US Market Recap | 2026-07-31 (ET) Friday

Coverage window: 2026-07-31 09:30–16:00 ET regular session, plus 16:00–20:00 ET after hours. The reconciliation benchmark is the same-day pre-market list reports/us/2026-07-31.md.

Data sources and methodology statement (important):

• Local yfinance was once again rate-limited by Yahoo throughout (YFRateLimitError; direct calls to query1.finance.yahoo.com also returned HTTP 429, failing repeatedly in testing). This report uses no yfinance data at all — the same situation as the pre-market piece. This is now the second consecutive trading day, and should be treated as a stability problem with the local environment rather than a one-off (see the to-do list at the end).
• Single-stock and ETF closing prices come from stockanalysis.com (scraped 2026-08-01, labeled as the "Jul 31, 2026 4:00 PM EDT" close); indices, VIX, the dollar index, the 10Y Treasury and Brent come from investing.com historical data tables; macro data comes from BLS primary releases.
• Every single-stock percentage change has been cross-checked one by one by back-solving the 7-30 close as "7-31 close ÷ (1 + same-day % change)" and comparing it with the 7-30 closes recorded in the pre-market piece; all 18 names line up (see the note below the table in §2). This makes the reconciliation in this report reproducible, and it also validates in reverse that the back-solving method used in the pre-market piece is reliable.

---

0. Recap in one line

1. Today was risk-on, but it was an "extremely narrow risk-on held down by rates." The S&P was +0.70% and the Nasdaq Composite +1.0%, but the Russell 2000 closed down −0.45%, XLK tech closed down −0.22%, utilities −0.69% and healthcare −0.59%. The gains were highly concentrated in two stocks and one underrated theme: AMZN +15.32% (its best single day since 2012), GOOGL +6.73%, ETN +7.32%. The indices rose while the median stock fell.

2. The pre-market list's hit rate was 12/18 ≈ 67%, but that number is inflated and must be discounted. Three of them — KPTI / BIOA / MYGN — were small-cap biotechs already down −69%/−58%/−38% pre-market, so those were "mechanical hits"; stripping them out gives a true hit rate of 9/15 = 60%. All five misses erred in the same direction — reading "catalyst already delivered / poor entry level" as "will underperform today," and every one of them outperformed instead: MSFT +3.02%, META +3.28%, VRT +6.18%, WDC +2.21%, AVGO +0.37% (AVGO was the mirror-image error: liked, but it underperformed).

3. The pre-market piece's biggest judgment error was ranking "memory price hikes" as the day's #1 theme — today it was the weakest line in the entire market. MU −5.90%, SNDK −5.09%, while the HDD group carrying the same "storage" label rose: WDC +2.21%, STX +0.52%. This is precisely the verification condition the pre-market piece wrote down for itself in §8: "if the HDD group and the DRAM/NAND group diverge after the open, that is exactly what shows the earlier co-movement was label-driven rather than the same fundamentals" — the divergence appeared on cue, and the 1a/1b split was confirmed by the market. So the accurate statement is: the theme ranking was wrong, but the theme split was right, and the split is the harder, more valuable half.

4. The pre-market piece's best judgment was its final paragraph. It wrote that "today's opportunity is more likely in names not yet repriced pre-market (AVGO, NVDA, GOOGL) and in ETN, the only name today with a real catalyst that the market is ignoring, rather than at the top of the gainers list" — ETN +7.32%, GOOGL +6.73% and NVDA +2.93% all delivered; only AVGO +0.37% fell short. At the same time, "avoid storage names up more than 20% over two days" was completely correct (MU −5.90%, SNDK −5.09%). The conclusion beat the scoring table: the scores in §3 ranked GOOGL 7th (65 pts), and it was the second-strongest large cap of the day.

5. The ECI was, as expected, the only macro variable of the day, and it was hawkish: Q2 employment cost index +0.9% QoQ vs. consensus +0.8% (the pre-market piece could not obtain that consensus; it is filled in here), +3.4% YoY. The 10Y Treasury closed at 4.738% (+7.5bp), the highest since January 2025; the 30Y around 5.25% (+4bp, in the 19-year high zone). But stocks did not play catch-down; they rose instead — the "rates vs. equities divergence" the pre-market piece flagged as its biggest blind spot did not converge today, it widened further. This is the single most important unresolved risk carried into next week.

6. Next session (Monday 8-3) tone: a cautious continuation. VIX closed at 15.99 (−6.44%) and is already back at low levels, the dollar index at 99.94 (+0.08%) was essentially flat, and Brent +3.22% strengthened again. Taken together the three read as "equity sentiment is loosening, the inflation side is not." Next week's calendar is front-loaded: PLTR after the close Monday, AMD Tuesday, SNDK/WDC Wednesday, July payrolls Friday — the real verdict on storage comes Wednesday, not today.

---

1. Market overview

1.1 Indices and turnover

⚠️ Three measurement caveats ① Intraday and closing prints must not be mixed: the widely circulated "S&P 7,505.66 / +0.91%" is an intraday reading at 15:49 ET, not the close. The close was 7,489.80 / +0.70% — roughly a 0.2pp fade into the bell, which is itself an information point about the day's sentiment (see §5). This report uses closing values throughout. ② The Russell 2000 initially showed an erroneous "+1.37%" reading, which has been disproven. The investing.com historical table governs, at −0.45%, corroborated by the IWM ETF close of 291.20 / −0.48% (the two should move in the same direction and magnitude, and they do). This report does not use +1.37%. ③ The Nasdaq 100's July performance: one source claims "about −7%, the largest monthly drop since March 2025," which is too far from the Nasdaq Composite's −2.38% and cannot be reconciled with the verified closing series in this report. This report does not cite that number and flags it as an unresolved conflict.

1.2 Market breadth

⚠️ Advance/decline counts for 7-31 could not be obtained for this report; marked "no reliable data."

One data trap must be spelled out: the set of figures that turned up in searching — "NYSE 348 advancers / 271 decliners, advance-decline ratio 2.2:1, volume 18 billion shares" — is on verification the breadth for 7-30 (Thursday), not today. The evidence: the same article contains "VIX down 17.28% to 17.09" and "MSFT surged 15.5%," both of which have been confirmed as belonging to 7-30. This report does not use that set of figures. (This "headline date = publication day, content = prior trading day" writing convention is the single easiest source category to misuse when doing a recap; it has been added to the to-do list.)

Usable substitute evidence on breadth (indirect but directionally clear):

• The equal-weight view is missing, but the large-cap/small-cap gap is clear: the S&P was +0.70% while the Russell 2000 was −0.45%, a gap of 1.15pp.
• At least 5 of the 11 sectors closed lower (XLK / XLF / XLV / XLU plus small caps), with gains concentrated in XLY / XLC / XLE / XLI.
• XLK (tech) at −0.22% moved opposite to the Nasdaq Composite's +1.00% — that combination by itself shows the rally was not a "tech sector" move, but was driven individually by a handful of heavyweights that are not in XLK (AMZN sits in consumer discretionary, GOOGL/META in communication services).

Conclusion: today was a textbook "strong index, weak breadth" session, not a broad rally.

1.3 Volatility, rates and the dollar

⚠️ Brent's absolute price level carries an unresolved source conflict: the pre-market piece cited Fortune/Rigzone at 92.27 (7-31 06:45 ET, prior close 92.65), while the investing.com historical table shows 7-30 close 86.88 → 7-31 close 89.68, a level difference of roughly $2.6–3.0. The most likely explanation is a different contract month (front month vs. continuous contract; end-July is exactly a roll period), but this report could not confirm it. This report therefore uses only the "+3.22% same-day change" and draws no conclusion about Brent's absolute price level.

Sentiment call: risk-on on the surface, risk-mixed underneath.

• Supporting risk-on: VIX down sharply two days running to 15.99, Brent and XLE strengthening together, consumer discretionary and communication services leading.
• Against risk-on: the Russell 2000 closed lower, utilities −0.69% and healthcare −0.59% — both high-dividend/defensive sectors — fell together (this is not the healthy rotation of "money moving from defense to offense"; against a backdrop of the 10Y +7.5bp on the day it looks more like duration assets being directly suppressed by rates), and the fade from +0.91% into a +0.70% close.
• The most honest characterization: today was not a broad recovery in risk appetite; it was money concentrating chips heavily into the few stocks with hard catalysts, under the constraint of rising rates. VIX at 15.99 reflects "the index isn't panicking," which is not the same as "individual stocks aren't panicking" — MU −5.90% and AAPL −7.35% happened on the very same day the VIX fell.

---

2. Reconciliation against the pre-market list

Reconciliation basis (stated first, to avoid hindsight bias): the pre-market piece gives observation-intensity labels (priority deep-dive / watch closely / watch only / avoid), not buy-sell instructions. This table therefore reconciles against the relative strength direction implied by each label, benchmarked to the S&P 500's +0.70%:

• priority deep-dive / watch closely → implies "should outperform, or at least not lag, the index" → outperformance scores ✅
• watch only → implies "not worth chasing today, likely to underperform" → underperformance scores ✅, outperformance scores ❌
• avoid → implies "clearly weak" → a decline scores ✅

Note below the table (data self-consistency check): the "7-30 close" column for all 18 names above was back-solved as "7-31 close ÷ (1 + same-day % change)," and compared one by one against the 7-30 closes independently recorded in the pre-market piece, 18/18 match (all discrepancies within $0.2, arising from decimal rounding in the percentage changes). This both guarantees the reliability of the percentage changes here and validates in reverse that the back-solving method the pre-market piece used while yfinance was unavailable holds up.

Hit rate

A one-line self-critique

The real problem is not the number 60%; it is that all five errors share the same root.

MSFT, META, VRT and WDC were all labeled "watch only," on the grounds of, respectively, "catalyst already delivered," "weak bounce," "own fundamentals have not improved," and "up too much over two days" — the shared structure of all four is: using level / valuation / catalyst timing to negate direction. The market's answer today was: on a day when the theme is right, stocks with poor entry levels outperform anyway. And AVGO is the mirror image of the same error — the reason the pre-market piece liked it was precisely "good level, no gap up," and it turned out a good level didn't help, because it had no catalyst that day.

Read the five errors together and the conclusion is consistent: what determined moves today was "is there a same-day catalyst," not "is the level good." AMZN (earnings), GOOGL (data center news), ETN (earnings), VRT (theme spillover) and META (oversold bounce) all had same-day events; AVGO and STX did not, and went flat. In the pre-market piece's scoring model, "catalyst directness" is worth 20 pts and "expectation gap" 10 pts, but in practice the explanatory power of "catalyst directness" today far exceeded its weight — that is what the next version of the scoring model should adjust, rather than continuing to patch up low-discrimination filler items like "trading characteristics."

One more element of luck that must be acknowledged: GOOGL's +6.73% was driven mainly by the Anthropic data center news that only broke after the pre-market piece was filed. The pre-market judgment of "best expectation gap" left room for it, but counting it as a full stock-picking hit would not be honest — the correct statement is "the directional positioning was right, the source of the catalyst was not foreseen."

---

3. Verifying today's themes

3.1 Did the pre-market piece pick the strongest theme? — No, and the way it got it wrong is worth recording

§9② of the pre-market piece ranked "memory price hikes (DRAM/NAND) + AI nearline storage (HDD)" as the #1 theme of the day. The actual outcome:

• DRAM/NAND group: MU −5.90%, SNDK −5.09% — the weakest group of large-cap names in the whole market
• HDD group: WDC +2.21%, STX +0.52% — modest positive returns

Why did it get this wrong? The pre-market piece had actually already written down the answer; it just didn't apply it to the ranking. The original text stated explicitly that this chain "is missing an independent price anchor: third-party DRAM/NAND contract prices could not be obtained for this report," and it actively downgraded the relationship among the three pieces of testimony (Samsung's warning, Cook's "once-in-a-century flood," Jassy attributing the capex overrun to memory) from "cross-verification" to "aligned incentives, not independent verification."

— That downgrade was right, but after the downgrade it was still ranked first. That is the real methodological flaw: the logical hardness had already been marked down to "medium-high," yet the theme ranking was not adjusted along with it. A line of logic that you yourself concede is "the softest piece of foundation" should not simultaneously be the day's #1 theme.

At the same time, credit where it is due: in §2 it held firm against the surface appearance of "four names with similar pre-market gains" and insisted on splitting HDD and DRAM/NAND into two independent sub-chains, 1a and 1b, and in §8 it wrote down an explicit falsification condition — "if the HDD group and the DRAM/NAND group diverge after the open, that is exactly what shows the earlier co-movement was label-driven rather than the same fundamentals." Today the two groups printed +2.21% vs. −5.90%, an 8.1pp gap, and that split was fully confirmed.

Net assessment: the ranking was wrong (−), the split was right (+), and the split is the harder half. A recap that only looks at moves would say "the memory theme was a miss"; an accurate recap should say: the pre-market piece's "structural understanding" of this theme was right, and its "weighting understanding" was wrong.

3.2 Was there a surprise theme the pre-market piece missed? — Yes, and it was the second strongest of the day

GOOGL's +6.73% was driven by news that only appeared after the pre-market piece was filed: Google backing Anthropic's $15 billion data center in Hubbard, Texas (the campus includes a 1.6GW gas-fired power plant); Google may reportedly take roughly 20% equity in the project in exchange for guaranteeing part of the lease.

The significance of this news goes beyond a single stock; it hits three lines at once:

1. The return-on-investment reckoning for AI cloud — Google's "lease guarantees for equity" structure is a form of spending with more return discipline than META's "pure lease obligations";
2. Data center power — a self-built 1.6GW gas plant points to the same bottleneck as the CVX–Microsoft 20-year power supply agreement and ETN's electrical equipment orders;
3. Anthropic as a hidden thread running across multiple companies' income statements — AMZN (a $53.4 billion investment revaluation), MSFT (about $3.2 billion) and GOOGL (a reported stake of about $124 billion) already share the same valuation source.

It is forgivable that the pre-market piece missed it (the news broke after filing), but the pre-market piece could have noticed point 3 earlier: in §5⑦ it itself wrote that "GOOGL's TTM net profit of $244.12 billion shows an anomalous YoY increase; strongly suspect it includes AI investment revaluation," yet it treated that only as a "valuation may be distorted" risk and did not realize it simultaneously means GOOGL is one of the core assets of the Anthropic theme. From the same fact it read only the risk side and not the elasticity side.

3.3 An honest line about the "quality audit" theme

The pre-market piece ranked "earnings quality audit" as the 3rd theme, arguing that "the market accurately saw through one-off gains." Today's data does not support that argument, but it does not refute it either.

AMZN's headline EPS of 5.75 includes $53.4 billion of one-off gains; stripping them out, the true beat is only +7%~+13% — and the stock rose 15.32% today. Looking at today alone, the market behaved more like it did not see through it, or saw through it and did not care.

But one day's price cannot settle whether this theme is true: the explanation the pre-market piece itself offered (the market is buying the +$132 billion sequential increase in AWS backlog, not the quarter's EPS) still holds and is more persuasive — the AWS acceleration is real, the one-off gain is not, and the market is paying for the part that is real, which does not contradict "seeing through." The correct treatment: the verification cycle for this theme is quarterly, not daily; today it is recorded as neither a hit nor a miss, and marked "verification point not yet reached."

---

4. Post-close earnings moves

⚠️ Conclusion first: 7-31 was a Friday, and the regular after-hours session (16:00–20:00 ET) produced no earnings event worth treating as a next-day catalyst. This section states that plainly and does not invent entries to fill out the structure.

4.1 Today's after-hours session (after 16:00 ET)

No significant company was found to have reported after the close on 7-31. US earnings convention rarely schedules releases for Friday after the close (news value gets diluted by the weekend); this week's dense window runs Tuesday through Thursday next week.

After-hours price drift (low signal, recorded only):

How to read this (restraint required): after-hours volume is extremely thin, so the drifts above have no statistical significance. The only mildly directional observation is that the two energy majors strengthened together after hours while the three storage names weakened together — both consistent with the intraday sector direction, i.e. continuation rather than reversal. This should not be used to predict Monday's open.

4.2 Earnings released pre-market today but priced during the session (filling a coverage gap in the pre-market piece)

Line 9d of §1 of the pre-market piece admitted that "ABBV/CL/CCJ/LEA were not individually verified." The most important of those is filled in here:

This case follows the same pattern as AAPL: beat on the quarter, guidance cut → down. The market's pricing rule today of "guidance > current quarter" was applied consistently across both mega-cap tech and pharma.

4.3 Next week's real catalysts (inputs to the next-day outlook)

⚠️ To be verified: the exact release times for ISM manufacturing (typically 8-3) and ISM services (typically 8-5) could not be officially confirmed for this report; they are listed by convention only and should not be used as a confirmed calendar.

---

5. Flows and sentiment

5.1 Sector rotation (sorted by today's move)

The core fact about the rotation — it must be spelled out: today the "tech sector" (XLK) fell while "tech stocks" rose. There is no contradiction: AMZN sits in consumer discretionary and GOOGL/META in communication services, none of them in XLK; and XLK's largest weight, AAPL, crashed 7.35%.

This yields an inference that is useful for the next session: using XLK or QQQ to judge "how strong the AI theme was today" gives the wrong answer. Today's real vehicles for the AI theme were the highly unconventional combination of XLY (AMZN) + XLC (GOOGL) + XLI (ETN / power equipment).

5.2 VIX and Treasuries

• VIX 15.99 (−6.44%), sharply lower two days running (7-30 −17.28%, 7-31 −6.44%), a fast retreat from post-FOMC panic levels on 7-29 back to lows.
• 10Y at 4.738% (+7.5bp, highest since January 2025), 30Y around 5.25% (+4bp).
• The driver is clear: ECI Q2 +0.9% QoQ vs. +0.8% expected (+3.4% YoY, with private-sector real wages −0.4% YoY, the first decline since 2022). Wage inflation not cooling → reinforces the hawkish pricing behind the FOMC's "three votes for a hike" → the long end moves up.

⚠️ A divergence that must be called out (the pre-market piece flagged it as "today's biggest blind spot"; today it did not converge but widened):

The pre-market piece's exact words were "Nasdaq futures +1.11% coexisting with the 30Y Treasury at 5.23% (a 19-year high)... historically this kind of divergence rarely persists for long — either rates fall back or stocks play catch-down."

Today's outcome: neither happened, and the divergence deepened instead. Rates rose rather than fell (10Y +7.5bp), stocks rose rather than fell (Nasdaq Composite +1.0%), and the VIX fell 6.44% on top of it.

An important methodological correction is required here, rather than simply saying "the blind spot is still there": treating "the stock-bond divergence" as a signal that must converge itself carries too strong a prior. What the market actually offered today is a self-consistent explanation — the rise in rates is driven by "sticky wages / nominal growth that isn't weak," not by "recession / credit risk"; under that cause, long-duration growth assets suffer (XLU −0.69% and IWM −0.48% really were suppressed), but the giants with genuine earnings acceleration (AWS +37%, Azure +45%) can outperform through the numerator even as the denominator worsens. The divergence did not converge because it was never the same set of assets diverging — what got suppressed (small caps, utilities) and what rose (AMZN, GOOGL) simply are not the same duration bucket.

So the more accurate formulation is: not "stocks and rates have diverged and the bill will come due," but "rates are creating stratification inside the market — they have already delivered their suppression on small caps and defensive sectors, they just have not yet reached the giants backed by earnings acceleration." This stratification will most likely persist until the 8-7 payrolls; the real risk is payrolls also running hot and pushing rates above some threshold, at which point even the giants' denominator gives way.

5.3 risk-on / risk-off characterization

Characterization: structurally risk-on, breadth-wise risk-off.

Overall: money is not adding to total risk exposure; it is reallocating it — out of small caps, defensives and DRAM/NAND, into the handful of large caps with hard same-day catalysts (AMZN/GOOGL/ETN). VIX at 15.99 describes calm at the index level, and it masks violent divergence at the single-stock level (AMZN +15.32% coexisting with MU −5.90% and AAPL −7.35% on the same day).

---

6. Next-day outlook (next trading day: Monday 8-3)

① Theme persistence

② Next-day and this week's calendar

8-3 (Monday)

• After the close: PLTR Q2 earnings (call around 17:00 ET) — a sentiment gauge for high-multiple AI software
• ISM manufacturing PMI (conventionally Monday 10:00 ET; not officially confirmed for this report)

Later this week (already affecting Monday's pricing)

• 8-4 Tue: AMD Q2 earnings — PE 161.86, already −1.90% today on de-risking
• 8-5 Wed: SNDK + WDC earnings — verdict day for the storage theme
• 8-7 Fri 08:30 ET: July non-farm payrolls report — the biggest macro event of the week

Why payrolls matter more than any earnings report: the FOMC has already produced "three votes for a hike," today's ECI came in at +0.9% above consensus, and the 10Y is already at an 18-month high of 4.738%. If payrolls also run hot, rate-hike pricing gets reinforced further, and the "rate stratification" described in §5.2 could spread from small caps to large-cap growth. Conversely, if payrolls soften, the IWM/XLU names suppressed today would be the side with the most elasticity.

③ Key names to watch (ticker + verification point)

④ What to avoid

• Avoid chasing SNDK / WDC ahead of the 8-5 earnings: SNDK was −5.09% today, but the volatility profile of a 52-week range of 40.10–2,354.39 is unchanged, and two-way risk into earnings is extreme.
• Avoid chasing AMD ahead of the 8-4 earnings: PE 161.86, already −1.90% today; the pre-market piece's discipline of "don't touch high-multiple names in the 48 hours before earnings" was validated today.
• Avoid chasing AMZN's +15.32%: this must be kept separate from the "don't chase" error made pre-market — not chasing at +11.58% pre-market was conservative; chasing after a +15.32% close is a level problem. The two red flags (negative FCF, and midpoint Q3 operating profit guidance of $24.5 billion < the $27.5 billion actually delivered in Q2) have still not been reflected in any single day's price.
• Avoid buying the AAPL dip: there was almost no dip-buying all day (−7.65% pre-market → −7.35% at the close), and the view that memory cost pressure extends into FY27H1 has not been falsified.
• Avoid small caps and high dividend yield (IWM/XLU/REITs) until the 8-7 payrolls: the 10Y is at 4.738% with direction undecided.
• Do not use XLK/QQQ to judge the strength of the AI theme (see §5.1) — today that would have given a completely wrong answer.

⑤ Input notes for the next pre-market list

1. Third-party DRAM/NAND contract prices (TrendForce and similar) must be obtained first. This is the item the pre-market piece listed as "highest priority to verify," and it has now gone unobtained for two consecutive days. Until it is in hand, the memory theme should not be ranked #1 again — today we paid the price for that.
2. GOOGL's income statement quality must be verified. It is currently both "the best stock pick" and "valuation authenticity unknown," and that combination cannot be carried into a third day. Check the contribution of investment revaluation line items in the 10-Q to the TTM net profit of $244.12 billion.
3. The scoring model needs to raise the weight on "catalyst directness." All 5 errors today stemmed from "using level to negate direction" (§2 self-critique), and whether there was a same-day catalyst had far more explanatory power than the level.
4. Establish a consistency check between "theme strength" and "theme ranking": the pre-market piece had already marked memory's logical hardness down to "medium-high" yet still ranked it first — once hardness is marked down, the ranking must be marked down with it, and this should become a hard rule.
5. Data source discipline: ① yfinance has been rate-limited two days running and the pre-market process needs to default to a backup source; ② beware reports where "the headline date is the publication day and the content is the prior trading day" (§1.2 of this report discarded a set of breadth data for exactly this reason); ③ intraday prints must not be treated as closing values (§1.1 of this report already removed the 15:49 print of 7,505.66).
6. Fill in the items not obtained today: 7-31 advance/decline counts, the Brent contract basis, and the Nasdaq 100's July performance.

---

Appendix: this report's to-be-verified list

Still unresolved (carried over from the pre-market piece, in descending priority)

1. Third-party DRAM/NAND contract / spot prices — the only independent verification source for the memory theme, unobtained for two consecutive days.
2. Whether GOOGL's P/E of 16.74 is contaminated by AI investment revaluation — today it was the second-strongest name, and the authenticity of its valuation is still unverified.
3. The tax treatment of AMZN's $53.4 billion Anthropic gain — it determines whether the ex-item EPS lands at 1.95 or 2.06.
4. MU's TTM net profit — this report obtained a market cap of 929.52B, TTM revenue of 90.27B and a PE of 19.74 (implying TTM net profit of roughly $47.1 billion), but that back-solved figure has not been verified against primary filings and is not used as a conclusion.

Newly added in this report 5. NYSE/Nasdaq advance/decline counts for 7-31 — not obtained; the set of figures found in searching was verified as belonging to 7-30 and discarded. 6. The source conflict on Brent's absolute price level — 92.27 (Fortune/Rigzone) vs. 89.68 (investing.com), likely a contract-month difference; this report uses only the +3.22% change. 7. The Nasdaq 100's July performance — one source claims about −7%, irreconcilable with the Nasdaq Composite's −2.38%; discarded. 8. The exact release times for ISM manufacturing/services — listed by convention, not officially confirmed. 9. The precise closing value for the 30Y Treasury at 5.25% — taken from media paraphrase ("about +4bp"), with no original historical-table value obtained; the 10Y's 4.738% is a primary data-table reading.

Known limitations of this report

• The definition of the hit rate leaves room for post-hoc interpretation: the pre-market piece gives observation-intensity labels rather than trading instructions, and this report judges delivery by "strength relative to the S&P 500," which is a reasonable but not the only possible mapping. If "watch only" is read as "does not constitute a buy recommendation" rather than "expected to underperform," then the four "watch only" names that rose today (MSFT/META/WDC/VRT) should be scored neutral rather than misses, and the result becomes 12 hits / 1 miss (AVGO) / 5 neutral — the miss count drops from 5 to 1. This report adopted the basis that is harsher on itself and declares that choice here; under either basis "12 hits" is unchanged, and the difference is only in how errors are counted.
• STX was recorded as "roughly flat" rather than a miss, a borderline judgment (+0.52% vs. the index's +0.70%, a difference of 0.18pp).
• After-hours drift data is on extremely thin volume and has no statistical significance, as already noted in §4.1.

---

⚠️ Risk disclosure: this recap is only a post-close review of information and observations, and does not constitute investment advice. Data may differ in timeliness or measurement basis; company disclosures / SEC filings govern, and this must not be used directly as a trading basis. Labels such as "priority deep-dive / watch closely / watch only / avoid" are this report system's observation-intensity tags, not trading instructions; automatically generated content may contain stale information or factual errors.