U.S. Pre-Market Brief | 2026-08-19 (ET) Wednesday

Coverage window: 2026-08-18 16:00 ET regular-session close → 2026-08-19 08:34 ET (includes the 8/18 after-hours session, the overnight Asia/Europe sessions, and this morning's pre-market). About 56 minutes remain until the open (09:30 ET).

Quote conventions: pre-market prices, pre-market volumes and prior closes for single stocks and ETFs are taken from the stockanalysis.com quote API (Pre-market field. Important: the update timestamps of individual tickers are NOT synchronized — measured spread reaches 08:17–08:34; every pre-market move quoted in this brief carries its read timestamp, and where none is marked the figure comes from the full re-pull at 08:30–08:34. Cross-ticker comparison is indicative only; snapshots taken at different moments do not constitute a strict cross-section); index futures, Treasury yields, VIX, crude, the dollar and gold are taken from the CNBC quote API (read 07:51–08:02 ET). Earnings figures are taken exclusively from company press releases / SEC 8-K originals, links in the body.

Methodology statement for this brief (please read first): there are two mutually independent main lines today — do not read them as one. ① Biotech: the first Phase III success for an mRNA cancer vaccine, but the companies did not release any quantitative efficacy data from this Phase III (no HR, no absolute benefit). The market is pricing the Phase III off the Phase II KEYNOTE-942 HR=0.51 — that Phase II anchor is not stale: Merck released a 5-year (median 60.3 months) update at ASCO on 2026-06-01 and the HR is still 0.51, a point that matters to both bulls and bears. But using a point estimate from an n=157 Phase II to price an n=1,137 Phase III that has added a lower-risk stage IIB/IIC population remains the largest evidence gap in this event. ② Retail earnings: the EPS of HD (yesterday), TGT and TJX today are all contaminated by the one-time IEEPA tariff refund. All retail EPS in this brief are given on an "ex-refund" basis, because the three companies do not account for the same refund consistently — TJX stripped it out of adjusted EPS, Target did not. Reading the headline EPS alone leads to exactly the opposite conclusion.

---

0. Today in one sentence

Today is a split tape of "a single-point biotech detonation + weak stabilization with violent dispersion in AI hardware," not a broad recovery in risk appetite. The strongest catalyst is undisputed: Merck/Moderna announced at 06:45 ET that the Phase III INTerpath-001 trial of the personalized mRNA cancer vaccine intismeran autogene plus Keytruda met its primary endpoint. MRNA in the pre-market spiked as high as +108.86% to $131.50 and had pulled back to $113.88 (+80.87%) as of 08:19 (14.33 million shares); MRK +7.27% at $145.00 (1,128,784 shares, now above its 52-week high of $137.98); BNTX +14.96%; XBI +2.37%, XLV +1.34% — this is the only direction today with real volume behind it.

⚠️ Primary trading note: MRNA's pre-market price traveled $131.50 → $113.88 → $118.90 within 30 minutes, an amplitude equal to 28% of the prior close, on 16.33 million shares (2.4× its full-day average volume). Any single-point citation (whether "+108%" or "+81%") is misleading — the only reliable statement is a range, and which end of that range the opening print lands on cannot be predicted. See the timestamped trajectory table in 5.2.

But the other half of the tape is not "AI hardware keeps falling" — it is "weak stabilization + violent single-stock dispersion," and this is a conclusion that was revised by this brief's own data source within the hour before the open; the version below is the one to use. Around 08:00 yesterday's worst-hit names were indeed uniformly negative, but by 08:30–08:34 they had been systematically revised one-way higher: SNDK +1.89%, MU +1.21%, CRDO +1.05%, LITE +0.81%, FN +0.66%, STX +0.26%, TER +0.18%, COHR +0.01%, SMH +0.16%, with WDC narrowing from −1.41% to −0.19%. FN, down −19.38% yesterday, is positive this morning, and SNDK, down −9.01% yesterday, is leading — the claim of "zero bounce, no bid" no longer holds as of 08:30.

What actually deserves attention is the extreme dispersion inside the same sub-sector, and I was unable to find a same-day primary catalyst for it:

Both are core custom AI ASIC names, moving in exactly opposite directions, and neither on light volume. The pattern closely resembles 2026-04-20 "Google in talks for Marvell to participate in custom AI inference chips, Broadcom under pressure", but I could not verify a same-day primary news source for 8/19 — this brief therefore reports the fact and offers no attribution. Treat it as the first thing to confirm after the open, not as an established main line.

ADI must be downgraded in how it is described: it did not report after yesterday's close; it reported pre-market this morning at 07:01 ET (revenue $4.02 billion / +40%, Q4 guidance $4.3 billion vs the Street's $4.08 billion), pre-market +1.43%. This means: ① yesterday's −3.50% in ADI happened before the print, so it cannot be paired with this morning's gain to form a "good news ignored" contrast; ② the reaction window is only about 90 minutes of thin pre-market; ③ the 10:00 ET call has not yet been held, so segment detail and order commentary are not out yet. This brief therefore withdraws the inference that "a big beat bought only 1%, showing the market will not price AI hardware fundamentals" — 90 minutes of pre-call pre-market cannot support that terminal judgment.

Driver types: ① product/clinical (biotech, grade S) ② earnings and guidance (retail + semis) ③ macro (14:00 ET July FOMC minutes) ④ geopolitics (day 4 of the U.S.–Iran stalemate lifting crude).

Pre-market state: S&P futures 7,716.25 (+0.03%), Nasdaq futures 29,517.25 (−0.23%), Dow futures 53,457 (+0.10%) — Dow up, Nasdaq down, same direction as yesterday; the rotation is still running. 10Y Treasury 4.686% (−2.0bp), 30Y 5.27% (−1.5bp), yields falling for a second straight day; VIX 15.72 (−0.76%); dollar index 99.419 (−0.24%); Brent $91.56 (+0.59%, fourth straight gain), WTI $85.48; gold 4,418.60 (−0.05%). Asia was soft overnight (Nikkei −2.5%, Korea's KOSPI −1%, taking the handoff from yesterday's U.S. semiconductor rout).

---

1. News overview

Two known gaps in this table (clients please note): ① this brief contains no analyst rating changes for today (8/19). The "Wednesday's Top Wall Street Analyst Research Calls" returned by search was verified to have been published on 2026-08-05, and the 12 items in it including BBY, BURL, TDG, HUM and DELL were all two-week-old news and have been removed in full; no reliable same-day rating source was found afterwards. This is a coverage gap, not the same as "no rating changes today." ② this brief contains no intraday data for today (the 10:30 EIA crude inventories, the 13:00 20-year auction and the 14:00 FOMC minutes are all released after this brief was finalized).

---

2. Strongest themes, descending

---

3. Single-stock strength ranking

3.1 Positive direction

3.2 Negative direction

Note: COHR / MRVL / WDC / FN / MU / SNDK / LITE / CRDO have been removed from this table — in the 08:30–08:34 re-pull they all turned positive or were roughly flat, and no longer qualify as "negative direction." This contradicts the 08:00 version of this brief; this table is the one to use.

---

4. Single-stock scoring model (100 points)

---

5. Detailed analysis of top names

5.1 MRK (Merck 默沙东) — priority deep-dive | 87 pts

Related news: 2026-08-19 06:45 ET, Merck press release. Phase III INTerpath-001, n=1,137, fully resected stage IIB–IV cutaneous melanoma, 2:1 randomization, intismeran 1mg q3w (up to 9 doses) + Keytruda 400mg q6w (up to 9 cycles, about 1 year) versus Keytruda monotherapy. The RFS primary endpoint was met and the DMFS key secondary endpoint was met, with no new safety signals. (link¹)

First, falsify a popular explanation: what rallied today is not "the patent cliff pushed out." The intuitive story goes — a personalized vaccine must be combined with Keytruda, so Keytruda turns from "a monoclonal about to be replaced by biosimilars" into "the mandatory backbone of a combination regimen," and the patent cliff gets rewritten. But today's price data directly rejects that explanation:

The absolute dollar increments of the two are close to 1:1 (1.18:1), and the Merck–Moderna contract is a 50/50 split of global costs and profits (Merck 10-K original wording: "share costs and will share any profits equally", exercised in 2022 for $250 million). If the market truly believed the vaccine can defend Keytruda's $32.9 billion installed base, that value accrues 100% to Merck and Moderna gets nothing, so MRK's increment should be several times MRNA's. It is not. So what the market is pricing is "a new-product option split 50/50 + a platform read-across," not an extension of exclusivity.

"Combination equals life extension" also has four hard holes at the mechanism level: ① a combination cannot force a tie-in to the branded drug — after December 2028 doctors will pair intismeran with cheap biosimilar pembrolizumab, and Merck has no legal tool to stop them; ② the 10-Q explicitly states intismeran is being evaluated in combination with "Keytruda or Keytruda Qlex," so the vaccine is agnostic to formulation and will not lock patients into the protected one; ③ the IRA is a parallel second blade — Keytruda is expected to be selected for government price setting in 2027, effective 2029-01-01, and proposed CMS rules may sweep in the subcutaneous Qlex as well, against which the combination narrative is entirely ineffective; ④ the adjuvant melanoma market cannot support $24.2 billion (see below).

So is +7.3% justified? Change the denominator and the answer changes. MRK's GAAP TTM P/E of 109x carries no information at all — it is contaminated by two fully expensed IPR&D acquisitions (Cidara $9.0 billion, Terns $5.7 billion, together $6.05/share). Stripping out the one-time items, underlying FY26 EPS ≈ $8.71–$8.81, which puts the prior close of $135.17 at 15.4x and the pre-market $145.00 at 16.5x (the pre-market peak of $147.98 would be 16.8x). That is, +7.3% ≈ +1.1 PE turns. Turning a company with "no story after 2028" into a company with "one platform story" and re-rating it by 1.4 turns is not a demanding partial release of a discount factor — it does not need a rigorous NPV to justify it. That is a more likely true cause than "a $24.2 billion upward NPV revision" (the latter requires extrapolating all 9 trials to success).

But most of the room for discount repair has already been used up (contrary to the popular impression that "MRK is a cheap cliff stock"): the forward P/E has already gone from 8.83x (2025Q2) → 15.69x (now) in 12 months, a +78% multiple expansion, all of it before today's news; the stock is already +74% from its 52-week low of $77.58 to yesterday's close, and the 52-week high was set yesterday. Today's move pushes it to roughly 16.5x, close to ABBV (17.1x) and already above AZN (14.8x), BMY (10.3x) and PFE (9.8x). The pre-market price of $145.00 is already above the sell-side average target of $136.85.

Theme stage: launch day (news released 06:45, first candle of the day), but the valuation stage is already mid-to-late.

Attribution limit (clients please note): a single-day +7.3% in a $333.5 billion company is an extreme move. This brief attributes it entirely to the INTerpath-001 readout, but has not exhaustively checked whether MRK has any other catalyst today. If another piece of news surfaces after the open, the whole attribution in section 5.1 needs to be re-assessed.

Pre-market and technicals: pre-market $145.00 (+7.27%, 08:18 read), 1.129 million shares, now above the 52-week high of $137.98. The pre-market path was $147.98 (peak) → $147.35 (08:01) → $145.70 (08:08) → $145.00 (08:18), a fade of only −2.0%, far smaller than MRNA's −13.4% — large-cap buying is clearly steadier. Note that the pre-market price is already above the sell-side mean target of $136.85.

5.2 MRNA (Moderna 莫德纳) — watch closely (do not chase) | 80 pts

Related news: same as above; MRNA is the co-developer.

Catalyst logic: this is the first oncology Phase III success for the mRNA platform, and its meaning goes beyond a single indication — the INTerpath program comprises 9 Phase II/III trials covering NSCLC, bladder cancer and renal cell carcinoma, plus Phase I studies in pancreatic and gastric cancer. Once the platform is validated, the probability of success (PoS) for subsequent indications should be revised systematically higher.

How much the market actually priced this asset at today (shares out 399,235,889, per the 10-Q cover):

The two increments are close to symmetric (1.19:1), consistent with the 50/50 split, and the ratio stayed stable through the pre-market fade — which says the market is pricing the asset itself rather than a one-sided sentiment re-rating of MRNA.

The limits of this reasoning (which must be stated, or it will be taken as stronger evidence than it is) — this brief downgrades it from "falsification" to "suggestive observation," for four reasons:

1. The numerator is contaminated, and this brief admits it itself: section 5.2 states that "a substantial part of this price is short covering and market-maker hedging colliding in the least liquid part of the session" (short interest is 13.37% of the float). If MRNA's increment is mechanically amplified by a squeeze, then the true fundamental increment is smaller and the true ratio is larger — which is entirely compatible with "some exclusivity value is being priced." Dividing by a denominator known to be contaminated yields an unusable quotient.
2. It attacks a straw man: nobody argues that "100% of the $32.9 billion installed base is priced in." The present value of an exclusivity extension is a forward cash flow that is "probability × post-2029 × already compressed by the IRA," and its PV could plausibly be only a few billion dollars — exactly inside the noise of that 1.1–1.2:1 band. This test cannot distinguish "zero exclusivity value" from "$5–10 billion of exclusivity value"; it has no resolving power.
3. There is no null hypothesis, no error band, and the ratio itself drifts: computed with prices from different moments, the ratio swings between 1.09 and 1.19. A statistic that swings 9% on its own within an hour cannot carry a conclusion as heavy as "falsification."
4. The signal-to-noise ratio differs by an order of magnitude, and the prior points the opposite way to intuition: $24 billion is +7.3% for MRK, inside the intraday noise band of a $333.5 billion company (and mixed in with XLV +1.34% of sector beta and its own momentum from setting a new high yesterday); MRNA's increment is nearly pure event. More importantly: the incremental Keytruda volume that the combination generates accrues 100% to Merck and does not enter the INT profit pool — so the prior should be "MRK's share > 50%" to begin with. Observing 1.1–1.2:1 is therefore weak support rather than a refutation.

The correct statement is therefore: the approximate symmetry of the market-cap increments is an observation compatible with "a 50/50 new-product option + platform read-across as the main driver"; it weakens but does not falsify the "exclusivity extension" narrative. Truly testing the latter would require a ratio significantly and stably above 1.5, not today's 1.1–1.2 sitting inside the noise.

Backing out an implied peak sales figure — the entire passage below is a hypothetical exercise, not market consensus and not the actual output of any sell-side report: the multiple used is 3× MRNA's attributable peak revenue (range 2–4×), a 50% split, with no probability of success applied and no discounting. On that basis today's pricing implies global peak sales for intismeran of roughly $10–20 billion, midpoint about $13.5 billion; if the multiple is changed to 2× or 4×, the implied peak becomes roughly $20 billion or $10 billion — showing that the conclusion is extremely sensitive to this unverified multiple. And the numerator (the $20.3 billion increment) itself contains a squeeze component, so this "implied peak" is an amplified ceiling, not a neutral estimate.

Another calibration point: the "most bullish target price on the Street before the readout, $77," cited in this brief is the maximum; the consensus target is actually about $48.92, with a consensus rating of Hold. Anchoring on the consensus rather than the maximum, the current price deviates from the Street by even more than described above.

And that number cannot possibly be reached on melanoma alone: at the third-party-cited pricing assumption of roughly $200,000 per patient (that pricing assumption comes from media accounts of a Jefferies view, not from a primary report), $13.5 billion of peak sales requires about 68,000 full-price patients per year. U.S. invasive melanoma incidence is about 112,000 cases a year, of which stage IIB/IIC is about 7% and regional (stage III) about 10%, so the addressable adjuvant pool is only about 19,000–20,000 cases a year; scaling that up across developed markets and applying 60% penetration still yields only about 26,000 patients → global end-market annual revenue of roughly $5.2 billion, and profit attributable to MRNA (the 50% split) of roughly $1.3 billion a year.

Rough conclusion: adjuvant melanoma on its own is worth only about $3–6 billion of EV, meaning roughly 60%–80% of today's move depends on NSCLC / bladder cancer / renal cell carcinoma — and not one of those has read out. The order-of-magnitude fact supporting the bulls is: the addressable population of NSCLC alone (about 50,000–60,000 cases a year) is roughly 3× melanoma's, and the two Phase III trials INTerpath-002 and -009 are already enrolling, with the renal trial INTerpath-004 fully enrolled.

The biggest problem — the price reaction far exceeds the strength of the evidence: the company explicitly did not release the HR or the magnitude of absolute benefit, and STAT News recorded that it "did not immediately release detailed data." The market is currently pricing an n=1,137 Phase III off Phase II KEYNOTE-942 (n=157) with HR=0.51 (RFS) / 0.411 (DMFS). "Statistically significant" is perfectly compatible with an effect size far smaller than the Phase II's — a Phase III effect size smaller than Phase II is the norm in drug development, not the exception. On top of that this is an interim analysis with OS immature, and the adjuvant setting has precedents where an RFS benefit failed to convert into an OS benefit.

Fundamental verification (SEC 10-Q, 2026-06-30) — three facts that run against the mainstream narrative:

1. The cash is not as thick as imagined. Cash and investments are $6.91 billion, but the 10-Q states explicitly that the $950 million litigation settlement was paid in July 2026, leaving roughly $5.96 billion truly available. The company itself guides to cash falling to $4.7–5.2 billion by the end of 2026. Normalized annual cash burn is about $1.9–2.0 billion, giving a runway of roughly 2.4–3.5 years. The credit agreement contains one more hard constraint: cash may not fall below $500 million at any week-end (rising to $750 million if more than $1 billion is drawn), and that money is unusable.
2. The revenue collapse has in fact already bottomed, and this is widely misread. FY2021 $18.47 billion → FY2025 $1.94 billion (−89.9%) is history, but Q2'26 revenue of $145 million was +2.1% year over year, and the company's 2026 guidance is for "up to 10% growth." "Revenue is still collapsing" no longer holds for 2026.
3. There are no convertibles; the only debt is the November 2025 $591 million senior secured term loan at roughly 9.20% (SOFR+5.50%), secured by a first lien on substantially all assets. Borrowing at that rate and on those collateral terms is itself hard evidence of the financing position it was in at the time.

How to read the issuance risk correctly: there is no near-term forced-issuance cliff (cash runs out around 2029). But raising $2 billion at $62.96 dilutes 16%–20%, while at $122 it dilutes only 5.5%–6.5% — opportunistic financing after good news is a standard industry move. What could actually trigger a raise is not routine R&D burn but the capex for personalized manufacturing: current capex guidance is only $200–300 million a year, an order-of-magnitude gap versus "supporting tens of thousands of one-patient-one-batch runs a year," and MRNA is the party responsible for manufacturing in the collaboration. The buffer is that INT costs and profits are split 50/50 with Merck, so MRNA's net quarterly burden is only $97 million.

Two recent facts buried by today's headline (pointing in opposite directions; both must be stated):

• Positive: the flu vaccine mFLUSIVA (mRNA-1010) was approved by the FDA on 2026-08-05 — the first mRNA flu vaccine ever and the company's fifth approved product.
• Negative: the interim analysis of norovirus candidate mRNA-1403 failed to meet the early success criteria in Phase III (disclosed 2026-07-31, less than 3 weeks ago). That is a major setback in a large indication.

Pre-market and technicals — the one table in this brief readers most need to look at. Prior close $62.96, 52-week high only $85.60. MRNA's pre-market price is not a number; it is a curve moving one way, down:

Falling from the 08:02 peak of $131.50 to $113.88 at 08:19 (−13.4%) and then rebounding to $118.90 within 13 minutes is not a one-way decline; it is high-volatility range trading. What I wrote in the 08:19 version — "one-way down and still falling" — has been overturned by the 08:32 data and is corrected here. Pre-market volume has already reached 2.4× the 3-month average full-day volume (6.86 million shares), so the quote is credible, not thin-volume noise.

This curve is itself the most important trading information today: any single-point citation (whether "+108%" or "+81%") is misleading. The only reliable statement is a range: the pre-market traded between $113.88 and $131.50, an amplitude equal to 28% of the prior close. Which end of that range the opening print lands on cannot be predicted — this is a textbook case for the rule "do not chase a stock that has already gapped up hugely," and the reason is not "it is falling" but "its price has not been discovered yet."

There is also a layer of "already partly front-run": the price path is $22.28 (2025-11-17) → $85.60 (2026-07-06, +284%) → $62.96 (8/18). That move in early July was very likely already front-running this readout, and before that move the EV was already $18.8 billion — the market had already paid nearly $19 billion for the pipeline before today.

The hardest single point on each side (presented side by side, not reconciled):

The hardest bull point — the regulatory path does not need to wait for OS. In adjuvant melanoma the FDA has twice granted full approval (not accelerated approval) on RFS alone: pembrolizumab in stage IIB/IIC, 2021-12, based on KEYNOTE-716; nivolumab in stage IIB/C, based on CheckMate-76K (HR 0.42). The endpoint precedent is already laid, on top of which mRNA-4157 received FDA Breakthrough Therapy designation and EMA PRIME back in 2023-02. The capacity side is not hand-waving either: the Marlborough dedicated individualized-therapy plant is 140,000 sq ft plus a 60,000 sq ft expansion, and began supplying clinical batches in 2025-09. And the most bullish target price on the whole Street before the readout was Piper Sandler's $77, with a consensus rating of Hold — not a single existing sell-side model has the read-across of 9 trials built in; the revision is a rebuild, not a tweak.

The hardest bear point — a structurally near-identical precedent just blew up. IMbrave050 (atezolizumab + bevacizumab, adjuvant HCC): the interim analysis (median follow-up 17.4 months) met the RFS primary endpoint with a 28% reduction in the risk of recurrence/death, and was called a landmark at ASCO that year; after longer follow-up (median 35.1 months) the RFS HR retreated to 0.90 and the OS HR was 1.26, with the conclusion that the data "do not support use in the adjuvant setting." Adjuvant setting + RFS primary endpoint + pre-specified interim analysis + positive topline + immature OS — all five features match today.

But that analogy has one key asymmetry that must be stated at the same time: all of IMbrave050's lethality came from "the effect size collapsing once follow-up lengthened." Whereas intismeran's Phase II anchor has precisely survived the long-follow-up test — Merck released the 5-year (median 60.3 months) update of KEYNOTE-942 at ASCO on 2026-06-01, with the RFS HR still 0.51 and the DMFS HR still 0.411. In other words, this anchor is not "old 2023 data" but data reaffirmed 2.5 months ago, and its failure mode is the opposite of IMbrave050's. The bear's correct formulation should be: "the Phase II effect size is robust, but the Phase III switched to a lower-risk population and is 7× the sample size, so the risk of effect-size dilution comes from the population, not from follow-up duration" — which still holds, but is weaker than the IMbrave050 analogy.

There is one more piece of effect-size dilution arithmetic that the market as a whole skipped today:

Two key points: ① the 0.51 anchor comes from a 50-patient control arm and the upper CI bound is 0.887, so the true effect could easily be 0.85, yet the market is extrapolating from the point estimate; ② the Phase III diluted the population toward lower risk — stage IIB/IIC patients have a far lower baseline recurrence rate, so even with an unchanged HR the absolute benefit is mechanically compressed by the baseline event rate. At n=1,137, an HR of 0.75–0.80 is enough to reach p<0.05 at an interim analysis.

The manufacturing bottleneck is the real killer for commercialization, and it has a tombstone: literature turnaround is 4–6 weeks and manufacturing cost is >$100,000 per patient. Moderna's own wording already concedes the problem is unsolved — the company says it is working to "right-size the intismeran manufacturing process to improve turnaround time and lower cost." Compare Provenge / Dendreon: FDA approval in 2010 (on OS, a harder endpoint than RFS), $93,000 per course, COGS at one point reaching 77%; analysts once forecast $4.3 billion of annual sales by 2020, but actual 2013 sales were $284 million and the company went bankrupt in November 2014 — the cause of death being exactly complex, expensive individualized manufacturing.

Mechanics: short interest is 49.77 million shares = 13.37% of the float with 7.21 days to cover, and nobody got out before the readout; pre-market volume has already reached roughly 2× the full-day average. The current price is about 60% above $77, the highest target on the Street before the readout — a substantial part of today's price is short covering and market-maker hedging colliding in the least liquid part of the session, not a clearing price.

Conclusion: the direction is confirmed, but the risk/reward of chasing in the pre-market is terrible. This stock's value distribution is bimodal (melanoma-only approval is worth roughly $7–12 billion of EV, a cross-tumor platform is worth $66–120 billion); today's news shifts probability mass from the bear peak to the bull peak but eliminates neither peak. The core disagreement compresses into one sentence: when that HR is presented at an academic conference, will it be the validation of this trade or the end of it.

5.3 ADI (Analog Devices 亚德诺) — watch closely | 78 pts

Related news: press release pre-market at 07:01 ET on 2026-08-19 (not after yesterday's close; the earlier draft was wrong here). FQ3 revenue $4.02 billion (+40% YoY), GAAP gross margin 67.3%, operating margin 40.1%; adjusted gross margin 72.5%, adjusted operating margin 50.0%; GAAP EPS $2.74 (+163%), adjusted EPS $3.45 (guidance was $3.30). Q4 guidance: revenue $4.3 billion ± $100 million, vs the FactSet consensus of $4.08 billion; adjusted EPS $3.86 ± $0.15; adjusted operating margin about 52.0%. Growth was led by data center and industrial.

The print itself is very strong: revenue above its own guidance, Q4 guidance about 5.4% above the Street, FQ3 actual adjusted operating margin 50.0% (Q4 guided 52.0%) — a rare combination in analog semis, and one delivered through forward guidance rather than a current-quarter beat (by this desk's convention, guidance > current-quarter beat).

But this brief must withdraw an inference from the earlier draft, and explain why. The earlier draft wrote that "a big beat bought only +1%, showing the market will not price AI hardware fundamentals." That inference does not hold, for three reasons:

1. Timing mismatch. ADI reported this morning at 07:01 ET, not after yesterday's close. Yesterday's −3.50% in ADI happened before the print (it fell with the semiconductor group), so "−3.50% yesterday → +1.4% this morning" is not a "good news ignored" contrast at all; they are two unrelated events.
2. The window is too short and liquidity too thin. Less than 90 minutes elapsed from 07:01 to the time of writing, and pre-market volume was only 68,000 shares — a quote on that volume is not enough to represent the market's pricing verdict on an earnings report.
3. The most important information is not out yet. The 10:00 ET call has not been held. By this desk's experience, the 8-K and press release give the numbers, while segment detail, orders/backlog and commentary on demand durability all come on the call — the real pricing moment for this print is after the call, not now.

ADI's correct positioning is therefore "watch closely" rather than "counter-trend sample": it remains the strongest fundamental disclosure of the day, but using it to argue a sector-level conclusion is out of bounds before the 10:00 call.

⚠️ In this data pull, ADI's SEC 8-K original (EDGAR adi3q26exhibit991earnings.htm) and the investor.analog.com press release returned HTTP 403 and a timeout respectively, and stocktitan.net hit DNS timeouts twice in a row. ADI's segment revenue breakdown (industrial/auto/communications/consumer YoY individually) was therefore not obtained, and the Q3/Q4 figures were taken instead from the PRNewswire release and MarketScreener's account of the FactSet consensus, without word-for-word verification against the 8-K original. If ADI segment data is to be cited later, it must be re-pulled.

Segment breakdown: the individual YoY growth rates for industrial/auto/communications/consumer were not obtained this time, and this brief makes no segment-level judgment. **Note: ADI's Q3/Q4 figures come from the company's PRNewswire release (2026-08-19 07:01 ET) and have been checked item by item; but "the Street consensus of $4.08 billion," the denominator used to compute the "5.4% beat," comes from a single secondary account with no timestamp — the precise magnitude of this expectation gap should be used with caution.

5.4 TGT (Target 塔吉特) — watch closely | 66 pts

Related news: 2026-08-19 pre-market, Target press release / SEC 8-K.

The restatement that must be done (the core of this brief):

Refund amount: $994 million pre-tax / $752 million after tax / $1.65 per share.

Conclusion: 80% of the $2.06 year-over-year increment comes from the tariff refund ($1.65); measured against headline EPS it is 40%. But the underlying business really is improving. The genuinely hard evidence consists of three numbers that have nothing to do with the refund: comps +3.8% (+2.4% expected), traffic +3.6%, digital +8.7% (same-day delivery +25%) — positive traffic means this is share-driven rather than price-driven. Full-year guidance was still raised by $0.75 after stripping the refund out, which the refund also cannot explain.

Why is the pre-market nonetheless −0.50% (08:18, 468,000 shares): the stock is already up more than 50% year to date and near a two-year high (52-week high $156.47, prior close $152.48), so the good news in this print was already priced. This is the classic "delivered the numbers but the stock does not rally" pattern.

5.5 TJX — avoid | negative

Related news: Q2 FY27 comps +4%, revenue $15.2 billion ($15.19 billion expected), GAAP EPS $1.36 / adjusted $1.22 ($1.19 expected), adjusted pre-tax margin 11.9% (+0.5pp), adjusted gross margin 31.4% (+0.7pp).

Two traps that must be separated:

1. $1.36 is GAAP, while the $1.19 consensus is on an adjusted basis. TJX received $331 million of IEEPA tariff refunds in the quarter, a net contribution of $0.14, and the company proactively stripped it out of adjusted EPS. On a comparable basis it is $1.22 vs $1.19, a beat of only +2.5%, not the +14% the headline suggests.

2. "Raised full-year guidance" is very likely a misleading formulation, but the evidence here is weaker than for point 1: the raised FY27 range of $5.15–$5.20 versus a consensus of $5.22 — but I found another source putting the FY27 Street number at $5.19, and if it is $5.19 then the top end of guidance, $5.20, is actually slightly above consensus and this claim does not hold. The two consensus figures used in this brief, $5.22 and $1.34, come from a single secondary account with no timestamp, so do not size a position on them. The direction of Q3 guidance $1.30–$1.32 below the $1.34 expected is more consistently sourced.

3. A basis issue that must be flagged: per secondary accounts, TJX's Q3 guidance itself is still "aided by additional IEEPA tariff refunds benefiting cost of sales." If true, then this already-below-consensus Q3 guide is still being propped up by the next tranche of refunds, meaning the underlying business is weaker than it appears; it would also overturn the characterization that "the refunds are one-time and nearly finished." This brief did not obtain the TJX press release original for word-for-word confirmation, so this is marked as pending verification.

4. Readers cannot reconcile $331 million of refunds → a net $0.14/share directly: on roughly 1.103 billion diluted shares, $0.14 corresponds to about $154 million after tax and, backing out, about $205 million pre-tax, leaving a gap versus $331 million. The missing link is the company's disclosure that it was "partly offset by related incremental compensation accruals," which this brief had not previously spelled out and now adds. This also shows that TJX's "refund" and Target's "$994 million directly credited against COGS" are simply not the same basis, so cross-company comparison needs care.

Conclusion: a print that is "a small beat this quarter, guidance short next quarter, and a full-year raise that still sits below consensus." Pre-market −4.67% (08:18,318,000 shares, with the decline still widening) is fair pricing, not a mistake. What deserves credit is TJX's accounting honesty — it stripped the refund out, and Target did not.

5.6 LOW (Lowe's 劳氏) — avoid | negative

Q2 comps only +0.2%, GAAP EPS $4.27 / adjusted $4.40. Full-year guidance cut across the board to the low end of the range: revenue $92 billion (was $92–94 billion), comps flat (was flat to +2%), adjusted EPS $12.25 (was $12.25–12.75). Management explicitly described the DIY customer as under "pressure," and online +15.7% cannot offset it.

The restatement required by this brief's own convention (omitted earlier, added now): Lowe's likewise disclosed that IEEPA tariff refunds contributed about $0.11/share, and since GAAP $4.27 < adjusted $4.40, the adjustments are add-backs of expenses rather than a removal of the refund — meaning that $0.11 is still sitting inside the $4.40. LOW's comparable EPS is therefore about $4.29. This does not change the direction of the conclusion (comps +0.2%, guidance cut to the low end), but it makes the LOW-versus-TJX comparison more accurate: TJX proactively stripped the refund out; LOW and Target both did not.

Position matters too, and was omitted earlier: LOW's prior close was $215.64, only 8% above its 52-week low of $199.40 and already 26% below its 52-week high of $293.06 — the "guidance cut to the low end" is happening in a stock that has already broken down. Compare TGT, which is only 2.5% off its 52-week high: both are retail prints, but the two sit in completely different positions, and the same intuition about "good/bad news already priced" cannot be applied to both.

Reading it together with HD: HD closed −0.12% after yesterday's print, −2.02% relative to its pre-market price, and is down another −1.10% this morning. The home-improvement chain has been falsified two days running: HD only held its guidance by crediting $685 million of refunds against COGS, and LOW could not hold guidance even with refunds.

This morning's 08:30 macro data gives this chain an independent third-party check, but the direction is split: July housing starts of 1.239 million came in far below the 1.350 million expected, −12.4% MoM and −13.5% YoY, with single-family starts −9.9% MoM — corroborating LOW's "DIY under pressure" and HD's low-single-digit comps. But building permits over the same period were 1.443 million, +5.0% MoM and +3.1% YoY, which is strengthening.

The correct reading: current construction activity is clearly weakening (starts), but the forward-looking planned volume is recovering (permits). So "the demand side of this chain really is weak" holds only for the present and cannot be extrapolated to the next two quarters — permits typically lead starts. This is also the reason to avoid the home-improvement chain today: not that it has no future, but that current data and forward data are fighting each other, and there is no tradable direction right now.

5.7 BNTX — watch only | 68 pts

Pre-market +14.96% at $106.63 (08:18, 732,000 shares; peaked at +17.52%). After checking SEC filings, confirmed: BNTX's most recent 6-K is dated 2026-08-04 (Q2 results), and there is no filing or announcement of its own on 8/19 — today is a pure platform read-across.

But the sign of this read-across is not one-directional, and this is the layer most easily missed:

• Positive: the modality (individualized neoantigen) has been validated in Phase III for the first time, which directly raises the prior probability of success for BNTX's own BNT122 (autogene cevumeran, partnered with Genentech).
• Weakening: BNT122's lead indication is colorectal cancer in Phase II, with a different molecule, different delivery and a different partner. Modality validation ≠ molecule validation. And that trial is still blinded and has not read out (enrollment completed 2026-04; after a pre-specified interim analysis in 2026-06 the trial continued, and the final analysis has been pushed to 2027).
• Negative (almost nobody mentions this): BNTX's most heavily backed asset is not a cancer vaccine but pumitamig (BNT327, a PD-L1×VEGF-A bispecific, partnered with BMS, with 7 global registrational trials) — it is positioned to challenge/replace the position of PD-1 monoclonals. A validated INT that must be combined with Keytruda actually reinforces the pembrolizumab-anchored standard of care, which is not necessarily good news for the competitive landscape of BNTX's largest asset.

Financial comparison — this set of numbers is the sharpest in the brief (BNTX's reporting currency is the euro, converted at EUR/USD 1.1576 as of 2026-08-18):

Debt-basis note (affects the comparison above): on the MRNA side this brief uses the $591 million term loan disclosed in the 10-Q. If the total-debt basis including lease liabilities ($1.287 billion) is used instead, MRNA's EV rises by about $700 million and its net cash falls correspondingly. Both bases can be found in public filings; this table uses the narrow one; on the BNTX side the €316 million already includes lease liabilities, so the EV comparison above is slightly favorable to MRNA, and readers should adjust as they see fit.

BNTX's EV is only about 19% of MRNA's, while its cash is 2.8× MRNA's. BNTX is buying back its own stock; MRNA is borrowing at a 9.2% rate against all of its assets. But this is exactly why: BNTX's low valuation is not a problem today's news can solve — its discount comes from its own revenue decline (Q2'26 revenue €105.6 million, −59.5% YoY) and from the full-year guidance it cut just on 8/4 (revenue €1.6–1.9 billion, down from €2.0–2.3 billion).

Another base effect: BNTX was already a weak stock before today (prior close $92.75 vs a 52-week high of $124), so part of the +17.5% is oversold-bounce sector beta rather than a full asset re-rating. The risk/reward is asymmetric — watch only.

5.8 WYFI (WhiteFiber) — avoid | negative

Announced a proposed $250 million convertible on 8/18, then on 8/19 upsized the pricing to $270 million with a 5.00% coupon, due 2032, conversion price $33.84. Down −10.81% yesterday and −22.31% this morning (655,000 shares, 08:31), roughly −31% over two days.

One number that is easy to get wrong must be corrected: the conversion premium should be measured against the pricing reference price (prior close $27.07), i.e. $33.84 / $27.07 = +25.0% — this is about as standard a convertible premium range as it gets, not "deeply out of the money." Using the post-crash pre-market price as the denominator produces "more than 60% above," which is an arithmetic artifact of the subsequent share-price decline, not a signal from the terms of the deal.

This brief therefore does not treat it as evidence of "a tightening financing window for AI infrastructure" — a 25% premium plus a 5.00% coupon is expensive for a small-cap AI infrastructure name but not abnormal. Its only real significance is the dilution itself: upsizing the deal (from $250 million to $270 million) alongside −31% over two days says the market does not currently welcome incremental financing in this space. The characterization stops there; it is not extrapolated into a sector-level conclusion.

---

6. Negative / avoid list

Nine names — COHR / LITE / FN / CRDO / MU / WDC / STX / SNDK / DELL — have been deleted from this table: in the full re-pull at 08:30–08:34 they all turned positive or were roughly flat, so the factual basis for "avoid" no longer applies. The earlier draft's "all still negative this morning, zero bounce" was based on asynchronous snapshots from 08:0x and is void.

---

7. Intra-theme ranking

Theme 1: personalized mRNA cancer vaccine

Theme 2: retail (true ranking after stripping out tariff refunds)

Theme 3: AI hardware (direction revised from "broadly damaged" to "weak stabilization + internal dispersion")

All reads standardized to the full re-pull at 08:30–08:34 ET (the earlier 08:0x version gave the opposite conclusion and is void):

---

8. Verification signals at the open

Pre-market

• MRNA's gap-fade is already under way, and it is the single strongest signal this morning: peak $131.50 (08:02) → $113.88 (08:19), a 13.4% fade on rising volume in 17 minutes, with volume going from 11.26 million to 14.33 million shares. The "breaks $115" trigger fired before the open. The question now is not "will it gap-fill" but whether it can stop falling after the open: if it loses $105 (≈ prior close +67%) within the first 30 minutes, today's pattern is "good news fully discounted"; only if it can build a platform on volume at $110–115 is there anything to call a trend.
• MRK is a cleaner signal than MRNA: large cap, 1.129 million shares of pre-market volume, above its 52-week high, and a far smaller fade than MRNA's (peak $147.98 → $145.00, only −2.0%). If MRK can hold above $140 into the close, the re-rating has been accepted by institutions; and if the ratio of MRK's to MRNA's market-cap increment deviates clearly from 1:1, the market has changed how it prices the 50/50 split — that is the direct verification point for the reasoning in section 5.1.
• How to use ADI correctly: it only reported this morning at 07:01 ET, and the 10:00 ET call has not been held. The pre-market +1.43% represents nothing more than an initial reaction in 90 minutes of thin liquidity and cannot be used to judge "whether the market prices AI hardware fundamentals." The real observation point is the second round of pricing after the 10:00 call (segment detail, orders/backlog commentary), plus the full-day close — by this desk's rule "the 8-K has no guidance, the call does," the post-call price is the valid sample.

Intraday

• Sector ETF confirmation: XBI (pre-market +2.37%, 280,000 shares) and XLV (+1.34%) are the test of whether theme 1 broadens. Note that both are also fading with MRNA (XBI has slipped from +2.57% to +2.37%). If XBI's gain is clearly smaller than MRNA's/MRK's, this is a three-stock event, not a sector move — do not buy the second tier as if it were a sector move.
• SMH's direction: −4.09% yesterday, −0.05% at 08:17 and already +0.16% by 08:31 — this signal flipped within the hour before the open; do not treat it as confirmation. Only one usable test remains: whether SMH closes red or green for the day, and whether the dispersion between MRVL (+11.22%) and AVGO (−4.74%) narrows or widens after the open. If it widens, this is rotation inside custom ASIC rather than a sector-level judgment, and the narrative carried over from yesterday that "AI hardware is being de-levered as a whole" should be abandoned.
• Dispersion inside retail: TGT (−0.50%) vs TJX (−4.67%) vs LOW (−3.25%). If TGT turns green while TJX/LOW keep falling, the market has learned to read these prints ex-tariff-refund — which is itself a validation of this brief's methodology.

Options sentiment

• MRNA's IV will balloon today, but this is a clinical event, not an earnings report, so there is no standard IV-crush path — the data-release day (a future medical conference) is the real IV event. Buying MRNA options in the pre-market means paying an extremely high premium.
• IV crush after TJX's print will amplify the sideways drift after the drop.

Risks

• The 14:00 ET FOMC minutes are today's hard timestamp. At the July meeting 3 members argued for a hike, and if the minutes show the hawkish camp is stronger than the market understands, long-duration assets (including today's exploding biotech) will come under pressure. Price action before 14:00 does not represent the full day.
• Reversal after a gap: MRNA's +95% gap has very rarely been held intact into the close historically.
• A lone winner with no sector follow-through: if XBI spikes and fades while MRNA/MRK stay strong alone, the theme lacks breadth.
• Index futures diverging from Treasuries: today's falling yields + falling Nasdaq futures contradict the "rate-driven" narrative — consistent with yesterday's recap conclusion that the decline in AI hardware was not caused by rates, which is why an improvement in rates will not rescue it either.

---

9. Final conclusions

① The 5 stocks most worth watching today

② The 3 strongest themes today

③ What to avoid today, and why

1. "Bottom-fishing" AI optical interconnect and memory — this item has been overturned by the data and is retained to show the correction: at 08:0x these names were indeed uniformly negative, but after the 08:30–08:34 re-pull FN/LITE/CRDO/MU/SNDK/STX/COHR all turned positive or flat. The correct statement is: yesterday's selloff got an initial bid this morning, but the strength of that bid has not yet been tested at the open. What should actually be avoided is AVGO (−4.74%, still falling on volume with no same-day primary catalyst), and chasing MRVL before the reason for its +11.22% has been established.
2. The second and third tiers of the mRNA theme (BNTX/ARCT/NVAX): no catalysts of their own, already up a lot, pure read-across. NVAX +6.68% purely because of the "vaccine" label, with no technical connection to personalized neoantigen therapy.
3. The home-improvement chain (HD/LOW): two consecutive days and two independent companies pointing to soft current demand, with the 08:30 housing starts print of −12.4% (far below expectations) providing third-party corroboration — but building permits over the same period were +5.0%, so the forward indicator is strengthening. Current and forward data are fighting each other, there is no tradable direction right now, and that is the reason to avoid.
4. Every ticker with pre-market volume below 10,000 shares (AMGN 330 shares, REGN 229 shares, BBIO 45 shares, DG 370 shares): the quotes are unreliable and cannot be a basis for decisions.
5. Large bets before 14:00 ET: the FOMC minutes contain 3 members who argued for a hike, and the hawkish risk is not priced.

④ Final one-sentence judgment

Only one thing is truly certain today: a press release that did not even disclose a hazard ratio moved MRNA through an amplitude equal to 28% of its prior close ($113.88–$131.50) within 30 pre-market minutes, while its consensus target price still sits at $48.92. The direction is most likely right — the first Phase III success for a personalized mRNA cancer vaccine means the probability of success for all 9 pipeline trials should be revised up; but the price has already awarded full marks to data that has not been released, and the market itself revised that full-marks score twice before the open.

Every other judgment is weaker than that one; please trust this brief in the following order: ① the retail restatements are the hardest — the EPS of TGT, TJX and LOW are all contaminated by IEEPA refunds, the three companies account for them differently, and the ex-refund ranking (TGT strong / TJX and LOW weak) is supported by primary disclosure. ② MRK's re-rating is better explained by multiples than by NPV (+7.3% ≈ 1.1 PE turns of discount repair), but the claim that "the 1:1 market-cap increment falsifies the patent-cliff narrative" has been downgraded by this brief itself to a suggestive observation and should not be used as a conclusion. ③ AI hardware direction is TBD — this one reversed on itself within the hour before the open, and neither MRVL +11.22% nor AVGO −4.74% has a same-day primary catalyst yet, so until the cause is established it is not a basis for trading either side.

---

Data-pull and process log (internal)

This brief went through one round of substantive revision after risk-auditor QC; the judgments overturned/downgraded are listed below for review purposes:

1. 🔴 ADI's release timing was written wrong, and one major conclusion was withdrawn with it. The first draft said "released after the close on 08-18," when it was actually 07:01 ET pre-market on 2026-08-19 (confirmed by the PRNewswire timestamp), with the call at 10:00 ET not yet held. On that basis the inference that "a big beat bought only +1%, showing the market will not price AI hardware fundamentals" is withdrawn — yesterday's ADI −3.50% happened before the print, so the two do not form a contrast. Lesson: attribution can only be done after the release timing has been aligned with the price-reaction window.
2. 🔴 "Zero bounce in AI hardware" was falsified by the brief's own data source 12 minutes after it was written down. The cross-section captured at 08:0x was uniformly negative; after the full re-pull at 08:30–08:34, FN/LITE/CRDO/MU/SNDK/STX/COHR had all turned positive and SMH went from −0.05% to +0.16%. Root cause: the first draft applied the discipline "pre-market volume and price are cumulative" only to MRNA, while using single points for the other 20-plus names, and those single points' update timestamps spanned 28 minutes (BURL frozen at 07:54, MU updated to 08:22) — asynchronous snapshots treated as a cross-section. All have been re-pulled and timestamps standardized.
3. 🟠 "1:1 market-cap increment ⇒ exclusivity extension falsified" was downgraded to a suggestive observation. Four flaws: the numerator contains a short squeeze (short interest 13.37%), which the brief itself had admitted; it attacks a straw man (nobody argues 100% is priced); the ratio itself drifts between 1.09 and 1.19 with no error band; and the prior actually points the other way (the incremental Keytruda volume from the combination accrues 100% to Merck, so >50% should be expected). Lesson: the most strongly worded sentence in the whole brief rested on its weakest reasoning — that is precisely the signal that cross-checking is needed.
4. 🟠 WYFI's conversion premium was computed off the wrong basis. Using the post-crash pre-market price as the denominator gave "63% above"; correctly measured against the pricing reference price (prior close $27.07) it is +25.0%, a normal range. The inference of "a tightening financing window for AI infrastructure" was withdrawn accordingly — a textbook case of reaching the conclusion first and then finding the data.
5. 🟠 The IMbrave050 analogy suppressed the counter-evidence. The 5-year update of KEYNOTE-942 (ASCO, 2026-06-01) still shows an HR of 0.51, meaning it precisely survived the long-follow-up test, the opposite of IMbrave050's failure mode. This has been added and the bear case narrowed to "population dilution" rather than "decay with follow-up."
6. 🟠 LOW's refund restatement was omitted, violating the brief's own stated methodology (it declared that "all retail EPS are stripped of refunds" and then missed one company). The $0.11 restatement and the 52-week position have been added.
7. 🟠 The internal-only wrapper was applied in the wrong direction: four pieces of "basis/coverage information the client should know" (the rating-coverage gap, MRNA's debt basis, the incomplete MRK attribution, the $200,000 pricing assumption not being primary) had been wrapped into the internal block by mistake. That is the opposite of failing open — hiding them makes clients overestimate reliability. All have been moved into the body. Reverse check: no 403/timeout/rate-limit narrative remains in the body; that side is clean.
8. 🟡 Several basis corrections: ADI's adjusted operating margin of 50.0% is FQ3 actual and 52.0% is FQ4 guidance (the first draft mixed them); TGT's "67%" was changed to "80% of the year-over-year increment"; three mutually contradictory statements of MRNA's pre-market volume/peak/fade have been reconciled.

Data-pull failures (unrelated to conclusions, recorded only):

• Three failures on ADI primary sources: SEC EDGAR 8-K exhibit 403, investor.analog.com 60-second timeout twice, stocktitan.net DNS timeout twice; the segment revenue breakdown was therefore not obtained.
• CNBC's Target earnings page returned 403.
• The stockanalysis.com screener page returned 429 on the second request, but the /api/quotes/s/<sym> quote API worked throughout.
• CNBC futures symbols @ES.1/@NQ.1/@YM.1/@RTY.1 all returned empty and were replaced with @SP.1/@ND.1/@DJ.1; Russell 2000 futures were not obtained and IWM pre-market was used instead.
• CNBC's ExtendedMktQuote_* fields were all empty in the pre-market, again confirming the memory that "CNBC extended-hours fields cannot be used pre-market", so the MRVL/AVGO moves could not be cross-checked against CNBC.

Sub-agent conflict handling (conservative basis taken in all cases): BNT122 interim analysis (one side read "DSMB recommended continuation," the other read "crossed the futility boundary") → only the part both agreed on was written, with no judgment on efficacy; MRNA debt basis ($591 million term loan vs $1.287 billion total debt) → the narrow basis was used, with a note added in the body; MRNA TTM net loss (−$3.15 billion vs −$1.47 billion) → citation avoided, replaced with the company's own cash guidance.

Still unresolved, must be handled immediately after the open:

• MRVL +11.22% / AVGO −4.74% have no same-day primary catalyst. The pattern closely resembles 2026-04-20 "Google in talks with Marvell on custom inference chips, Broadcom under pressure," but no 8/19 news source was verified. This is the brief's largest unresolved open question.
• Whether TJX's Q3 guidance itself contains additional IEEPA refunds (secondary accounts say yes, unconfirmed against the original); and whether the FY27 consensus is $5.19 or $5.22 — if it is $5.19, then "still below consensus after the raise" does not hold.

⚠️ Risk disclaimer: this list is a pre-market information review and watchlist only and does not constitute investment advice. U.S. equities are volatile and pre-market gap risk is high; after earnings there can be IV crush and guidance reversals. Automatically generated content may contain stale information or factual errors — company disclosures / SEC filings prevail, and this must not be used directly as a basis for trading.