that's the whole point of this series.
Get the next note by email →   All writing

Sources & footnotes

  1. The July employment report (BLS Employment Situation, Aug 7): payrolls −23K vs ~+83K consensus — the first negative monthly print of this expansion; May revised −66K to +129K, June −37K to +57K (−103K combined); unemployment 4.1%; private +30K, government −53K (NBC News; CNBC; Fox Business). Summer context as previously footnoted: FOMC 9–3 hold with Hammack/Kashkari/Logan dissenting for +25bp (Jul 29); Warsh “no soft inflation target” (press-conference transcript); the 30-year's first 5.2%+ close since 2007 (Jul 29). ↩
  2. Fed repricing: September hike odds >50% pre-report → ~40% after (CNBC, Aug 7); pre-FOMC ~76–82% and post-hold ~57–60% as previously footnoted (CME FedWatch/Bloomberg/Polymarket snapshots). Morgan Stanley quote as reported (Aug 7 coverage). 30-year ~5.17% (Aug 5; FRED DGS30). August refunding: $125B held steady — 3yr $58B (Aug 11), 10yr $42B (Aug 12), 30yr $25B (Thu Aug 13, 1:00pm ET) (U.S. Treasury quarterly refunding statement). July CPI: Wed Aug 12, 8:30am ET; consensus +0.1% m/m (Bloomberg survey); PPI Thursday; retail sales Friday. ↩
  3. The gates: CEG Q2 (Aug 6 pre-open): adj. operating EPS $2.55 vs ~$2.33 est; revenue $7.50B vs ~$7.70B; FY guide raised to $11.50–$12.50 from $11–$12; Calpine close, higher capacity revenue, 920 MW of new long-term nuclear PPAs (company 8-K; earnings-call transcript). VST Q2 (Aug 7): revenue $4.0B and net income $305M below estimates (largely hedge-accounting optics), adjusted EBITDA $1.8B vs ~$1.64B est (+30% y/y), FY26 EBITDA/FCF guidance reaffirmed ($6.8–7.6B / $3.925–4.725B); shares −9.2% to $140.59 (company release; PRNewswire; coverage). Gate terms as published in the series opener: guidance intact + price ≤ the Jul 30 reference (CEG $263.56 / VST $148.62). Deployment status: no executions this week — cash intact, per the half-cash-through-CPI rule. ↩
  4. The AI prints: AMD Q2 (Aug 4): record revenue $11.5B (+50%), data-center $6.7B (+107%, 58% of revenue), non-GAAP EPS $1.66 — sold ~7–9% after-hours/next-day on ~54x forward valuation and margin concerns (company release; Investing.com; TradingKey). PLTR Q2 (Aug 3): revenue $1.94B (+92.8%), U.S. commercial +149%, FY guide raised; +~15% after hours (Fortune; CNBC). Prior-week reactions (MSFT +16.9%, AMZN +9% AH, GOOGL −7.1%, META −8.6%) as previously footnoted. ↩
  5. Marks (Aug 7 closes, bpleon price feed): DRAM $50.60 vs $59.73 fill; ASML $1,740.99 vs $1,752.87 fill; QTUM $153.46; MU $877.57; VST $140.59; CEG $269.89; GOOGL $354.30; AMD $483.36. China-spot tripwire status per DigiTimes weekly tracking — no flip. ↩
  6. Oil: WTI settles $82.49 (Jul 17), $92.19 (Jul 23 peak, through the published ~$90 wire), $79.26 (Jul 28), $84.46 (Jul 29) as previously footnoted (exchange settlements/Reuters/CNBC); ~$78.31 and Brent ~$82–83.5 in the Aug 4–7 window (Forbes Advisor / Fortune oil trackers). Brent peak $100.69 (Jul 23). ↩

Methodology: marks are settled closes vs. disclosed fills; wires are graded against their published wording; responses are pre-committed before the events that test them. Single-source figures are attributed inline. Nothing here is a price target.

Nothing on this page is investment advice. We work in scenarios and trip-wires, not predictions. See disclaimer.