US Market Recap | 2026-08-07 (ET) Friday

Coverage window: the 2026-08-07 09:30–16:00 ET regular session, plus the 16:00–17:40 ET after-hours session. Quote basis: closing prices and intraday highs/lows for indices, sector ETFs and single names all come from the CNBC quote API (exchange basis), snapshot taken at 17:40 ET; all percentage changes are measured against the 8/6 close. Single-name percentage changes have been cross-checked against the stockanalysis.com gainers/losers list for the day and match (TEAM +35.31% / DOCS +32.62% / PUBM +31.90% / SEZL −33.89%, all four compared one by one with no discrepancy). Data not obtained for this issue: NYSE advancing/declining issues (only Nasdaq obtained), single-name dollar volume and bid-ask spreads, NYSE new highs/new lows. The related passages are treated as "not obtained" — no estimates are made.

---

0. One-Sentence Recap

Today was risk-on, and a "breadth-type" risk-on at that — but with a stagflation hedge sitting underneath it.

July nonfarm payrolls at −23,000 were read by the market as a complete removal of the rate-hike threat: the S&P 500 rose +0.62% to 7,757.64, an all-time closing high; the Nasdaq Composite +1.30%; the Russell 2000 +1.10%; while MSFT +0.03%, AAPL +0.29%, META +0.37%, GOOGL −0.96% — megacap tech was essentially flat across the board. The Nasdaq's 1.3% gain was not driven by the heavyweights, which means what rallied was the "mid-tier" of mid-cap software and semis: Nasdaq advancers vs decliners came in at 1,537 : 665 (about 2.3:1), a complete reversal of the 8/6 structure of "flat index, 57% of names down internally."

The strongest theme was software — and stronger than the pre-market brief expected. IGV (software ETF) +3.29% was the single strongest of all the sector ETFs, well ahead of SMH +1.96% and XLK +1.42%; the tail of the sector table was XLE −1.13%, XLF −0.36%. The pre-market call of a "SaaS narrative-falsification repricing" not only held, it spilled over from single-name events into a sector move (MDB +7.78%, SNOW +3.93%, ZS +3.74%, OKTA +3.35%, CRM +3.20%).

Pre-market list hit rate at a glance: of 15 independent conclusions, 9 clearly delivered, 4 partially delivered, 2 clearly wrong (SPCX underrated, APP short direction wrong), plus 1 major miss (TWLO, closed +24.89%). The prettiest call was ABNB (+17.43%, closing at a 52-week high and at the high of the day); the most valuable one was the "watch only" on DOCS — it opened +88%, hit +94% intraday, and closed with only +32.62% left, a 29.5% decline from the open.

Next-session tone: constructive but reduce the slope of position building. VIX closed at 14.90 (−1.65%), the 2-year Treasury at 4.199% (−4.6bp), the 10-year at 4.649% (−2.1bp), and the dollar index at 99.604 (−0.33%) — the rate tailwind is real. But COMEX December gold rose +2.37% to $4,401.30, a big gain on the very same day the index made a new high. That is not a pairing risk-on should produce; it is pricing "weak growth + supply-driven inflation." Next Wednesday's (8/12) 08:30 ET July CPI is the adjudication point for this divergence.

---

1. Market Overview

Indices and Volume

Week to date: S&P +3.58%, Nasdaq +5.19%, Dow +2.96% (Reuters basis¹). The Nasdaq posted its strongest week since April.

Breadth

⚠️ Breadth is the one item most worth remembering today. The 8/6 structure was "the S&P down only 0.18%, but 59% of NYSE and 57% of Nasdaq names lower" — the index shielded by the heavyweights while the internals weakened. Today is the exact opposite: the heavyweights (MSFT/AAPL/META/GOOGL) were essentially flat, yet the index made a new high, and the Nasdaq's 2.3:1 advance ratio says the mid-tier did the pushing. Section 8 of the pre-market list flagged "deteriorating breadth" as a risk signal; today that signal is cleared.

Rates, FX, Volatility, Commodities

Three readings:

1. The front end fell more than the long end, steepening the curve by 2.5bp — this is the standard pricing of "policy path turns dovish, term premium unchanged." What got stripped out of hike expectations is the stretch before 2027, not the whole curve.
2. ⚠️ The 10-year backed up from 4.623% at 08:33 to 4.649% at the close, retracing 2.6bp intraday. The first reaction to payrolls was the most dovish moment; the long end walked back over the session — the market did not read this print as "recession," only as "no hike." That matters for next week's CPI: the long end is not pricing inflation coming down.
3. ⚠️ Gold +2.37% happening at the same time as a record index close is the most incongruent pair of signals today. The pre-market list had noted that "gold +1.89% and equity futures +0.18% are diverging; they cannot both be right" — today's answer is: both rose on the day, but they are buying two sides of the same thing. Equities are buying "the Fed can't hike"; gold is buying "the Fed can't hike, but the inflation is coming from an energy supply shock and can't be pushed down." The name of that combination is a stagflation hedge, not risk-on.

Sentiment Assessment

Qualitatively: risk-on with healthy breadth, but not a "growth is back" risk-on — a "policy constraint removed" risk-on.

The evidence is in the sector dispersion: XLF −0.36% (banks pressured by the falling front end), XLE −1.13% (oil pulled back), while XLY +1.49%, IGV +3.29%, SMH +1.96% led. If the market were really trading "the economy is improving," financials and energy would not be the only two sectors down. The market is trading the discount rate, not the numerator.

---

2. Pre-Market List Reconciliation

Basis: the 8/6 close. The "pre-market" column is the pre-market list's 08:37 ET snapshot.

2.1 Long Side (Single-Name Strength Ranking)

2.2 Negative / Avoid List

2.3 Hit Rate and Self-Review

A one-line self-review:

What really deserves to be remembered today is not the 86.7% figure, but that "the most expensive call right" and "the most expensive call wrong" happen to be two sides of the same capability.

The right side: the pre-market judged DOCS at +85% as "watch only," TEAM at +31% as "watch closely," and put ABNB at +7% in the No. 1 slot — and the market executed those three rankings to the letter today. DOCS was −29.5% from the open, TEAM was gap-and-go, ABNB closed at the high — the ranking criterion, "how well the size of the gap matches the fundamental substance," proved to be effective.

The wrong side: this method only works on names already on the list; it cannot protect me from the names I missed. I gave SPCX 50 pts and "watch only" on the grounds of "no earnings catalyst" — while on the day it had an Argus upgrade, a Terafab plant plan, and an Nvidia exclusivity deal from three days earlier, none of which I checked; TWLO, inside my coverage window (8-K filed after the 8/6 close), published results that raised full-year revenue, operating profit and free cash flow all at once, and I did not even mention its name. The proudest assertion in that pre-market piece — "verified against the SEC originals, only ABNB of the four had a genuine guidance raise" — was drawn from a sample that had dropped TWLO. No amount of verification depth can make up for a gap in the sample.

Lesson: verification depth (vertical) and coverage breadth (horizontal) are two independent failure modes, and yesterday I put the entire budget into the former.

---

3. Theme Validation

3.1 Theme Reconciliation Table

3.2 ❌ The Missed Surprise Theme: Orbital AI Compute (SPCX)

The pre-market list characterized SPCX as "a lockup-supply event, no earnings catalyst, total score 50, watch only." That characterization is incomplete on three levels:

⚠️ Attribution discipline (this must be spelled out, otherwise it is stitching old announcements onto today's tape): ③ the Nvidia exclusivity deal and Starmind were announced on August 4, not today's news; Nvidia was up about 3% and SPCX about 9% at the time, and that round was already priced. Today's (8/7) +15.83% has ① the Argus upgrade as its direct catalyst, with ② and ③ as continuing background narrative. Writing an 8/4 announcement up as "today's positive" is misattribution, and this issue does not do that.

This theme matters more for the next session than for today: it extends the "AI compute" narrative from "power and cooling bottlenecks in ground data centers" into orbit, and next week happens to be neocloud earnings week (CoreWeave, Nebius, Cerebras, SMCI) — see Section 6.

3.3 ✅ The Most Precise Line: Ad Tech's "Demand Side Collapses, Supply Side Rallies"

Item 5 of the pre-market list read: "do not extrapolate TTD's crash into 'the ad market has collapsed' and short the whole chain." Today's data validated that sentence down to the decimal:

Same day, same value chain, −21.9% and +31.9% side by side — this is not a sector move, this is a company-level share shift.

⚠️ But this issue must also execute the second half of its own sentence: the pre-market said "cannot be extrapolated to the whole industry" while simultaneously putting APP into "short watch (demand-side theme resonance)" — that is self-contradictory, and the market punished the contradictory half (APP +3.32%). The correct formulation should be: TTD's problem is TTD's; it does not resonate to PUBM, and it does not resonate to APP.

3.4 ⚠️ The Missed Single Name: TWLO (Twilio), and It Overturns the Pre-Market's Core Assertion

Fact check: Twilio's 8-K was filed on 2026-08-06 (SEC EDGAR accession 0001447669-26-000088; the 10-Q was filed on 8/7). That falls inside the coverage window declared by the pre-market list (8/6 16:00 ET → 8/7 08:34 ET), so it should have been covered and was not.

Why this matters — it is not "one more limit-up name I failed to recommend," it knocks out the pivot of that pre-market piece:

The core assertion in item 2 of Section 0 of the pre-market list was: "the only name with what can genuinely be called a substantive full-year raise is ABNB," and ABNB was ranked No. 1 on that basis. But TWLO's upgrade is harder than ABNB's on all three dimensions:

1. ABNB raised a revenue growth bracket ("low-to-mid double digits" → "at least mid double digits") and the EBITDA margin (35% → "at least 35.5%"); TWLO raised the absolute ranges for revenue, operating profit and free cash flow — three items.
2. ABNB's +17% revenue growth includes roughly 3–4pp of FX tailwind, and net profit includes a $77 million one-off tax benefit (disclosed pre-market); TWLO raised the organic-basis guidance on its own by 3.5pp (9.5–10.5% → 13–13.5%), and that basis by definition already excludes M&A and FX.
3. ABNB's Q3 margin is guided slightly lower y/y; TWLO's operating profit and FCF guidance were both raised in the same direction.

⚠️ This is a "conclusion correct but sample incomplete" failure: the pre-market compared SEC originals line by line and overturned the "upgrade" narrative at TEAM / NET / DOCS, and all of that work was validated by the market today (TEAM rose on repricing, NET fell back sharply, DOCS topped out at the open). But the universal claim "only ABNB" holds only inside that closed sample of 8 names. The correct wording should have been "among the 8 names covered in this issue, only ABNB…" — the universal quantifier is the error this issue made, not the verification work.

The market's own ranking confirms this: the two names that genuinely had full-year guidance raises (TWLO +24.89%, ABNB +17.43%) both rallied hard and closed near their highs; while among the three that rose on valuation repricing, DOCS fell 29.5% from the open and NET fell 5.6% from the open. The difference between an upgrade and a repricing showed up more clearly in the intraday path today than in the closing percentage change.

3.5 A Methodology Note: Today, "Closing Change" and "Intraday Path" Give Opposite Rankings

Ranked by closing change, the repricing group (TEAM/DOCS/PUBM) sweeps the top three; ranked by the path after the open, the genuine-upgrade group (ABNB) and the new-catalyst group (SPCX) come out in front instead, while two of the repricing group are negative after the open.

This says that on "day one after earnings," the closing gain comes almost entirely from the gap, not from that day's trading. What can actually be carried into the next session is the stretch that is still moving up after the open. This criterion is more direct than the "size of the gap" used by the pre-market list, and I suggest folding it into the next pre-market list's validation signals.

---

4. Post-Close Earnings Moves (16:00–17:40 ET)

⚠️ There were no heavyweight earnings after today's (Friday) close, so this section has no substantive content — that is a fact, not missing data.

All 20 focus names snapshotted at 17:40 ET moved within ±2.5% after hours, on extremely thin volume:

Reading: these small after-hours gains (PUBM +2.25%, DOCS +1.35%) do not constitute a catalyst for the next session — they are thin post-auction prints with directional meaning close to zero. The Friday after-hours void means the pricing input for the next session's (Monday's) open is the weekend news flow, not post-close earnings.

The next real batch of post-close catalysts comes after Monday 8/10's close (ASTS, BBIO, HIMS, RKLB) — see Section 6.

---

5. Flows and Sentiment

5.1 Sector Rotation (SPDR Sector ETFs + Thematic ETFs)

Three features of the rotation structure:

1. Software > semis > large-cap tech — IGV outperformed XLK by 1.87pp. That ordering says money is buying mid-cap long-duration software, not the tech heavyweights. It is the purest expression of "rates down → duration assets benefit."
2. Financials and energy were the only two sectors down — that is the test for a "discount-rate tape" rather than a "growth tape." If the market were trading a recovery, banks (procyclical + credit) and energy (demand) would not be at the bottom.
3. Defensives (XLP +0.01%, XLU +0.53%, XLRE +0.38%) all underperformed — the risk-on is real; money is not positioning defensively.

5.2 Heavyweights vs the Mid-Tier

⚠️ The Nasdaq rose +1.30% while four of the six largest weights gained less than 0.4% and one fell. Nearly all of this advance was contributed by things outside the heavyweights. Combined with the Nasdaq's 2.31:1 advance-decline ratio, today's breadth is the healthiest of the week — a mirror image of the 8/6 structure of "index shielded, internals falling."

But note what produced the healthy breadth: it came from the broad benefit of falling rates for long-duration small and mid caps, not from a broad improvement in earnings. This kind of breadth is equally sensitive to rates moving the other way — if next Wednesday's CPI comes in hot, the mid-tier that rose the most today will fall the most.

5.3 Risk-On / Risk-Off Assessment

Conclusion: risk-on (of the policy-constraint-removed type), but with an explicit layer of stagflation hedging on top.

In one line: today the market deleted the "rate hike" risk, but it did not delete the "inflation" risk. What gold is buying is exactly that difference.

---

6. Next-Session Outlook (Monday 2026-08-10)

① Theme Continuity

② Next-Session and Next-Week Calendar

⚠️ Structural note for this week: today is Friday, so the next session is Monday 8/10. With no earnings after Friday's close, Monday's opening input is the weekend news flow rather than post-close catalysts.

🔴 The 8/12 CPI is what this issue considers the most important event of next week, because it directly adjudicates the legitimacy of today's tape:

• Consensus: headline easing to 3.4% y/y, core 2.5% (June headline was −0.4% m/m, with y/y falling from 4.2% to 3.5%).
• The asymmetric risk is in core: if core rebounds toward 2.9%, the probability of a September hike gets repriced, and the "policy constraint removed" that was bought today has to be handed back.
• ⚠️ Today's market structure is very poorly protected against that risk: VIX at just 14.90, gains concentrated in long-duration mid-cap software, financials already falling. Once rates reverse, whatever rose most today falls most.
• Gold +2.37% is already voting that "CPI will not be that obedient" — this issue considers it the leading signal most worth watching.

③ Focus List (Ticker + Validation Point)

④ What to Avoid

1. Chasing today's extreme gap names (DOCS +32.62%, PUBM +31.90%, TEAM +35.31%) — DOCS already demonstrated it once today: buying at the open lost 29.51% on the day. All three of the top gainers came from the gap rather than from intraday trading, and the next day's pricing starting point is the closing price, not those pretty closing gains.
2. Adding to long-duration growth ahead of the 8/12 CPI — today's entire advance rests on the single assumption of "no hike in September," and with VIX at just 14.90, protection against that assumption being overturned is close to zero.
3. Shorting TTD — the negative holds, but with the 52-week low already broken + 17.61% short interest + a 7.56% bounce off the low today, the odds are no longer on the short side.
4. Extrapolating TTD's problem into shorting APP or the whole ad chain — this issue already missed once today because of exactly that (APP +3.32%). PUBM +31.90% and APP +3.32% both prove: this is TTD's company-specific problem.
5. Energy (XOM / CVX / OXY) — XLE was the weakest today, the Hormuz clauses have not landed, direction is decided by negotiations.
6. Storage / AI hardware (WDC / SNDK) — two straight down days, ebb confirmed.
7. Chasing SPCX — the catalyst is a single sell-side upgrade; the lockup window for 911.5 million shares runs to 12/8.

⑤ Input Notes for the Next Pre-Market List

1. ⚠️ Coverage breadth takes priority over verification depth (the most important item in this issue). Today's lesson is TWLO: an earnings report that fell inside the coverage window, raised three full-year guidance items at once, and rose +24.89% on the day was missed entirely, while the same report did extremely deep SEC-original verification on TEAM/NET. The first step of the next issue should be "pull the full list of every 8-K filed inside the window from SEC EDGAR first, then decide who to dig into," rather than reverse-engineering the sample from media headlines.
2. ⚠️ Ban universal quantifiers. "Only ABNB raised guidance" should be written as "among the 8 names covered in this issue, only ABNB…". The credibility of the verification work cannot be lent out to the completeness of the sample.
3. Fold "open→close" into the validation signals. Today proves that the closing gain comes almost entirely from the gap, and it is the path after the open that separates a "genuine upgrade" from a "pure repricing" (ABNB +8.12% vs DOCS −29.51%). I suggest the next issue write this criterion explicitly into its validation signals.
4. Set up a theme tracking level for SPCX. It is no longer "a lockup-supply event," it is the vehicle for the "orbital AI compute" theme; Monday's ASTS / RKLB earnings are the first validation point.
5. Next Wednesday's CPI should be front-loaded as the framing variable for the whole week. Today's sector structure (software leading, financials and energy at the bottom, gold sharply higher) is essentially an early bet on CPI, and the pre-market list should carry "if CPI comes in hot, whatever rose most today falls most" as a risk warning running through the whole week.
6. Keep the discipline of "no causal judgment when information is insufficient." SEZL fell −33.89% today (the 6th-largest decline in the market), and the pre-market explicitly wrote "earnings details not obtained, no causal judgment" — that was right; do not go back and retrofit a cause just because it fell afterwards.

---

Operations Log (not sent to clients)

Data-sourcing status this run: yfinance was entirely unavailable; the CNBC quote API single-handedly carried all the market data.

• ⚠️ yfinance threw YFRateLimitError from the very first call, and could not even get a crumb for a single Ticker('SPY').history() (the error point is data.py:_get_crumb_basic). Unlike the 8/7 pre-market run where "batch calls failed but single names could be worked around," this time it was rejected at the cookie/crumb handshake stage, so splitting requests does not help.
• Leftover processes have been ruled out: ps aux showed only 6 Python processes, all system services (networkd-dispatcher, unattended-upgrade) plus this project's uvicorn API — no stuck yfinance leftovers. Conclusion: this was pure IP-level rate limiting, not a local process problem, consistent with the second root cause recorded in memory under yfinance-rate-limit-root-cause.
• ✅ The CNBC quote API was stable throughout, returning indices / rates / FX / sector ETFs / single names in a single request, and the ExtendedMktQuote field gives the after-hours price and after-hours change directly, saving a separate after-hours lookup. The | separator must be encoded as %7C (done per memory, worked first try); the @ in the futures symbols @CL.1/@LCO.1/@GC.1/@SP.1/@ND.1/@DJ.1 must also be encoded as %40, otherwise the whole symbol string fails to parse — this is not in memory yet and should be added.
• CNBC's open/high/low/yrhiprice/yrloprice fields are the key increment for this issue: it is exactly those that gave DOCS's open of $38.87 / high of $40.00 / close of $27.40, which made the most valuable reconciliation conclusion in the whole issue — "−29.51% from the open" — writable at all. Taking only the closing change would have missed this completely (a +32.62% close looks like a successful recommendation). I suggest making OHLC a mandatory field for recap data pulls.
• Cross-validation: the stockanalysis.com gainers/losers list and CNBC agree exactly on the percentage changes for the four overlapping names (TEAM/DOCS/PUBM/SEZL); dual-source confirmation passed.
• WebFetch failure list (same as pre-market, can now be treated as a stable reproduction): Benzinga (403), CNBC article pages (403), thestreet (403), wsj.com (rejected at the tool layer). stockanalysis.com's gainers/losers pages fetch fine and are currently the best source for the gainers/losers list.
• SEC EDGAR data.sec.gov/submissions/CIK*.json came through again: it confirmed TWLO's 8-K filing date as 2026-08-06, which nails the "miss" down as an in-window miss rather than an out-of-window event — without that step it would have been easy to let myself off the hook with "TWLO was outside the coverage window." This technique should be made standard.
• ⚠️ Attribution discipline triggered an interception once (memory ai-attribution-stitches-old-news in effect): while searching SPCX, several 8/7 articles were discussing the Nvidia Vera Rubin / Starmind exclusivity deal, which could easily have been written up as "today's positive." On checking, that announcement was published on 8/4 (Nvidia IR press release), when NVDA was about +3% and SPCX about +9% — already priced. Today's direct catalyst is the Argus upgrade on 8/7. The dates and the attribution distinction are written out explicitly in Section 3.2. This memory directly prevented a misattribution today.
• Data that could not be obtained (already disclosed to clients at the top of the report): NYSE advancing/declining issues (only Nasdaq obtained, 1,537:665), S&P 500 volume, single-name dollar volume and bid-ask spreads. The WSJ market diary page is rejected at the tool layer and is the main gap for NYSE breadth data; I suggest finding an alternative source (e.g. barchart, or a WSJ API path).
• Handling of source conflicts: ① one search summary said DOCS was "up over 200% pre-market," which badly contradicts the measured pre-market +88.72%, open +88.1% and high +93.6%, judged to be summary distortion and not used; ② two mutually exclusive versions of the SPCX price target appeared, Morgan Stanley $300 and $140, so this issue cites only the dual-source-confirmable Argus $160 and cites nothing else; ③ two versions of SPCX revenue appeared, $18.67B and $7.81B, and the dual-source-consistent quarterly basis of $7.81B (+92%) was adopted.
• Sub-agents: none dispatched this run. The session-level instruction explicitly requires not invoking the Agent tool without a user request, so all verification was done in the main process. The cost is that the supplementary verification on TWLO (the prior organic guidance of 9.5–10.5%) went only through search summaries plus call transcripts, with no line-by-line comparison against the SEC 8-K original, so its confidence is below the verification level applied to TEAM/NET in the pre-market piece. This is presented in Section 3.4 of the body as a "source," but the confidence difference is not flagged — a sentence should be added next time.

---

⚠️ Risk warning: this recap is post-close information review and observation only and does not constitute investment advice. Data may differ in timeliness or basis; please refer to company disclosures / SEC filings as authoritative, and do not use this directly as a basis for trading.