US Pre-Market Brief | 2026-07-29 (ET) Wednesday

Coverage window: 2026-07-28 16:00 ET regular-session close → 2026-07-29 08:00 ET (includes 7/28 after-hours earnings, the Asia/Europe overnight session, and this morning's pre-market) Data pulled: 2026-07-29 at approximately 12:00 UTC (= 08:00 ET, 90 minutes before the open) Reconciliation baseline: input prompts come from the 7 retraining feedback items in reports/us/2026-07-28-recap.md §6⑤, each of which has been acted on (see "Retraining Execution Checklist" at the end). This edition is the revised version after QC and fundamentals re-verification: ① the first draft was QC'd by risk-auditor, which found 14 red flags, of which 5 🔴-level items were externally re-verified and corrected one by one; ② fundamentals-analyst completed an independent yfinance re-verification of BE / QCOM / ARM, correcting 5 further substantive points (including a media-sourced error in BE's pre-market gain). The full correction list is in "QC Revision Log" (R1–R15) at the end — that section records which conclusions this report changed; please read it first. ⚠️ The yfinance re-verification for MSFT / META did not return, so the numbers for those two (tonight's main event) are not independently cross-checked; see §10 open item 10. ⚠️ Data-source blacklist (yesterday's §6⑤ item 7): in the Zacks "Stock Market News for July X" series the body date runs one day later than the headline; this report does not use it. This edition adds ts2.tech to the same treatment (it self-rated low confidence yet the first draft accepted it one-sidedly; see revision log R14).

---

0. Today in One Sentence

1. Today is the highest event-density day of the quarter, with three events interlocking: the 14:00 FOMC decision + the 14:30 Warsh press conference + after-hours earnings from five companies, MSFT/META/QCOM/ARM/LRCX. Positioning before the open is merely placeholding for what comes after 14:00; pre-market price discovery is incomplete.

2. Geopolitics escalated overnight, but the nature of it is "two-way," not one-sidedly bullish for oil. Iran-linked militias launched drones from inside Iraq at oil facilities in Saudi Arabia's Eastern Province (Saudi Arabia says all were intercepted); US and Saudi jets then struck multiple logistics and weapons sites in eastern Iraq; an IRGC ballistic-missile "surprise attack" on US forces was intercepted by US forces. Before this, the US and Iran had gone roughly three days without attacking each other. Brent September contract about +3.5% to near $87.0 (CNBC¹). ⚠️ Two opposing facts that must be stated side by side: ① all attacks were intercepted, with no production loss; ② Iran's foreign minister held separate talks with Saudi Arabia and Oman over the same period. The Strait of Hormuz, per Iran's foreign ministry, "remains closed" — this is a pre-existing carry-over fact, not something new overnight.

3. Therefore: the driver of yesterday's "beneficiaries of falling oil" theme has reversed, but it should be downgraded to "watch and see" rather than zeroed out. Yesterday's recap raised that chain's persistence to "medium" on the grounds that "the driver is still there today." The driver did indeed reverse overnight, but: the pass-through of fuel costs to airlines/cruise lines is hedged and lags by quarters, so a single-day +3.5% does not constitute a same-day fundamental reversal; and RCL's raised full-year EPS guidance is a company fact already on the ground, which should not be overridden by one trading day of commodity prices. The first draft mechanically cut this theme to 0 and judged RCL a casualty — that is taking the rule too far, and this edition has corrected it.

4. The strongest catalyst is BE (Bloom Energy), but it is also the most contested name today. Q2 revenue $1.0654 billion (beat of about 30%), +165.5% YoY, adjusted EPS $0.78 vs. est. $0.4066 (a beat of 91.8%), full-year revenue guidance $3.4–3.8 billion → $3.9–4.2 billion, adjusted EPS $1.85–2.25 → $2.55–2.85. ✅ All four of the above have been independently re-verified against yfinance and match. ⚠️ But the pre-market gain needs correcting: measured at +8.45% to $180.94 (12:29 UTC), not the media's "+11% to $185.20"; pre-market high $182.19 (+9.14%, 12:11 UTC). And that gain only recovers part of the −11.34% drop on 7/28 — the net price level is still below the 7/27 close of $188.18. ⚠️ Three more important pieces of background (entirely missing from the first draft, supplied by yfinance): ① BE is already −52.5% from its 52-week high ($351.28, 6/25) and −33.8% over the past month — this "beat and raise" comes after the stock has been cut in half, not at a high; ② diluted share count rose from 232.5 million to 319.7 million over one year, +37.5%, while total debt over the same period rose from $1.52 billion to $2.95 billion, leaving a net position of net debt of $458 million (not net cash); ③ what the short seller (Hunterbrook, 7/8) questions is the backlog itself (scandium oxide supply), which the company has publicly rebutted.

5. Pre-market state: disagreement, not direction. S&P futures +0.2%, Nasdaq 100 futures roughly flat (SPY $742.65 +0.24%, QQQ $675.59 +0.02%); the Nasdaq 100 formally entered a technical correction on 7/28 (more than −10% from its high) (NBC²). ⚠️ VIX 18.21, 10Y 4.606%, DXY 101.393 are all 7/28 closing values; real-time pre-market values were not obtained. Asia fell hard overnight (KOSPI about −6%, SK hynix closed −9.6% (down as much as −20% intraday), Samsung −5.2%, Nikkei −1.5%), while Europe was firmer (FTSE +0.6%). Driver types ranked: macro > earnings/guidance > geopolitics > industry (China DUV) > ratings.

---

1. News Overview

Impact level uses a single six-tier scale, S / A+ / A / B+ / B / C (the first draft used four tiers in §1 and six in §3, running two systems in parallel; they are now merged).

Note A — FOMC pricing: three markets disagree sharply; do not take a range and do not mix them. ⚠️ The first draft simply carried over yesterday's recap figures of "hold 63.5–65% / hike 35–36.5% / September 82%," which are 7/28 stale values; this edition retracts them. The readings currently obtainable (none is an 08:00 ET primary snapshot; timestamps noted below):

• CME FedWatch: 35.4% hike five days before the meeting; fell to about 30% as the meeting approached; the July peak reached 46.5%
• Polymarket: 25.1% five days before the meeting; 6.6% hike / 93.2% hold as the meeting approached
• Kalshi: about 38%
• September: Polymarket about 70%; another source reports 77% Two readings: ① the three markets range from 6.6% to 38%, so no single number can serve as the basis for a conclusion; ② more importantly, the drift direction — hike pricing was moving down going into the meeting (46.5% → about 30%), the opposite of the intuition that "the oil spike reinforces the hawks." On that basis this report lowers the weight it puts on hawkish risk (see §2 Theme 1). Also: the contested question at this meeting is not only energy inflation. Employment is an equally major variable (media accounts describe the labor market at multi-year lows); the first draft reduced the FOMC to a single "energy inflation" story, which was missing a leg — now added.

---

2. Strongest Themes, Descending

---

3. Overall Single-Stock Strength Ranking

⚠️ Yesterday's §6⑤ item 2: every single stock must first clear the sector-direction gate. §3.1 keeps the "Sector ETF (7/28)" column, and conflicts have been proactively downgraded. ⚠️ Pre-market column convention: only genuine pre-market quotes go in; where none is available, write "no pre-market quote." The first draft put Ford's after-hours price into the pre-market column and STX's full-day 7/28 gain into the pre-market column, both cross-session mixing, now corrected. Positive levels use one six-tier scale: S / A+ / A / B+ / B / C (the first draft's A− and C+ were out of bounds and have been fixed).

3.1 Bullish Side (descending by total score)

3.2 Bearish / Avoid Side (ascending by total score; the lower the score, the more negative)

---

4. Single-Stock Scoring Model (100 points)

4.1 Line-Item Detail (sums re-checked item by item)

⬇/⬆ are adjustments relative to the first draft, with reasons:

• BE moat 12→8, risk −10→−14: ① the core supporting metric, backlog, is exactly what the short seller disputes; ② ✅ new from the yfinance re-verification: net debt $458 million (the first draft implicitly assumed net cash, which is wrong), D/E 311.5%, diluted share count +37.5% in one year, one quarter contributing 140% of TTM operating cash flow (extremely unstable), and on FY26 guidance P/E 62–67x / P/S ~12x / P/B 51.5x / EV/EBITDA 207x. Total 76→71→68.
• TER elasticity 15→14, expectation gap 5→6, risk −12→−15: ⚠️ the first draft used "+104% YoY" as the expectation gap, violating this model's convention — Q2 revenue actually beat consensus by only 8.9%; added the "Q3 guidance down QoQ on both lines" risk. Total 72→69.
• F elasticity 12→11, expectation gap 7→5, risk −6→−8: revenue −4%, the FCF raise includes a $500 million one-off refund, and this is the second raise this year (the expectation gap should be discounted). Total 71→66.
• CSGP moves from the bearish side to the bullish side: profit guidance maintained + net income +817%; the first draft checked only the revenue guidance line before assigning "negative A / avoid," violating this report's own iron rule stated in §5 that "guidance outweighs the current-quarter beat." The risk deduction is taken at the model cap of −15. Total 30→55.

⚠️ One self-exposure in the scoring model: in the first draft's §4.1 all six line-item sums were correct, yet for two of them (TER, F) the "expectation gap" input itself was wrong (YoY substituted for vs.-consensus). A correct sum ≠ a correct score — a prohibition clause has been added to the model's convention column.

---

5. Detailed Analysis of Top Names

1Bloom EnergyBE68 pts · watch closely

• Related news (7/28 16:05, Businesswire⁶): Q2 revenue $1.0654 billion (est. $822.8 million, a beat of about 30%), +165.5% YoY; adjusted EPS $0.78 (est. $0.40); product revenue +215% to $935.4 million; gross margin 33.4% (+668bp); operating profit $182.2 million (versus −$3.5 million a year earlier). Full-year revenue $3.4–3.8 billion → $3.9–4.2 billion; adjusted EPS $1.85–2.25 → $2.55–2.85.
• Catalyst logic: a double of current-quarter beat + guidance raise, where the size of the guidance raise (EPS midpoint +38%) exceeds the size of the current-quarter beat — satisfying this skill's iron rule that "guidance outweighs the current-quarter beat."
• ⚠️ The dispute that must be stated alongside it (missing from the first draft, added here): what the short seller Hunterbrook Media (7/8) questions is the backlog itself — the contested point is whether global supply of scandium oxide can support the backlog scale the company claims. The company has publicly rebutted this, saying supply is ample, the supply chain does not depend on China, and visibility supports 25GW of annual production. (Globe and Mail²³) This report does not adjudicate that dispute, but since the backlog is the contested item, it cannot simultaneously be treated as clean moat evidence — the moat score has been cut from 12 to 9.
• ✅ The H2 arithmetic problem is solved (listed as an open item in the first draft, closed after the yfinance re-verification): H1 revenue booked = Q1 $0.7511 billion + Q2 $1.0654 billion = $1.8165 billion. To reach the $3.9–4.2 billion guidance, the second half needs to do $2.0835–2.3835 billion. Sell-side consensus before the print had Q3+Q4 = $0.9337 + $1.1630 = $2.0967 billion, sitting exactly at the low end of the guidance range. → Reading: the guidance floor is essentially equal to the prior consensus, so clearing it is not hard; but the guidance ceiling implies the second half running about 14% above prior consensus, so the range itself is aggressive. The contested question is not "can they hit it" but "how good is the ceiling." → ⚠️ Also clarifying one questionable figure from the first draft: the pre-print (7/27) media line of "H1 revenue $751.1 million" is arithmetically incompatible with Q2 alone at $1.0654 billion — $751.1 million is in fact Q1 standalone revenue (+130.4% YoY), mislabeled as H1. The first draft was right to discard the number, but wrong about why.
• ⚠️ Three substantive risks added by the yfinance re-verification (all missing from the first draft):
  1. Position: the 52-week high is $351.28 (6/25) and the current price is −52.5% from it; −33.8% over the past month, −42.1% over the past three months; but still +92.0% YTD and +379.7% over the past 12 months; beta 3.74. This "beat and raise" comes after the stock was cut in half, not at a high — the payoff structure is completely different from what the first draft implicitly assumed.
  2. Dilution and leverage: diluted share count went from 232.5 million → 319.7 million (+37.5%) in a year; over the same period cash went from $575 million → $2.491 billion and total debt from $1.522 billion → $2.949 billion — the company did both a large equity and a large debt raise within two quarters. The net position is net debt of $458 million, not net cash; D/E 311.5%. "Revenue doubled" has to be discounted at the per-share level.
  3. Cash-flow quality: TTM operating cash flow is only $298.2 million, of which 2025Q4 alone contributed $418.1 million (140% of the total), while 2026Q1 was only $73.6 million. This is not a smooth curve; Q2 operating cash flow is the number to watch (Yahoo has not yet recorded it).
• Valuation (on the company's own FY26 guidance, not Yahoo's forward convention): P/E 62–67x, P/S ~12x, P/B 51.5x, EV/EBITDA 207x, short interest 7.7% of float. ⚠️ The forward P/E of 36.7x that Yahoo shows uses FY2027 consensus (EPS 4.545), and that number implies FY27 revenue growing another 72.8% and EPS another 109.7% — an assumption of two consecutive doublings, with very little margin for error. Whenever you cite a forward P/E, you must state which year it is.
• One positive cross-check on earnings quality: Q2 GAAP net income $196.3 million + stock-based compensation of about $45 million ≈ $241 million, which lines up with "adjusted EPS 0.78 × diluted shares of about 320 million ≈ $250 million" → this beat is mainly operational, not something propped up by a one-off tax item. (Inference; must be confirmed against the non-GAAP reconciliation table in the 10-Q.)
• Pre-market and technicals: 7/27 close $188.18 → 7/28 intraday −11.34% to $166.84 (pre-earnings panic + the short report + a pipeline regulatory rejection) → after hours +10.88% $184.99 → measured pre-market +8.45% $180.94 (pre-market high $182.19, +9.14%).
• Final judgment: the fundamental numbers are the strongest fact on the ground today — revenue, EPS and guidance have all three been independently re-verified against yfinance and match. But this is a stock that is down by half from its high, has diluted 37.5% in a year, operates with net debt, has an extremely unstable cash-flow curve, and whose core metric is being publicly questioned. ⚠️ The verification point set yesterday has still not been met as of writing — the measured $180.94 is still below the 7/27 close of $188.18, and by a wider margin than the media figure ($185.20) suggests. Per yesterday's §6④ item 6: until it reclaims $188.18, do not treat it as a clean positive trade.

2TeradyneTER69 pts · watch closely

• Related news (7/28, Businesswire¹¹): Q2 EPS $2.47 (est. $2.09), revenue $1.329 billion; Q3 guidance EPS midpoint $2.00 vs. consensus $1.44, revenue midpoint $1.25 billion vs. consensus $1.026 billion (a beat of about 20%). CEO Greg Smith said the Q3 guidance reflects "strong AI-related demand."
• ⚠️ Two omissions in the first draft, added here:
  1. The correct expectation gap is +8.9%, not +104%. Q2 revenue consensus was $1.22 billion against an actual $1.329 billion, a beat of 8.9%. "+104% YoY" is a growth rate, not an expectation gap — the first draft substituting YoY for the expectation gap is the scoring-method error this QC pass caught.
  2. The Q3 guidance midpoint is below Q2 actuals: revenue $1.25 billion vs. $1.329 billion (−6% QoQ); EPS $2.00 vs. $2.47 (−19% QoQ). "Guidance beat by a wide margin" and "guidance down QoQ" are both true at once — the first draft wrote only the first half.
• Theme and stage: "back-end test ≠ front-end inspection" — ⚠️ this hypothesis has no new evidence today. TER and KLAC are the same data set from after the 7/28 close, and the real test (intraday price action) has not happened yet. And on re-verification the KLAC leg is gross margin mix + guidance beating only marginally, with no direct connection to "front-end capex exposure" — the axis actually rests on TER as a single point of support.
• Pre-market: 7/28 close $320.65 (−4.22%) → after hours +13.53% $364.02 → pre-market +7.0% $343.02, nearly half already given back.
• Final judgment: the one most worth watching and least worth chasing. Yesterday's §6④ item 3 already listed "chasing TER's after-hours gain" as avoidance item 3; the give-back from +13.5% after hours to +7.0% pre-market is early confirmation of that avoidance. Watch first, do not chase.

3MicrosoftMSFT67 pts · watch only

• FQ4 after the close tonight. ⚠️ Consensus conventions conflict; stated side by side without adjudication: one set has revenue $89.37 billion / adjusted EPS $4.33; yesterday's recap recorded another set at revenue $87.50–87.67 billion / EPS $4.22–4.24. The company's own guidance is $86.7–87.8 billion. Azure guidance 39–40% (constant currency).
• Catalyst logic: what is decisive is not this quarter's Azure but FY2027 capex guidance. Capex this quarter is expected to exceed $40 billion; calendar 2026 is around $190 billion; the market expects FY27 growth of +20%~30% (about $220 billion). Microsoft Cloud gross margin guidance has already come down to 64%, and coming in below that would mean AI growth is still being swallowed by costs. Copilot paid seats exceed 20 million.
• ⚠️ The capex guidance is two-way for the market: above expectations → positive for semis, negative for software margins; below expectations → confirms "capex has peaked." The outcome of today's two main lines rests on the same number.
• Options: the 7/31 weekly 392.50 straddle prices about ±7%, with a call/put ratio of about 1.9:1; the implied market-cap swing is about $189 billion. 7/28 close $393.35 (+1.09%).
• Judgment: watch only. Buying before the print means paying an IV premium for a two-way, unpriceable binary event, and there will certainly be IV crush afterwards.

4FordF66 pts · watch closely

• Related news (7/28, CNBC¹²): adjusted EPS $0.42 (est. $0.35); adjusted EBIT $2.5 billion (+17%), 5.2% margin; Q2 adjusted FCF $2.1 billion; cash $22.3 billion, total liquidity $43.4 billion. Full-year adjusted EBIT $8.5–10.5 billion → $10.0–11.0 billion; full-year adjusted FCF $5.0–6.0 billion → $6.0–7.0 billion.
• ⚠️ Three omissions in the first draft, added here: ① revenue $48.3 billion, −4% YoY — the media line was "margins expand despite revenue dip," with volumes down because of the Novelis disruption and some models being discontinued, and the first draft never mentioned Ford's revenue or volumes anywhere; ② the FCF raise includes the $500 million of the IEEPA tariff refund already received early (the Supreme Court struck down the IEEPA tariffs; the total is $1.3 billion, with the remaining $800 million falling in 2027) — that is, half of the roughly $1 billion increase in the FCF midpoint comes from a one-off refund; ③ this is the second full-year guidance raise of 2026, so the expectation gap should be discounted accordingly.
• Judgment: a $1.5 billion lift in the guidance floor is still a hard signal and the sector is a tailwind (XLY +1.48%), but the "clean positive" quality is lower than the first draft judged. The growth quality is "margin expansion + one-off items," not "volume and price rising together."

5CoStarCSGP⚠️ the single largest conclusion correction in this edition · 55 pts · watch only

• The first draft graded it "negative A / avoid" on the basis of "both Q3 and full-year guidance below expectations + the CFO leaving." On re-verification, that judgment checked only one guidance line.
• The complete facts (Businesswire¹⁵): the company's own headline is "a profitability inflection." Q2 revenue $925 million (+18%); net income $55 million vs. $6 million a year earlier (+817%), diluted EPS $0.14 vs. $0.01; adjusted EBITDA $184 million (+116%), 20% margin. Full-year adjusted EBITDA guidance of $780–820 million is unchanged (21–22% margin, which the company calls its highest ever). Full-year revenue guidance was cut to $3.715–3.755 billion, but that still represents +14~16% growth, which the company attributes to efficiency optimization at Homes.com and to the transaction-based Ten-X business as a deliberate strategic choice. The CFO change is an internal succession (Lown moving to Allstate, Rossmann taking over), not an abrupt departure.
• Judgment: this is a "cut revenue guidance, protect profit guidance" structure, not a one-sided negative. The −14.2% pre-market reflects a market looking only at the revenue line. But this report does not turn bullish on that basis — the revenue guidance cut is a fact, the pre-market decline is still widening (after hours −13.19% → pre-market −14.2%), and XLRE at only +0.55% offers no support. "Watch only" rather than "avoid."
• ⚠️ Methodological lesson: this report's own iron rules say "guidance outweighs the current-quarter beat," yet on CSGP it checked only the revenue guidance line. §10 open item 4 has added "whether FVRR is the same pattern" as a must-check item.

6VertivVRT65 pts · watch only

• Reported pre-market on 7/29. Revenue $3.274 billion (+24%, +18% organic) missed the $3.38 billion estimate; adjusted diluted EPS $1.52 (+60%); adjusted operating margin 22.6% (+410bp); full-year guidance raised across the board (second consecutive quarter); Q3 guidance revenue $3.65–3.85 billion, adjusted operating margin 24–25%.
• Judgment: excellent on profit, short on revenue — in the current environment a revenue miss in the AI chain carries more weight than an earnings beat, because the bear narrative is precisely "demand has peaked." It and BE are both AI power yet moving in opposite directions, and that divergence is one of today's most informative observations: the market is currently rewarding "visibility on new orders" and punishing "margin delivery on existing orders." Already −6.27% on 7/28, still +66% YTD.

7MetaMETA64 pts · watch only

• Q2 after the close tonight. Consensus EPS about $7.18–7.24, revenue about $60.16 billion (guidance $58.0–61.0 billion). Full-year capex guidance $125–145 billion against $72.2 billion actual in 2025 — nearly a doubling within a year, already raised twice this year.
• Judgment: the core question is whether advertising growth can cover the capex expansion, and how much negative free cash flow the market will tolerate. META's capex controversy is sharper than MSFT's — MSFT's capex has Azure revenue directly against it, while META's mainly corresponds to internal model training, a longer ROI path. Watch only.

⑧ Energy chain (XLE / MGY, 58 pts, watch closely)

• Brent's September contract rose about +3.5% to $87.0 overnight; the drivers were Iran-linked militias striking Eastern Province oil facilities in Saudi Arabia from Iraq, US/Saudi strikes on eastern Iraq, and the IRGC ballistic-missile surprise attack. Magnolia Oil & Gas was upgraded to buy at Truist.
• ⚠️ This edition's downgrades relative to the first draft: ① all attacks were intercepted and no production loss was reported — a completely different pricing basis from "oil facilities were hit"; ② Iran's foreign minister held separate talks with Saudi Arabia and Oman over the same period, so diplomatic channels are active; ③ the Hormuz closure is a carry-over fact, not a new overnight positive; ④ sector direction conflicts: XLE was still −1.35% on 7/28. All told: the theme strength stays at the first draft's A, but logic robustness is cut from "medium-high" to "medium" and persistence from "medium" to "medium-low."

9Intuitive SurgicalISRG60 pts · watch closely

• UBS upgraded to buy, target $500 → $550. The sector, XLV, is +2.36% and at an all-time high — the only name in this report with the double tailwind of "single-stock positive + sector all-time high." Ratings-driven moves are inherently weakly persistent (typically 1–2 days), so the weight comes mainly from sector momentum rather than from the upgrade itself.

10NvidiaNVDA55 pts · watch only

• On 7/28 it was the only gainer (+0.25%) among 20 semiconductor names, closing at $197.01. On 7/27 its 5-year CDS widened 14bp in one day to 82bp, the largest single-day move since the contract began trading in November 2025; the stock fell −5% the same day.
• Judgment: the price decoupling and the credit widening are happening at the same time — this is not a positive, it is a disagreement. Nvidia holds equity in or guarantees debt for customers such as OpenAI and CoreWeave, and those customers then use the money to buy Nvidia hardware; that loop makes demand look stronger than true end demand.
• ⚠️ Data grading note: this CDS figure is a second-hand source, TNW relaying Bloomberg, and dated 7/27 (two trading days ago). Yesterday's recap listed it as "not obtained" on two consecutive days. This edition does not judge it as "gap resolved," but downgrades the label to "second-hand relay, awaiting primary confirmation" (the first draft marked it ✅ resolved, too loose a standard, now fixed). From the same source, "CoreWeave's 5-year implied default probability of 42%" is an extreme figure from a single source, for reference only, not entering conclusions.

---

6. Negative / Avoid List

---

7. Within-Theme Rankings

Theme 1: AI power / data-center electricity

The most informative point in this theme: BE and VRT are in the same theme, reported in the same period, and are moving in opposite directions. The market is currently rewarding "visibility on new orders" and punishing "margin delivery on existing orders."

Theme 2: Semiconductors (direction leaning negative, internally split)

⚠️ The ranking contains an unverified-hypothesis component: TER's position depends partly on the "back-end vs. front-end" hypothesis, which currently rests on TER as a single point of support (see §2 Theme 5).

Theme 3: Value rotation / momentum breakdown

Theme 4: Energy / geopolitics

---

8. Open-Session Verification Signals

Pre-market (before 09:30)

Intraday

• First 30 minutes: focus on whether BE (+8.45%) and TER (+7%) gap-and-go or gap-fill. ⚠️ Yesterday's most serious self-criticism was that "names that delivered generally printed their high at the open" — if that happens a third time this week, it should be hardened into a scoring rule.
• Sector ETFs: whether the all-time highs in XLV / XLF can extend past the FOMC.
• Volume: the pattern over the past two days is "volume is still on the falling semiconductors, while the rising sectors are absorbed on lighter volume." If the rising sectors trade heavy volume for the first time, that is the key signal that the style shift has been upgraded from "money shuffling" to "money inflow."
• Breadth: ⚠️ NYSE advance/decline counts not obtained for a third consecutive day.

Options and volatility

• MSFT ±7% (call/put 1.9:1, implied market-cap swing about $189 billion), ARM ±14.9%, QCOM ±12.8% (historical average 7.1%).
• ⚠️ IV crush: MSFT / META / QCOM / ARM / LRCX all report tonight, so buying options before the print can lose money even with the direction right, simply from IV collapsing; overnight IV typically drops 30–50% after a print.
• VIX 18.21 (7/28 close) — it eased yesterday even on a down day for the Nasdaq, which says "the market is selling AI, not selling stocks." ⚠️ Real-time pre-market VIX not obtained; if it breaks above 20 around the open, that means this meeting's pricing has shifted from "non-event" to "risk event."

Risk signals

1. Gap and then reverse: BE (+8.45%), TER (+7%) — historically this week such names printed their high at the open.
2. A lone gainer with no sector follow-through: BE against XLU, TER against SMH — the strongest names pre-market are all inside weak sectors, which is the structure this report most needs to watch.
3. ⚠️ Real-time pre-market rate/FX readings are missing: VIX 18.21, 10Y 4.606%, DXY 101.393 are all 7/28 closing values. The first draft's assertion on that basis that "the 10Y is flat" is an unsupported inference and this edition retracts it. They need to be re-pulled before the open: if the 10Y breaks above 4.70% on a fast move up in oil, a positive open in the indices will be hard to sustain.
4. Everything before 14:00 is low-information price action.

---

9. Final Conclusions

① The 5 names most worth watching today

Selection rule (missing from the first draft, added and actually applied here): take those tagged "watch closely" in §3.1 and take the top 4 by total score in descending order (TER 69 / BE 68 / F 66 / ISRG 60); plus MSFT as the "event anchor" (total 67, tagged "watch only") — because its capex guidance tonight simultaneously determines the direction of three themes, it is included to be watched, not to be bought. ⚠️ One note outside the rule: XLE / MGY (total 58, §3.1 rank 8) does not qualify under the rule, but its thematic importance today outranks its score — the overnight oil reversal is the direct basis on which this report overturned yesterday's conclusion. It is therefore listed separately below the table, explicitly marked as a thematic anchor rather than a selection result, and not mixed into the top 5.

The thematic anchor outside the rule — XLE / MGY (total 58): the overnight oil reversal is the direct basis on which this report overturned yesterday's "beneficiaries of falling oil," but its pricing basis is far weaker than the first draft assumed (all attacks intercepted, no production loss, diplomatic channels active, and the sector still pointing down over two days). Verification point: whether XLE can turn positive pre-market — if it cannot, the market is reading oil as "inflation risk" rather than "energy earnings," and that is a worse signal for the broad market.

② The 3 strongest themes today

③ What to avoid today

1. Heavy positioning in any direction before 14:00. Three events stack up (FOMC + press conference + five earnings reports), and pre-market price discovery is incomplete. Listed as the first avoidance item for a third consecutive day.
2. Chasing the gapping names (BE +8.45% / TER +7%). "Names that delivered printing their high at the open" has already happened twice this week; TER's give-back from +13.5% after hours to +7.0% pre-market is early confirmation.
3. Taking part in the binary events at ARM / QCOM / LRCX / CAPR before they land. ARM ±14.9%, QCOM ±12.8%, LRCX reports tonight, CAPR has its FDA advisory committee today. ARM comes with an extra warning: the 7/27 recap built on it precisely the theme that was later falsified.
4. Bottom-fishing KLAC. ⚠️ The rationale has changed from "reason unknown" to "reason known and valid": guidance beat only marginally + memory pricing may suppress gross margin into 2027.
5. Treating micro-cap pre-market moves as a signal.
6. Treating BE as a clean positive trade before it reclaims $188.18.
7. ⚠️ New: pricing the FOMC off a single prediction market's number. Same-day readings from CME / Polymarket / Kalshi run from 6.6% to 38%, and taking a range or mixing them will produce a wrong conclusion.

④ Final one-sentence judgment

There are in fact quite a few company facts already on the ground before today's open — BE, TER, Ford, VRT, CSGP, FVRR, Visa, UPS, KLA and SK hynix have all put out numbers; what is genuinely undecided is the macro (14:00) and tonight's five reports. (⚠️ The first draft wrote "the only certain facts are three," contradicting the dozen-plus landed earnings reports it listed itself; corrected.) The most dangerous thing structurally is that the two strongest names pre-market (BE against XLU, TER against SMH) are both lone gainers inside weak sectors, and yesterday's recap had just concluded that "every single-stock conclusion must first clear the sector-direction gate." Hence this report's tone: thematic judgment can be aggressive, position judgment must be restrained — today suits watching verification points, not establishing directional positions.

---

10. Open Items (flagged proactively by this report)

1. The conflict between the Nasdaq 100 correction and yesterday's "rotation, not panic" conclusion — the Nasdaq 100 formally entered a technical correction on 7/28, while yesterday's recap concluded qualitatively that this was "rotation, not panic" (basis: VIX easing, breadth positive, 7 sectors up, XLV/XLF at new highs). The two need a head-on re-examination, which this report has not completed.
2. ✅ Closed — BE's H1/H2 split: H1 = Q1 $0.7511 billion + Q2 $1.0654 billion = $1.8165 billion; H2 needs to do $2.0835–2.3835 billion; pre-print consensus for Q3+Q4 was $2.0967 billion, sitting at the low end of guidance. ⚠️ Also clarified: the pre-print media line of "H1 revenue $751.1 million" is in fact Q1 standalone, mislabeled as H1. (Closed via the yfinance re-verification.)
3. The reason for UPS's "beat on both lines yet down 6.57%" — exactly the same pattern as KLAC. KLAC has been established, UPS has not. The first draft handled the two by inconsistent standards; this edition lists both as open items.
4. Whether FVRR simultaneously maintained/raised EBITDA guidance — the same blind spot as CSGP (cut revenue guidance, protect profit guidance). If it holds, FVRR's negative grade needs to be lowered. This report has not verified it.
5. Conflicting consensus conventions for MSFT — $89.37 billion / $4.33 vs. $87.50–87.67 billion / $4.22–4.24. Stated side by side without adjudication.
6. Contradictory prior-close conventions for crude — yesterday's recap recorded Brent Sep'26 closing at $85.00, while today's source implies a prior close of about $84.09, a difference of about $0.91; the two cannot both be true. ⚠️ And this report did not do the cross-check with the three ETFs USO/BNO/OIH on a no-roll basis the way yesterday's recap did — the verification standard used to overturn yesterday's conclusion is lower than the standard used to establish it, which is a methodological asymmetry that must be made good at the recap.
7. Missing real-time pre-market readings — VIX / 10Y / DXY are all 7/28 closing values; 9 of the 13 names have no genuine pre-market quote.
8. NYSE advance/decline counts — not obtained for a third consecutive day.
9. NVDA 5Y CDS — the figure was obtained (82bp) but it is a second-hand relay dated 7/27, so it is not judged as a resolved gap.
10. Independent yfinance re-verification: BE/QCOM/ARM done, MSFT/META did not return —
    • ✅ BE / QCOM / ARM re-verification is complete, and it corrected 5 substantive points as a result (see revision log R15). ⚠️ During the process the Python wrapper layer for yfinance kept returning 429 Too Many Requests, so the subagent switched to hitting Yahoo's underlying interfaces directly (v8/finance/chart, v10/finance/quoteSummary, ws/fundamentals-timeseries); the data source is the same as yfinance, but the retrieval path differs from the established process and must be recorded.
    • ❌ The yfinance re-verification for MSFT / META had not returned as of writing — therefore all financial figures in §5③ (MSFT) and §5⑦ (META) come from media consensus and are not independently re-verified, and MSFT's consensus itself has conflicting conventions (open item 5). These two are tonight's main events, so the gap is worth closing at the recap.
    • ⚠️ Yahoo's fundamentals-timeseries has not yet recorded BE's Q2 (still stopping at 2026-03-31), so BE's Q2 gross margin, operating profit, operating cash flow, and period-end cash and debt cannot be confirmed for now; the 10-Q must be checked.
11. The share-count assumption implied by BE's FY26 per-share guidance — management gave per-share figures of $2.55–2.85, while the diluted share count is already +37.5% over a year. The share-count assumption implied by that guidance is not disclosed, and if the FY26 average diluted count keeps rising above 340 million, the same net income maps to a materially smaller EPS. Also needed: confirmation of the convertible conversion price and whether there is a capped-call hedge.
12. The nature of that $7.370 billion one-off net income in QCOM's FY26Q2 — it pushes the trailing P/E down to 17.5x, so every valuation comparison on a trailing basis is wrong until it is stripped out.

---

Appendix: Retraining Execution Checklist (acting on yesterday's 7 §6⑤ feedback items)

---

Appendix: QC Revision Log (first draft → this edition)

risk-auditor found 14 red flags; below are the substantive conclusion changes that have been externally re-verified and corrected. Items not listed were format/consistency fixes.

---

⚠️ Risk disclaimer: this list is pre-market information gathering and observation only and does not constitute investment advice. US equities are volatile with high pre-market gap risk, and post-earnings IV crush and guidance reversals occur; automatically generated content may contain stale information or factual errors. Rely on company disclosures/SEC filings, and do not use this directly as a basis for trading.