A-Share Pre-Market Brief | 2026-08-11 Tuesday

Coverage window: 2026-08-10 (Monday) 15:00 close → 2026-08-11 07:00. All quotes and fund flows are on an August 10 close basis (the "8/11 limit-up price" figures in the text are derived from the 8/10 close under each board's price limit); the overnight overseas read is the U.S. August 10 trading session (read timestamp: 04:00 Beijing time on August 11, at the close). Financial data come from the original CNINFO/exchange filings. Data this brief failed to obtain (best effort made, disclosed honestly): the August 10 Dragon-Tiger List (published after 18:00, not retrieved this time), the trailing 5-day continuous fund-flow series for individual stocks (only the single 8/10 day was obtained), and August 11 real-time pre-market quotes. Therefore "two consecutive days of net inflow" in the text can only be a item to be verified today, not a verified conclusion.

---

0. Today in One Sentence

1. The strongest catalyst is 甘李药业 Gan & Lee (603087)'s licensing of bofanglutide across 39 European countries — a non-refundable upfront payment of EUR 62 million (≈ RMB 485 million on a revenue basis, at the Bank of China 8/10 EUR mid-rate of 781.71), equal to 12.0% of 2025 revenue of RMB 4.052 billion; the company set a precedent in 2024 by booking a similar licensing payment under principal operating revenue.
2. But the strongest branch is not pharma — it is semiconductor equipment, and the evidence is mainly money, not news. On 8/10 pharma led the limit-up count with 12 names, yet the whole chain saw main-force net outflow of about RMB 2.2 billion; the same day semiconductor equipment took in RMB 3.688 billion, 6.31% of turnover — both ranked first market-wide, and among the top 30 stocks by net inflow, equipment took 8 slots plus materials 1. The limit-ups are in pharma; the money is in equipment.
3. The direction of money is "a one-way move inside electronics," not a broad rally: electronics −RMB 20.383 billion, communications equipment −RMB 16.710 billion, digital chip design −RMB 8.570 billion, printed circuit boards −RMB 8.474 billion, while semiconductor equipment + materials drew in RMB 4.723 billion against the tide.
4. Driver types: capital reallocation (semiconductor equipment) > corporate licensing and orders (innovative drugs) > policy (coal / power grid "15th Five-Year Plan", PBoC incremental policy) > event speculation (ordnance restructuring).
5. Pre-market state: a divergence day, not a resonance day. The Shanghai Composite posted a fifth straight advance and approached 4000 (3966.59), but the ChiNext Index was −0.73% and STAR 50 −0.36%; sentiment has entered a "cooling phase," with 14 broken limit-up seals, the highest ladder rung only 5 consecutive limit-ups and a gap at 3; the Philadelphia Semiconductor Index closed overnight at 11,993.9, −2.94% (it had been +0.24% early — an intraday plunge), which will weigh on the tech open. Today's core task is not to find a new theme; it is to verify whether yesterday's RMB 3.688 billion was continuous.

---

1. News Overview

⚠️ Three "fake positives" and unit traps that must be named — don't be led astray by the media

1. "JPMorgan raised its memory forecast 48%" is a compression error introduced in reprinting. The original text reads "raised its 2026-2028 global DRAM and NAND market-size forecasts by 4%, 6% and 8% respectively" — some reprints compressed this into "48%". The real upgrade is a mild single-digit revision, not a reset of industry assumptions. On that basis this brief downgrades the item from S to A and changes the primary evidence for Branch 1 from "a large investment-bank upgrade" to "that day's fund flow." (The report's absolute level of US$1.442 trillion for 2027 is unaffected; the components DRAM US$982.9 billion + NAND US$458.7 billion sum consistently.)
2. 江波龙 Longsys's "private placement at a 45% premium" has the sign backwards. The placement price was RMB 560.00 and the close on the pricing base date (6/29) was RMB 723.10, so it was in fact issued at a 22.56% discount (a 23.65% premium to the RMB 452.90 floor price). The "45%" must come from 560 ÷ 386.60 − 1 = 44.85%, and its real meaning is that the 21 institutions that took part in the placement are currently down 26.34% (412.50/560−1) — information on the negative side. The placement raised RMB 3.700 billion, dilution is 1.54%, and the 6-month lock-up expires around January-February 2027.
3. 江波龙 Longsys's "planned RMB 400-800 million buyback" is not news. The controlling shareholder proposed it back on July 23; the amount is only 0.23%-0.45% of market cap, and the purpose is equity incentives / employee stock ownership, not cancellation — low substance.

---

2. Strongest Positive Branches, Descending

Three branch-level notes

1. Why semiconductor equipment ranks ahead of innovative drugs, whose news is harder: the innovative-drug news is genuinely the strongest (Gan & Lee is the only certain cash), but it is a single-stock positive, and at the branch level money is flowing out. For semiconductor equipment, money (first market-wide) and position (still negative over 20 days) point the same way. Branches are ranked by branch strength and stocks by stock strength; today the two do not agree — and that is the single most important thing to note today.
2. This first place is a "candidate," not a confirmation. The 8/10 recap set "two consecutive days of main-force net inflow" as the test for semiconductor equipment, and explicitly noted that 8/10 was the first sample. Today's closing money flow is the answer. Also, this brief did not obtain the day-by-day flow series for 8/6-8/9, so no prior can be given — that is this brief's biggest evidence gap.
3. Earnings delivery is not a plus: yesterday's recap used three samples to show that already-disclosed results have no explanatory power for the next day's move, and may even work in reverse — Cambricon's interim report +122.61% yet −6.33% that day, 广合科技 Guanghe Technology ex-non-recurring +96.71% yet −0.92% that day, 盛美上海 ACM Research (Shanghai) ex-non-recurring −15.31% yet +12.16% that day. On that basis this brief has lowered the scores of pure earnings-delivery names such as 江波龙 Longsys.

---

3. Overall Stock Catalyst-Strength Ranking (descending by stock)

Board price limits: Shanghai/Shenzhen main board ±10% / ChiNext ±20% / STAR Market ±20% / BSE ±30%. Read the "8/10 Chg" column below against each stock's own limit — +12% on ChiNext is not a limit-up. "In-window" = whether the catalyst falls inside the 8/10 15:00 → 8/11 07:00 window. News outside the window was already traded once yesterday. All PE figures are on a TTM basis (through the latest disclosed period). For companies whose earnings are rising rapidly, trailing PE materially overstates the valuation level; this is flagged case by case in the text.

Table note: "Not in the top 30" means the stock did not enter the top 30 by main-force net inflow on 8/10; it does not mean net outflow. This brief only obtained the top 30 plus fund-flow data for a few spot-checked stocks, not a full stock-by-stock pull. The August 10 Dragon-Tiger List (published after 18:00) was not obtained this time, so every fund-flow conclusion in the text lacks cross-validation against exchange seat data.

---

4. Stock Scoring Model (100 points)

Component Breakdown for the Top Seven

Deduction detail (spelled out name by name; otherwise the scores are meaningless)

• 甘李药业 Gan & Lee −11: the selling window for seven directors and senior managers opens 8/31 (−4); its biological-products sector saw net outflow of RMB 455 million and the stock itself had main-force net outflow of RMB 54 million that day (−3); 2026Q1 net profit attributable to parent −38.29% (−2); the milestones, which are 91.5% of the total deal value, are pure options and the product has zero approvals and zero NDAs globally (−2).
• 北方华创 NAURA −5: RMB 551.5 billion market cap limits elasticity (−2); no named order, only an industry trend (−2); overnight SOX −2.94% (−1).
• 药明康德 WuXi AppTec −8: the catalyst is outside the window and the stock already rose +4.21% on 8/10 (−4); only −3.2% from the range high (−2); main-force net outflow of RMB 139 million that day, i.e. retail sentiment absorbing supply (−2).
• 拓荆科技 Piotech −7: already +23.0% over 5 days (−3); PE (TTM) 128.70 (−2); overnight SOX weakness (−2).
• 中微公司 AMEC −11: already +25.9% over 5 days (−4); PE (TTM) 136.89 (−3); overnight SOX weakness (−2); "two consecutive days of net inflow" not yet confirmed (−2).
• 雅克科技 Yoke Technology −6: already +28.4% over 5 days (−3); 11.04% turnover is on the high side (−2); the materials end is more indirect than the equipment end (−1).
• 江波龙 Longsys −12: the results merely deliver on the 7/3 guidance (−4); first disclosure of operating cash flow of −RMB 3.151 billion (−4); inventory of RMB 25.777 billion is 60.12% of total assets while the write-down provision is only 0.38% (−3); the sole survivor in its branch (−1).

---

5. Detailed Analysis of the Top 10 Stocks

1) 甘李药业 Gan & Lee60308770 pts | priority deep-dive · Shanghai main board ±10%

Related news: announcement No. 2026-062 (released the evening of 8/10, dated 8/11). The counterparty is A.Menarini International Licensing S.A. (annual revenue EUR 4.88 billion, present in 140 countries); the licensed territory is "the 27 EU member states plus the United Kingdom, Switzerland, Norway, Iceland, Liechtenstein, the Balkan region and others, 39 European countries and regions in total"; "Menarini will pay a non-refundable upfront payment of EUR 62 million"; milestones "up to a cumulative EUR 664 million"; after commercialization, "tiered royalties at up to a high double-digit rate" on annual net sales. The company states it "expects a positive impact on the company's operating results for 2026 and subsequent years," while cautioning that "the amount Menarini ultimately pays remains uncertain." Announcement PDF¹

The positive, strictly tiered

Component	Amount	Nature	Assessment
Upfront payment of EUR 62 million	≈ RMB 485 million (Bank of China 8/10 EUR mid-rate 781.71)	Non-refundable, already signed	Direct cash positive
Milestones up to EUR 664 million	≈ RMB 5.191 billion	Requires R&D/approval/sales milestones	A far-dated option; cannot be discounted into a near-term valuation
Tiered royalties	Rate not disclosed	Requires the product to be approved and sold in Europe	An even more distant option

Table scrolls horizontally

Magnitude (the basis must be stated clearly, or it will be systematically overstated)

• Revenue basis (directly comparable): RMB 485 million = 12.0% of 2025 revenue of RMB 4.052 billion. This is the only ratio that requires no assumptions.
• After-tax basis (requires assumptions; an estimate): the company's 2025 income tax / pre-tax profit was only about 7.01% (high-tech enterprise status plus super-deduction of R&D expenses). If the entire upfront payment hits the books and bears the same tax burden, the after-tax contribution to net profit would be about RMB 450 million, roughly 39% of 2025 net profit attributable to parent of RMB 1.144 billion and 58% of 2025 ex-non-recurring profit of RMB 780 million.
• ⚠️ Caution: EU withholding tax and treaty rates are not disclosed and were not obtained, so the after-tax estimate above may be optimistic. Do not simply divide "RMB 485 million" by net profit attributable to parent and claim "42% of profit" — that compares a revenue-basis figure with an after-tax-basis figure.

There is an accounting precedent (fact and inference, kept separate)

• Fact: the 2025 annual report's "principal business by product" separately lists "international — royalty service revenue" of RMB 4.7535 million, −96.49% y/y, implying about RMB 135.4 million for that line in 2024 at an 89% gross margin; the annual report's own note reads, "mainly because in 2024 the company recognized milestone revenue as agreed under the relevant contract, and there was no such revenue in the reporting period."
• Inference (not company disclosure): historically the company has booked licensing/milestone payments as principal operating revenue in the period the milestone is met, so this upfront payment is more likely than not to go into operating revenue as well, and hence into the ex-non-recurring line. But the timing of revenue recognition also depends on progress in fulfilling performance obligations such as technology transfer; the announcement itself only promises a positive impact on "2026 and subsequent years" and does not promise full recognition within the year.

Timing note: the company's 2026 interim report is scheduled for August 26, but the agreement was signed after the balance-sheet date (6/30), so this money will not appear in the interim report; the company issued no H1 guidance, and 2026Q1 net profit attributable to parent was already −38.29%, so the August 26 interim report will most likely still show a year-on-year decline. The pace at which the positive is realized is out of sync with the nearest earnings disclosure point.

One motive clue worth noting (inference, not company disclosure): the company's 2026 restricted-stock incentive plan (2026-06-11 draft) sets a company-level target of no less than RMB 1.43 billion of net profit in 2026. Actual 2026Q1 was RMB 192.5 million, so Q2-Q4 must deliver RMB 1.238 billion, +48.8% versus RMB 832 million in the same period of 2025 — very hard on the insulin core business alone (Q1 revenue was already shrinking). The licensing payment, estimated at about RMB 450 million after tax, could cover most of that gap but not close it alone; the core business must still contribute growth. This is reasoning, not a promise; what it flags is that the company has a subjective incentive to recognize a substantial part of the upfront payment in 2026.

Stage of the industry branch: innovative-drug overseas licensing is in the late fermentation / climax phase. On 8/10 pharma led with 12 limit-ups, with 百花医药 Baihua Pharmaceutical (600721) at 5 consecutive limit-ups and 开开实业 Kaikai Industry (600272) at 3 — the ladder is complete but already broken. Branch-level money has diverged (see risks).

Fundamental verification

• Business match: a complete match. The leading domestic insulin maker (second in market share, behind only Novo Nordisk); the bi-weekly GLP-1 formulation is a core in-house pipeline asset, not a cross-sector concept.
• Financials: 2025 revenue RMB 4.052 billion (+33.06%), net profit attributable to parent RMB 1.144 billion (+86.05%), ex-non-recurring RMB 780 million (+81.21%), gross margin 75.84%; 2026Q1 revenue RMB 878 million (−10.84%), net profit attributable to parent RMB 192 million (−38.29%), but ex-non-recurring only −13.31% — the apparent decline mostly reflects the high base from non-recurring gains such as litigation enforcement proceeds in 2025Q1, and the deterioration in the core business is far smaller than it appears. Gearing is 6.72% with no interest-bearing debt, and cash-like assets are RMB 3.416 billion. 2025 operating cash flow was RMB 784 million (OCF/net profit 0.69). Inventory turnover is 422 days (a biologics characteristic).
• R&D substantiation (hard evidence): GZR18 received a China IND in 2021-10; first patient dosed in the China Phase III in 2024-12; first patient in the U.S. Phase II in 2025-03; an OSA-indication Phase III started in 2026-04. On 2026-05-29 the pivotal China weight-loss Phase III GRADUAL-1 met its primary endpoint: 640 subjects, 52 weeks, weight loss of 15.12% in the 24mg arm, 18.54% in the 48mg arm and 1.11% on placebo, and the U.S. pivotal Phase II met its endpoint at the same time; the Phase IIb head-to-head data were published in the Annals of Internal Medicine in June 2026, with HbA1c reductions of 1.87%-2.28% versus 1.60% for once-weekly semaglutide. A BD deal backed by a top-journal head-to-head plus a successful Phase III is two orders of magnitude away from "a vague reply on an investor-interaction platform."
• One sentence that must be downgraded: "the world's first bi-weekly GLP-1RA" is a forward-looking statement, not an accomplished fact. The announcement itself uses "is expected to become" and "potential." The product is not approved in any country worldwide, and no NDA acceptance announcement has been seen. The longest dosing interval among approved GLP-1RAs globally is still once weekly; among longer-interval candidates, the Phase III readout for Amgen's MariTide (monthly) points to early 2027. This is a race, not a settled fact.
• This is the 4th out-licensing deal, but the scale is an order-of-magnitude jump: 2025-11 Latin America PC (upfront undisclosed, +0.26% that day); 2025-12 India's Lupin (undisclosed, −3.13% that day); 2026-04 Korea's JW (US$5 million upfront, US$81.1 million total, +2.22% that day). This upfront is about 14 times the Korean deal and the potential total about 10 times, and it enters the high-regulation, high-payer EU market for the first time. The single-day share reaction to all three earlier deals was within ±3%, showing that the market had been pricing this BD line as "small and regional" — the scale has not been fully anticipated.
• One calibration point: the announcement states that the contract "falls within the approval authority of the general manager … and does not require submission to the board of directors or the shareholders' meeting" (fact). What needs saying is this: a potential total of EUR 726 million ≈ RMB 5.68 billion far exceeds the common 10%-of-net-assets threshold yet still did not go to the board, which suggests the review threshold is most likely applied to the upfront/committed amount rather than the potential total. This brief did not obtain the specific authorization language in the articles of association, so it does not infer from this that "the scale is not material" — it is listed only as a fact to be verified.
• Centralized-procurement status (on the positive side): the follow-on insulin procurement cycle runs through 2027-12-31, so there is no new national price-cut round for the next 16 months; 2025 domestic insulin formulation revenue was RMB 3.430 billion (+40.72%) on volume of 97.63 million units (+30.58%), having moved from the 2022 price-cut shock into a volume-for-price delivery phase. Separately, in 2026-05 insulin aspart and insulin lispro actually received EU Commission marketing authorization (an approval, not an intention).
• Current GLP-1 revenue share ≈ 0: the company has no marketed GLP-1 product today; 100% of product revenue comes from insulin and related biologics. This is "licensing revenue," not "product revenue."

⚠️ An offsetting fact that must be disclosed: an after-close announcement on 2026-08-06 (No. 2026-060) states that seven directors/executives (including chairman 陈伟 Chen Wei, board secretary 邹蓉 Zou Rong and five vice presidents) each plan to sell no more than 120,000 shares, 840,000 shares in total (0.14% of share capital); all shares came from equity incentives, the stated reason in each case is "repayment of personal debt," and the selling window runs 2026-08-31 to 2026-11-30. The timeline is objective: 8/6 collective sale announcement → the evening of 8/10 a major licensing positive → 8/31 the selling window opens. This report makes no inference about motives, but the information overlaps closely with the period in which the positive is fermenting, and anyone tracking the stock must know it. (Note: Chen Wei is chairman but not the controlling shareholder or actual controller; a 0.14% sale ratio is not large in itself.)

Technicals, sentiment and money: Shanghai main board ±10%, the 8/11 limit-up price is about RMB 74.47 (67.70×1.1). It closed 8/10 at RMB 67.70, −2.86%, on 2.26% turnover, a volume ratio of 1.35 and RMB 872 million of turnover, with main-force net outflow of RMB 54 million (−6.22% of turnover). −0.3% over the past 5 days, +3.7% over 20 days, −13.3% from the 52-week high (78.12,2025-09-25) and +30.1% above the 52-week low (52.05,2026-05-28). Average turnover over the past 20 days was RMB 874 million versus RMB 638 million over 60 days, volume already up 37% — and the rebound began right after the Phase III readout on 5/29, showing the pipeline value has been partly priced in. No limit-up, no gap-locked board, no one-day-wonder pattern. The announcement came after the close, the stock actually fell that day, and there is no sign of front-running.

Valuation: PE (TTM) 39.53×, PB 3.42×, market cap RMB 40.480 billion. PE historical percentile: 37.8% over the full sample, 21.3% over the past 3 years; PB historical percentile: 63.8% over the full sample, 86.4% over the past 3 years.

The low PE plus high PB here cannot be read straight off the "cycle top" template — Gan & Lee's profits collapsed under centralized procurement in 2022-2023, leaving PE persistently distorted or even negative and PB compressed to 1.8×; the 86% profit recovery in 2024-2025 pulled PE down to a low percentile while PB recovered along with the share price to the 86% percentile of the past 3 years. The real risk is that the low PE is a product of high trailing profits. On an ex-non-recurring TTM basis, PE is 53.9×, which is closer to reality.

Final judgment: hard news, meaningful scale, a low position, a hot theme, and the only S-tier catalyst on the whole list that falls inside the pre-market window — all four boxes ticked, hence the S tier. But it is not flawless: the executive selling window opens 8/31, both branch and stock money flowed out, Q1 fundamentals are a headwind, and the milestones are pure options. Those four make up the −11 deduction and should not be drowned out by the EUR 664 million headline number.

2) 北方华创 NAURA00237164 pts | priority deep-dive · Shenzhen main board ±10%

• Related news: JPMorgan's memory super-cycle report (the cycle running through end-2028); KLA and Lam Research have already raised their 2026 global wafer-fab equipment (WFE) spending expectations above US$150 billion in their latest results. ⚠️ For the latter, this brief only obtained a Chinese-language paraphrase and did not obtain the primary source or publication date — please discount accordingly when citing.
• The positive: an industry-trend positive, not a direct order. The platform-type leader in domestic semiconductor equipment with the broadest product coverage, and the most certain beneficiary of a WFE upgrade; there is no named order announcement, so under the tiering it can only count as an industry trend.
• Technicals and sentiment (the core advantage): it closed 8/10 at RMB 760.00, up only +1.07%, on 1.20% turnover and a volume ratio of 0.72. +18.5% over 5 days, −0.6% over 20 days (the most resilient in the whole equipment group), −21.4% from the range high. Against a backdrop of the entire sector rising 18%-41% over five days, it rose the least and drew down the least — the classic "leader not yet stretched" pattern.
• Money: main-force net inflow of +RMB 319 million (17th market-wide). PE (TTM) 98.91, PB 14.24, market cap RMB 551.5 billion.
• Final judgment: if the semiconductor-equipment main line is real, the leader has to keep up — that is both its opportunity and the branch's litmus test. A RMB 551.5 billion market cap caps its elasticity, making it better suited as a branch-confirmation signal than as a first choice for elasticity.

3) 药明康德 WuXi AppTec60325963 pts | watch closely · Shanghai main board ±10%

• Related news (outside the window): on the evening of 8/9 the U.S. District Court for the District of Columbia granted the preliminary injunction motion; H1 revenue RMB 28.897 billion (+38.93%), net profit attributable to parent RMB 11.08 billion (+29.43%), full-year revenue guidance raised to RMB 58.5-60.5 billion (+35~39%). East Money evening wrap⁵
• The legal matter needs a qualifier: the ruling is a preliminary injunction — interim relief during litigation; the 1260H designation has not been vacated, and the appeal process still lies ahead. It cannot be described as "the case has been won."
• Fundamentals (two disclosure gaps that must be stated): PE (TTM) 22.22 against +35~39% full-year guidance is the best growth-valuation match in the pharma main line. But this brief did not obtain its H1 ex-non-recurring profit or the share of non-recurring items — an H1 net margin of RMB 11.08 billion / RMB 28.897 billion = 38.3% is markedly above the CXO norm, and a large investment gain cannot be ruled out. This brief ran a profit-quality check on 兆易创新 GigaDevice ("non-recurring items are 29.7% of net profit attributable to parent") but could not run the same check on WuXi AppTec; the standard is therefore inconsistent, and readers should discount accordingly.
• Why not "priority deep-dive": ① the catalyst is outside the window and was already realized on 8/10 with +4.21%; ② it is only −3.2% from the range high, +25.6% over 5 days and +58.9% over 60 days — a high position; ③ turnover on 8/10 was huge yet main-force money flowed out by RMB 139 million, i.e. retail sentiment absorbing supply rather than institutions adding.
• Today's verification point: whether main-force money flips from net outflow to net inflow.

4) 拓荆科技 Piotech68807260 pts | watch closely · STAR Market ±20%

Thin-film deposition equipment, a high-value link in memory-fab expansion. It closed 8/10 at RMB 728.00, +5.41%, with main-force net inflow of +RMB 306 million. +23.0% over 5 days, −13.8% over 20 days, −23.0% from the range high. PE (TTM) 128.70, PB 18.03. Judgment: it moves with AMEC, sits lower on a 20-day basis and offers more elasticity, but with less certainty than AMEC or NAURA. Watch closely.

5) 中微公司 AMEC68801260 pts | priority deep-dive · STAR Market ±20%

• Money verification (the entire reason for this entry): main-force net inflow of +RMB 891 million on 8/10, 2nd market-wide; its semiconductor-equipment industry took in +RMB 3.688 billion, 6.31% of turnover — first market-wide on both counts; over the same period electronics was −RMB 20.383 billion and digital chip design −RMB 8.570 billion. This is not a broad rally; it is a targeted move of money.
• Branch stage: launch / first confirmation phase. Yesterday's post-close recap already called it "the cleanest new direction" and set "two consecutive days of main-force net inflow" as the test — 8/10 was the first sample, and today the test comes due.
• Technicals and sentiment: it closed 8/10 at RMB 389.99, +6.08% (no limit-up), on 3.05% turnover and a volume ratio of 0.92. +25.9% over 5 days, −4.6% over 20 days, −22.0% from the range high (RMB 500). The key reading: a big 5-day bounce while 20 days remain negative — a repair after a deep drawdown, not late-trend acceleration. PE (TTM) 136.89, PB 13.55, market cap RMB 373.5 billion.
• Why it scores below NAURA: it is already +25.9% over 5 days, so the expectation gap has been consumed; its PE is higher than NAURA's. Its advantage is the clarity of the money signal (a full 10 on trading traits); its disadvantage is position.
• Final judgment: the most trackable name at the branch level today, with a single, explicit verification point: whether main-force money posts a second consecutive day of net inflow today.

6) 雅克科技 Yoke Technology00240956 pts | watch closely · Shenzhen main board ±10%

Electronic specialty gases and precursor materials, direct support for memory-fab expansion. It closed 8/10 at RMB 155.24, +4.34%, with main-force net inflow of +RMB 343 million (13th market-wide), while its semiconductor-materials industry took in +RMB 1.035 billion. +28.4% over 5 days, −17.9% over 20 days, and −37.0% from the range high, the lowest position in the equipment/materials chain. PE (TTM) 73.39 is the lowest in the branch. Judgment: the materials-end extension of the semiconductor-equipment main line, with both valuation and position low. Watch closely.

7) 江波龙 Longsys30130856 pts | watch only · ChiNext ±20%

• Related news: the interim report released after the 8/10 close — revenue RMB 24.088 billion (+136.26%), net profit attributable to parent RMB 10.577 billion (+71528.66%), ex-non-recurring RMB 10.047 billion, Q2 alone RMB 6.715 billion (+73.9% q/q), gross margin 58.90% (12.96% a year earlier), ROE 79.03%.
• Why only "watch only" — four independent reasons:
  1. This is old news being realized, not a brand-new positive. The company had already issued H1 guidance on 2026-07-03: net profit attributable to parent RMB 9.20-11.00 billion, revenue RMB 22-25 billion, ex-non-recurring RMB 9.0-10.5 billion. The actuals landed at the 76.5% / 69.6% / 69.8% positions of those ranges respectively, and none of the three exceeded the upper bound. The market has held the information "H1 earnings of RMB 9.2-11.0 billion" for more than five weeks. The all-time high of 749.88 came on July 1, i.e. before the guidance was disclosed; from the post-guidance high of 681.80 (7/6) to 412.50, it has already fallen 39.5%.
  2. The same batch of announcements disclosed two substantive negatives for the first time. Net operating cash flow was −RMB 3.151 billion (+RMB 693 million a year earlier), so OCF / net profit attributable to parent = −0.298; inventory of RMB 25.777 billion is 60.12% of total assets (of which raw materials are RMB 14.370 billion), while the write-down provision is only RMB 97.4 million, a 0.38% provisioning ratio. And this is not single-period noise: operating cash flow has been negative for four consecutive periods from 2023 to 2026H1 (−RMB 2.798 / 1.190 / 1.201 / 3.151 billion), so over 3.5 years cumulative reported profit of +RMB 11.671 billion compares with cumulative operating cash flow of −RMB 8.341 billion, a gap of RMB 20.012 billion.
  3. Almost all of the profit comes from price spread, not volume. Cost of sales rose only +11.55% while revenue rose +136.26%, implying volume growth of about +11.5% and unit-price growth of about +111.8% (this is an estimate, premised on broadly flat unit costs). The company confirms the mechanism in its own risk disclosures: "changes in product unit cost lag changes in product selling price, so memory makers' gross margins fluctuate with wafer prices." The 58.90% gross margin and the RMB 25.8 billion of thinly provisioned inventory are two sides of the same coin.
  4. The sole survivor in its branch: on the same day 长鑫科技 CXMT (688825) fell −2.25% with net outflow of RMB 3.266 billion, and 兆易创新 GigaDevice (603986) fell −3.36% with net outflow of RMB 3.057 billion. Longsys's +RMB 922 million of net inflow is an isolated event within a branch bleeding heavily overall. And on 8/10 it rose +6.70% with no public catalyst at all, outperforming peers by 8-10 percentage points — most likely the interim report being priced in early.
• The valuation basis must be stated clearly: the PB of 14.48 shown by quote vendors uses Q1 net assets; the interim report lifted net assets attributable to parent from RMB 12.05 billion to RMB 18.583 billion, so that PB will mechanically reset to about 9.5× today. On true H1 net assets PB is 9.54×, or 7.97× pro forma including the placement. PE (TTM) is 14.79× (on post-issuance share count).

  The cycle-top signature still holds, and none of the three bases dissolves it: the PE historical percentile is 0.0%-6.8% while the PB historical percentile is 77.3%-84.7% — a 70-85 percentage-point divergence. Net assets more than doubled in half a year and absolute PB fell from 14.48 to 7.97, yet the PB percentile only fell from 84.7% to 77.3% — a percentile that a surging denominator cannot pull down shows the high valuation is not an artifact of accounting lag. Cross-check: 佰维存储 Biwin Storage has a PE percentile of 3.0% and a PB percentile of 68.2% (the same signature); 兆易创新 GigaDevice is at 55.6% / 61.0% (the two agree). The divergence appears only at module makers, which is precisely the evidence that "module makers' PE cannot be compared historically." (Methodological limitation: percentiles are taken from the Baidu valuation series, whose sample starts at each company's listing date; Longsys has n=1466, only about 4 years, and PE was negative on 46% of trading days, so the PE percentile is a percentile "within positive-value samples only.")

• Technicals and sentiment: ChiNext ±20%, the 8/11 limit-up price is RMB 495.00 (412.50×1.2). It closed 8/10 at 412.50, +6.70%, on 10.42% turnover, with RMB 11.851 billion of turnover, the largest in 12 sessions. −45.0% from the all-time high (749.88, 2026-07-01); −21.0% over 20 days, −30.0% over 60 days, +43.7% over 120 days.
• Final judgment: fundamental numbers grade A, information freshness grade D, cash quality grade D, trading position grade C. If it cannot hold a gap-up on expanding volume today, this is the classic "good news, distribution" pattern.

8) 精测电子 Jingce Electronic30056752 pts | watch closely · ChiNext ±20%

Metrology and inspection equipment. It closed 8/10 at RMB 235.66, +5.25%, with main-force net inflow of +RMB 311 million. +41.1% over 5 days, the sharpest bounce in the whole equipment group; −11.6% over 20 days, −29.7% from the range high. PE (TTM) 756.58 — the company's earnings are in a rapid recovery phase, so trailing PE lags badly and should not be used as a valuation yardstick. PB 16.00. Judgment: the most elastic and also the most dangerous; a 5-day slope of +41% is steeper than AMEC's or NAURA's. Watch closely, not a first choice.

9) 盛美上海 ACM Research (Shanghai) (688082, STAR Market ±20%) | 52 pts | watch closely

Cleaning equipment. On 8/10 it was +12.16%, the brightest gain in the market, with main-force net inflow of +RMB 303 million. +30.7% over 5 days, −18.9% over 20 days, −29.7% from the range high. PE (TTM) 92.29. An honest disclosure: yesterday's pre-market list passed on it because of "ex-non-recurring −15.31%," and that day it rose +12.16%, the best in the market — that Pass criterion was falsified. Yesterday's recap concluded that "financial quality is not a short-term ranker," while also warning that "simply inverting it and ignoring the ex-non-recurring line is equally dangerous." Judgment: it already exploded yesterday, so today it sits in chase-the-high territory. Watch closely.

10) 长川科技 Changchuan Technology30060450 pts | watch closely · ChiNext ±20%

Test equipment. It closed 8/10 at RMB 288.08, +1.72%, with main-force net inflow of +RMB 316 million. +26.7% over 5 days, −11.4% over 20 days, +128.8% over 120 days, −27.1% from the range high. PE (TTM) 118.24, PB 22.90. Judgment: a back-row name in the equipment chain — the money is there but the gain lags, so it is a follower. Watch closely.

---

6. Pass List

---

7. Intra-Branch Rankings

7.1 Semiconductor Equipment / Materials (Branch 1)

This branch holds 9 of the top 30 by main-force net inflow on 8/10: 8 in equipment (AMEC #2, Jingzhida #7, NAURA #17, Changchuan #18, Jingce #19, Piotech #20, ACM Shanghai #21, 臻宝科技 Zhenbao Technology #27) plus 1 in materials (Yoke #13). Skyverse did not make the top 30.

• The hardest stock: 北方华创 NAURA (002371) — leadership plus not yet stretched.
• Clearest trading signal: 中微公司 AMEC (688012) — 2nd market-wide by net inflow, +6.08% without a limit-up; the direction is the clearest.
• Who is just following: 长川科技 Changchuan Technology, 精智达 Jingzhida and 中科飞测 Skyverse are back-row followers.
• Who to Pass: 中科飞测 Skyverse (688361) — almost no earnings support and no inflow; it only has beta when the branch is strong.

7.2 Innovative-Drug Overseas Licensing (Branch 2)

• The hardest stock: 甘李药业 Gan & Lee (603087). Clearest trading signal: Gan & Lee again.
• Who is just following: 凯莱英 Asymchem and 毕得医药 Bide Pharmatech. Who to Pass: 百花医药 Baihua Pharmaceutical.
• Branch-level warning: biological products −RMB 455 million, chemical pharmaceuticals −RMB 1.514 billion, traditional Chinese medicine −RMB 585 million, medical services only +RMB 252 million (0.36% of turnover), pharmaceutical distribution +RMB 58 million — a combined whole-chain net outflow of about RMB 2.2 billion. The right way to use this branch is "pick the stock, ignore the branch."

7.3 Memory (Branch 4)

The right way to read this branch (the single most important supply-chain mapping conclusion in this brief): the memory up-cycle is real (CXMT's H1 guidance of RMB 50.0-57.0 billion of net profit attributable to parent proves it), but its best expression in today's A-share market is not module makers or fabs — it is equipment and materials. All three module makers fully pre-announced H1 results in the first half of July, their share prices peaked together in late June and early July, and they then drew down 45%-53% — the module makers' profits are already in the financial statements; the equipment makers' orders are still ahead. That CXMT has extremely strong fundamentals yet saw RMB 3.266 billion of net outflow shows exactly this: at this level, the market is not trading "who earns the most," it is trading "whose expectations are not yet priced in."

---

8. Opening Verification Signals for Today

8.1 Call-Auction Signals

• 甘李药业 Gan & Lee (603087): Shanghai main board ±10%, limit-up price about RMB 74.47. A 3%-6% gap-up is the reasonable pricing range for the news; a gap-locked limit or a gap-up of 8% or more would mean the market is capitalizing the EUR 664 million of milestones in one go, and none of that has any approval behind it yet; a gap-up that fades below the 5-day line would mean money leaving the pharma branch has overwhelmed the stock-level positive.
• The semiconductor-equipment group: with SOX −2.94% overnight, a lower open is the base case. Whether it can turn green after opening lower is the key — open lower and rally = the money is real; open lower and chop = yesterday's RMB 3.688 billion was a one-day wonder.
• 江波龙 Longsys (301308): ChiNext ±20%, limit-up price RMB 495.00. The gap between the size of the gap-up and the subsequent absorption is the core point to watch.

8.2 Sector Signals

• Semiconductor equipment (top priority): whether main-force money posts a second consecutive day of net inflow today — the test set out in yesterday's post-close recap, coming due today. Also watch whether the equipment names produce their first limit-up (there was not a single one in the whole group on 8/10), and whether NAURA (002371) can keep up as the leader.
• Innovative drugs: watch whether the 20cm names (百普赛斯 ACROBiosystems 301080, 毕得医药 Bide Pharmatech 688073) can gap-lock or seal quickly — ACROBiosystems broke its seal 30 times intraday yesterday, and repeated breaks again today would be a topping signal. Also watch whether pharma-chain money can turn from negative to positive, and whether 恒瑞医药 Hengrui Pharmaceuticals (600276) — the leader that has not participated at all and still saw RMB 485 million of net outflow that day — plays catch-up.
• Ordnance: whether 长城军工 Great Wall Military Industry (601606) can post a 2nd consecutive limit-up, and whether 建设工业 Jianshe Industry (002265) can keep up to form a ladder.
• Precious metals: 紫金矿业 Zijin Mining (601899) is the only yardstick — a catch-up on expanding volume = a second wave for the branch; if it stays flat while back-row names such as 招金黄金 Zhaojin Gold (000506) go limit-up, that is the tail end of the catch-up.

8.3 Stock-Level Signals

• Front-row absorption: if Gan & Lee seals the limit, watch the size of the sealing order and the number of broken seals.
• Fast seals that hold: the 20cm innovative-drug names.
• Large-order net inflow: 中微公司 AMEC (688012), 北方华创 NAURA (002371), 雅克科技 Yoke Technology (002409).
• Displaced by a stronger name: if a new equipment limit-up leader emerges today, AMEC's status as "core heavyweight" needs reassessing.

8.4 Risk Signals (what needs particular caution today)

1. The Philadelphia Semiconductor Index closed overnight at 11,993.9, −2.94% (prior day 12,356.8), and it had been +0.24% early — an intraday plunge — the pattern matters more than the magnitude. Intel −4.06% (a US$15 billion offering), NVIDIA −2.86%, Micron −1.89%, ARM down more than 4%. Pressure on the entire A-share semiconductor chain in early trading is the base case.
2. Four clarification/negative announcements landed directly on yesterday's limit-up names: clarifications from 先导基电 Xiandao Jidian (600641) and 海鸥股份 Seagull (603269), a profit warning from 南模生物 Shanghai Model Organisms (688265), and a verification halt for 爱丽家居 Aili Home (603221). Yesterday's limit-up ladder will most likely shrink today, and both the seal rate and sentiment will step down a level.
3. Halted today: 爱丽家居 Aili Home (603221), 建设机械 Jianshe Machinery (600984).
4. Sentiment is already cooling: 99 names on the consecutive-limit-up basis, an 87.6% seal rate, 14 broken seals; the top rung is only 5 consecutive limit-ups and there is only 1 name at 3 — the ladder is broken and the structure is unhealthy.
5. Index divergence: the Shanghai Composite posted a fifth straight advance toward 4000 (3966.59), but the ChiNext Index was −0.73% and STAR 50 −0.36%. A strong index with weak growth names means rotation from high to low, not a broad rally. Combined turnover across the two exchanges was RMB 2,523.103 billion, down RMB 141.317 billion.
6. The selling window for seven 甘李药业 Gan & Lee directors and senior managers opens on August 31 — two weeks away, but anyone tracking the stock must factor it in.
7. This week's macro calendar: U.S. July CPI is released at 20:30 Beijing time on August 12 (consensus 3.4% y/y, core 2.5%); Applied Materials (AMAT) reports on August 13, a two-way risk for the semiconductor-equipment main line. Today is a wait-and-see window ahead of the data.
8. This brief's own evidence gaps: the 8/10 Dragon-Tiger List was not obtained, the day-by-day 8/6-8/9 fund-flow series for semiconductor equipment was not obtained, and Gan & Lee's 8/11 real-time pre-market quote was not obtained. For the core test of "two consecutive days of net inflow," this brief cannot supply the prior for the first half.
9. Yesterday's pre-market list had a hit rate of 2/12, underperforming random stock selection on a day when 73.4% of stocks rose. This brief's credibility should be discounted accordingly; please rely on your own judgment.

---

9. Final Recommendations

9.1 The 5 Stocks Most Worth Watching Today

The following are objects of observation, not buy recommendations, and carry no implication of position size, entry/exit points or target prices.

9.2 The 3 Strongest Branches Today

9.3 Directions Not Worth Chasing Today

1. Optical modules / CPO / communications equipment — 中际旭创 Innolight −RMB 7.347 billion (the largest market-wide, 2.2 times the second), 新易盛 Eoptolink −RMB 3.309 billion, 光迅科技 Accelink −RMB 2.276 billion, 天孚通信 T&S Communications −RMB 1.123 billion; the communications-equipment industry −RMB 16.710 billion. Outflows of this magnitude usually do not end in one day.
2. PCB — but today's evidence points both ways, so weigh it yourself: on the money side, the printed-circuit-board industry saw net outflow of RMB 8.474 billion on 8/10, 东山精密 Dongshan Precision −RMB 3.047 billion, 沪电股份 WUS Printed Circuit −RMB 1.075 billion, 生益科技 Shengyi Technology −RMB 1.057 billion; yet on 8/11 the four securities newspapers simultaneously ran front-page stories that "PCB upstream materials keep rising in price and the supply chain is under strain." Money outflow and upstream price increases point in opposite directions, and this brief draws no one-way conclusion.
3. Memory modules and proxy stocks (the whole chain except Longsys) — 长鑫科技 CXMT −RMB 3.266 billion, 兆易创新 GigaDevice −RMB 3.057 billion. The memory cycle's positives are expressed more fully at the equipment end.
4. Yesterday's limit-up names that issued clarifications/negatives last night — 先导基电 Xiandao Jidian (600641), 海鸥股份 Seagull (603269), 南模生物 Shanghai Model Organisms (688265). A company clarification is the most reliable Pass criterion, as verified in yesterday's recap. ⚠️ But note: on 8/11 the four securities newspapers ran front-page coverage that "indium phosphide is being driven by AI optical-module demand, with 9 concept stocks drawing institutional attention," which points the opposite way from Xiandao Jidian's clarification — what was clarified is that this particular company is not in that business, not that the theme does not hold.
5. The grid-equipment limit-up ladder — 风范股份 Fengfan Power Equipment and 长城电工 Great Wall Electrical are both loss-making, and 三变科技 Sanbian Sci-Tech trades at PE (TTM) 320.58; meanwhile the core heavyweights 中国西电 China XD Electric at 0.00% and 平高电气 Pinggao Electric at −0.88% are not following at all. Historically, themes whose core heavyweights do not follow have poor persistence in their limit-ups.
6. Tungsten hexafluoride / specialty gases — 中船特气 CSSC Special Gas +654.2% over 120 days and 和远气体 Heyuan Gas at 24.11% turnover are at extreme positions; the "permanent shutdown in Japan" narrative has been publicly disproven.
7. The 5-rung high names — 宝鼎科技 Baoding Technology (002552) and 百花医药 Baihua Pharmaceutical (600721); the ladder is broken at 3 rungs with no relay support.
8. Digital chip design — industry net outflow of RMB 8.570 billion, and Cambricon (688256) fell −6.33% on 8/10.

9.4 The Final One-Sentence Judgment

Today's hardest news (Gan & Lee's EUR 62 million) and today's hardest money (semiconductor equipment's RMB 3.688 billion) are not in the same branch, and each is still missing a piece — Gan & Lee lacks money confirmation (both the stock and the branch saw net outflow), and equipment lacks a second day of inflow (the test only comes due today); ahead of the U.S. CPI print (20:30 Beijing time on August 12), this is a session for testing hypotheses, not for confirming a trend.

---

Data-Retrieval and Failure Log (not sent to clients)

• The East Money main quote API (push2.eastmoney.com) returned 502 on this machine, so akshare.stock_zh_a_spot_em() / stock_individual_info_em() all failed; Sina quotes (hq.sinajs.cn) returned Forbidden.
  • Workable substitutes: ① quotes via Tencent qt.gtimg.cn (GBK, ~ delimited, field order in work/q2.py); ② fund flows via push2delay.eastmoney.com still return 200 — the key finding this time: when the main site is 502 the delayed site works, and pulling already-closed data after the session is unaffected by the "delay"; neither 1.push2 nor 82.push2 works.
• Code resolution: Tencent smartbox Chinese-name lookup returns v_hint="N" (it only supports codes/pinyin), so we switched to the official exchange directories stock_info_sh_name_code (the main board and the STAR Market require two separate calls; symbol='科创板' cannot be omitted, otherwise the entire 688 range is missing) + stock_info_sz_name_code (ConnectionReset on the first attempt, succeeded on retry) + stock_info_bj_name_code, 5540 entries in total. Every stock code in the text comes from the official directory, not from memory.
• The 7 errors caught by the risk-control review (risk-auditor), all fixed:
  1. [Fatal] "JPM raised its memory forecast 48%" is wrong; the Sina original reads "raised by 4%, 6% and 8% respectively." This was the basis for grading Branch 1; 5 instances across the text have been corrected, the item was downgraded from S to A, and the branch's primary evidence was changed from "a large investment-bank upgrade" to "that day's fund flow." Lesson: the WebFetch summary compressed "4%, 6% and 8%" into "48%," the same family as the "English summary billion/亿 tenfold error" — enumerated numbers get compressed into a single value by summarizers and must be checked word by word against the original.
  2. Longsys's limit-up price was written as 463.92 (using the 8/7 close ×1.2); the correct figure is 412.50×1.2 = 495.00. Gan & Lee's 74.47 was right, which is exactly the kind of single-point slip that is easiest to miss. Lesson: every price-derived figure must be recomputed on a single basis — "base date = the 8/10 close."
  3. Gan & Lee's cross-basis ratio: dividing the revenue-basis RMB 485 million straight into after-tax net profit attributable to parent to get "42.4% of profit" was a systematic overstatement. It has been changed to state the revenue basis (12.0%) and the after-tax estimate separately (about RMB 450 million / 39%, using the company's 7.01% effective tax rate), and the assertion that it "exactly fills the gap" was deleted.
  4. Compliance: removed "any direction is worth only a probing position" (a position recommendation) and phrasing such as "not worth chasing / do not open new positions / avoid," and added the "no position sizing or entry/exit points" disclaimer at the top of 9.1.
  5. A self-contradiction in trailing-PE methodology: the text warned about the low-PE trap for Longsys yet used PE 121.68 to Pass CXMT — while CXMT's H1 guidance is net profit attributable to parent of RMB 50.0-57.0 billion, making that PE badly distorted. It has been changed to "the Pass reason is that money and the event are already realized, not valuation," and loss-making stocks were rewritten as "PE not applicable."
  6. Date of the share-sale announcement: 8/7 → after the 8/6 close (announcement file name AN20260806…).
  7. In-window news that had been missed: the coal "15th Five-Year Plan," the indium-phosphide front page, the PCB upstream price-increase front page, the Jianshe Machinery halt and others have been added; the PCB conclusion was changed to present both sides.
• Four substantive corrections produced by sub-agents (all merged into the text): Longsys had already issued guidance on 7/3 (RMB 9.20-11.00 billion), so this is old news being realized; the "private placement at a 45% premium" is a media misreading and was in fact a 22.56% discount; Gan & Lee had a 8/6 announcement of share sales by seven directors and senior managers (entirely missed in the first draft); the vendor PB of 14.48 for Longsys uses Q1 net assets and will mechanically reset to about 9.5× today.
• Basis differences: this brief's fund-flow figures (Longsys +RMB 922 million, AMEC +RMB 891 million, WuXi −RMB 139 million) differ from those cited in the 8/10 recap (RMB 863 million / 890 million / −231 million) because the snapshots are on different bases; both are for the 8/10 close and the ranking is unaffected. Internally this brief uses the self-retrieved push2delay basis throughout.
• Still not obtained (explicitly disclosed to clients in the text): the 8/10 Dragon-Tiger List (published after 18:00); the day-by-day 8/6-8/9 fund-flow series for semiconductor equipment (the first half of the "two consecutive days of net inflow" test, missing); Gan & Lee's 8/11 real-time pre-market quote; WuXi AppTec's H1 ex-non-recurring profit and share of non-recurring items; the primary source and date for KLA/Lam Research's "US$150 billion WFE."
• Methodological limits on valuation percentiles: the PE-PB percentiles for Longsys/Biwin/GigaDevice come from the Baidu valuation series (akshare stock_zh_valuation_baidu), whose sample starts at each company's listing date (Longsys 2022-08-05, n=1466), so "the past ten years" is an API label and the real span is only about 4 years; PE was negative on 46% of Longsys's trading days, so its PE percentile is a percentile "within positive-value samples only."
• Two items actively verified to avoid misquoting: ① JPMorgan's US$1.442 trillion for 2027 was checked against its components (DRAM US$982.9 billion + NAND US$458.7 billion) and sums consistently on a revenue basis — the tenfold error is not here; the error was in the size of the upgrade; ② the "permanent shutdown of 2,200 tons of Japanese tungsten hexafluoride capacity" is an exaggerated narrative that has already been disproven (Guancha.cn), and this brief does not use it.

⚠️ Risk warning: this list is pre-market information gathering and observation only and does not constitute investment advice. A-share volatility risk is extremely high, and automatically generated content may contain stale information or errors in supply-chain mapping; it must not be used directly as a basis for trading.