A-Share Pre-Market Brief | 2026-07-27 Monday

Time-window note: This report was generated around 19:30 on 2026-07-27 (Monday); the primary window is 2026-07-27 15:00 close → 19:30 (post-close filings and evening news). All news from during or before the 7/27 session is tagged "out-of-window (carry-over)" and down-weighted, and is not treated as a fresh catalyst for 7/28. This list serves the 2026-07-28 (Tuesday) open; "tomorrow / next day" in the text always means July 28.

Verification tiers (applies throughout):

• ★ [primary-verified] — item-by-item verification by fundamentals-analyst against original CNINFO filings and financial-statement interfaces: 凯盛科技 (Kaisheng Technology, 600552), 东山精密 (Dongshan Precision, 002384), 东方精工 (Dongfang Precision, 002611), 神火股份 (Shenhuo Coal & Power, 000933), 通富微电 (Tongfu Microelectronics, 002156), 深科技 (Shenzhen Kaifa, 000021), 长城军工 (Great Wall Military Industry, 601606), 中曼石油 (Zhongman Petroleum, 603619) — 8 names in total.
• △ [secondary-verified] — risk-auditor searched and confirmed the key facts (code / board / main business / listing date) but did not pull financial-statement interfaces: 红板科技 (Redboard Technology, 603459), 西安奕材 (ESWIN Material, 688783).
• ○ [unverified] — only news-flow and tape evidence; financials, share-price position and business-match have not been confirmed by this report.

Hard scoring constraints: for ○ names, "earnings elasticity" is capped at 8 pts, "industry position & fundamentals" at 10 pts, and "expectation gap" scores 0. Therefore, when an ○ name appears near the top of this list, its score reflects the hardness of the branch, not the certainty of the stock.

---

0. Today in One Sentence

1. This window produced no new catalyst that changes the landscape; the real variable is a "driver switch" that was already completed during the 7/27 session: at 13:30 Iran sent a de-escalation signal ("as long as the US continues to suspend hostile actions, Iran will also stop its strikes"), WTI fell as much as 6.1% intraday and Brent -5.61%, and A-shares rallied in response, with 5196 advancers and 121 limit-ups (Sina Finance¹). Note: this repricing was fully completed the same day — 中曼石油 (Zhongman Petroleum, 603619) already hit limit-down and the ChiNext index was already +3.16%. The incremental information for 7/28 is not "de-escalation" itself but "whether de-escalation persists", and that cuts both ways.
2. The only new main-board industrial filing in this window is the TGV pilot line from 凯盛科技 (Kaisheng Technology, 600552) — but in its own filings on 6/18 and 7/1 the company twice stated in writing that TGV "has not yet generated any operating revenue". On the filings front, this window carried more negative information (loss pre-announcements, share reductions, risk warnings) than positive.
3. Strongest carry-over branch: the PCB / copper-clad laminate (CCL) / electronic-cloth price-hike chain (an out-of-window carry-over trend, not a new catalyst): in July, high-end AI CCL M8/M9 rose 12%–22% and low-dielectric electronic fiberglass cloth rose 12%–20%; electronic cloth saw its fifth price increase of the year, to an average price of RMB 7.4/metre (Sina²). Its value lies in being "checkable and able to show up in the accounts", not in "there being news tonight".
4. The flow structure must be seen clearly: today's turnover was RMB 2.08 trillion, "up RMB 145.5 billion" from the prior day, but 长鑫科技 (ChangXin Technology, 688825) alone traded RMB 141.1 billion — stripping out ChangXin, the rest of the market added almost nothing, or even shrank. At the same time there were 121 limit-ups but only 8 stocks with consecutive limit-ups — an extremely thin ladder.
5. Pre-market state: sentiment is repairing but the structure is unbalanced, and the FOMC on 7/28–7/29 hangs overhead. This is a position well suited to screening branches and very poorly suited to chasing high-level relays.

---

1. News Overview

Impact level: S = changes an industry trend or delivers direct order elasticity; A = clearly positive for a branch with multiple beneficiaries; B = positive but with a long chain that needs verification; C = sentiment stimulus with weak persistence. in-window = released after 7/27 15:00; out-of-window (carry-over) = before 7/27 15:00, background only, already down-weighted.

7/27 tape (out-of-window · background): SSE Composite +1.15%, SZSE Component +2.72%, ChiNext +3.16%, CSI 300 +2.49%; turnover RMB 2.08 trillion, up RMB 145.5 billion from the prior day — but 长鑫科技 (688825) alone traded RMB 141.1 billion, and after stripping it out the rest of the market added almost nothing. 5196 advancers / 121 limit-ups / only 8 stocks with consecutive limit-ups (JRJ limit-up recap¹⁹). Last Friday, 7/24, saw a broad single-day decline, with the SSE Composite -1.61% at 3814.20 and turnover shrinking to RMB 1.94 trillion.

---

2. Evidence-Backed Branches, in Descending Order

The heading has been changed from "strongest positive branches" to "evidence-backed branches" — because this window (after 7/27 15:00) produced no new strong branch, and the ordering below reflects evidence hardness, not "positives that came out tonight".

A contradiction about gold that must be stated: the 7/27 close report said "gold and other sectors led both markets" (The Beijing News²⁰), which is the opposite of the intuition "geopolitics cools → gold falls back". This report cannot explain the divergence and has not verified any gold stock, so it neither lists gold among the losers nor makes any directional judgement.

---

3. Overall Stock-Level Catalyst-Strength Ranking

"Main board": ✅ = Shanghai/Shenzhen main board; ❌ = ChiNext / STAR Market / BSE (observation only, not a primary recommendation). "Verification": ★ primary-verified ／ △ secondary-verified ／ ○ unverified (scores already capped per the hard constraints).

Observation zone (not main board; no primary recommendation per the iron rules)

---

4. Stock Scoring Model (100 points total)

Breakdown by item (top 6)

Two scores that must be read properly:

• 神火股份 has the highest fundamentals score on the board at 14/15, yet only 2 points for expectation gap (the exact-figure pre-announcement on 7/9 and the limit-up on 7/10 already played out the move) — a good company ≠ a good trade tomorrow.
• 宏和科技 climbed to 4th on "directness 17 pts" (its product really is going up in price), but earnings elasticity, fundamentals and expectation gap were all capped or zeroed — its score is the branch's score, not the stock's score.

---

5. Detailed Analysis of Key Stocks (top 6 overall + the primary-verified 凯盛科技)

Note: this section is not strictly "the overall top 10"; it covers ① the top 6 overall and ② 凯盛科技 (600552), the subject of the only new main-board industrial filing in this window. Names 7–10 (生益科技 600183, 中国巨石 600176, 沪电股份 002463, 深南电路 002916) are all unverified ○ names, and everything that can be written about them is already covered in Sections 2, 3 and 7, so they are not expanded separately.

1深科技000021Amemory packaging & testing ｜ A ｜ 70 pts ｜ watch closely · Shenzhen Kaifa

• Related news: 长鑫科技 (688825) debuted on 7/27 at +465.82% with turnover of RMB 141.1 billion (CNR¹¹); TrendForce, published 2026-07-03, expects 26Q3 DRAM contract prices +1318% and NAND +1015%, but explicitly notes the increase has narrowed from about 60% in Q2 because consumer-end price tolerance has hit its limit (TrendForce¹³). Both are out-of-window (carry-over).
• Positive tier: an indirect supply-chain positive, but with the highest evidence density on the whole list — the 2025 annual report formally disclosed memory semiconductor revenue of RMB 4.091 billion, 25.98% of the total, with a gross margin of 20.40% (over the same period high-end EMS manufacturing was 54.20% of revenue at a gross margin of only 9.71%).
• Branch stage: already fermenting for 3 months, and it has been falsified once. From 6/16 to 6/30 the stock rose 57.64% over 11 trading days, peaked at RMB 66.20 on 7/1, then fell over 13 trading days to RMB 35.14 (-46.9%), with two limit-downs along the way. It is now RMB 43.95.
• Fundamentals [★ primary-verified]: 2026Q1 revenue RMB 3.724 billion (+10.67%), net profit attributable to parent RMB 242 million (+35.35%), ex-non-recurring RMB 143 million (+14.9%), gross margin 17.07%, gearing 41.68%, cash of RMB 8.10 billion against interest-bearing debt of about RMB 6 billion (net cash).
• Four deductions you must know about:
  1. No 2026H1 pre-announcement was issued. The mandatory deadline for interim pre-announcements on the Shenzhen main board is within 15 days of the end of the first half, i.e. July 15 — that window has passed (Tongfu Microelectronics, also on the Shenzhen main board, pre-announced normally on 7/14). This implies Shenzhen Kaifa's H1 net profit attributable to parent most likely did not move more than ±50% YoY. But an exemption must first be ruled out: if the absolute value of EPS in the year-earlier period was ≤ RMB 0.03, mandatory disclosure is waived; this report did not verify whether its 2025H1 EPS triggers that exemption, so this inference is "highly suspicious" rather than "confirmed".
  2. 2026Q1 operating cash flow was only RMB 114 million, covering net profit attributable to parent 0.47x, -82.9% YoY; inventories rose RMB 439 million QoQ; the period expense ratio jumped from 5.55% to 11.12%.
  3. It holds only 55.88% of the memory entity Hefei Peiton Storage (Big Fund Phase II holds 31.05%). The discount must be split into two bases: the discount on existing profit is about 0.8x (the historical basis, with minority interests at 21.6% of net profit in 2025); the correct discount on the incremental profit from memory price increases is about 0.56x (directly on the 55.88% equity stake). The latter is the right yardstick for judging price-hike elasticity.
  4. The RMB 1.470 billion expansion projects in this round complete in June and December 2027 respectively, so they add almost no "volume" to 2026–2027 earnings; elasticity can only come from running existing capacity flat out plus passing through unit prices.
• The company's own cooling language (original text of the 7/1 unusual-movement filing): "the company's HBM technology is still at the R&D stage and is not expected to generate related sales revenue or profit in the near term"; on specific business questions involving ChangXin Memory and Yangtze Memory, the company's uniform reply in its 5/21 investor-relations record was "customer information is subject to confidentiality agreements between the company and its customers". The online claim of "60–70% of ChangXin's orders locked in" has no company disclosure behind it.
• Industry position: one of the few main-board names with integrated wafer testing + memory packaging + module capability; it is not on the ChangXin strategic-placement list. Sell-side coverage is extremely thin (the most recent in-depth report on the stock dates from 2023-11-23).
• Technical sentiment: Shenzhen main board; 7/27 +4.67% closing at RMB 43.95, no limit-up; turnover ratio 14.82%, turnover RMB 9.77 billion; -33.6% from the 7/1 high, but already +25.1% from the July low of RMB 35.14 — read the two anchors together and the position is "low-to-mid but has already bounced some way".
• Final judgement: the best "position × evidence density" on this list, but "no interim pre-announcement" is a negative signal the market has not yet fully digested, and the 8/26 interim report is the only point of adjudication. Positioned as an observation.

2神火股份000933B+electrolytic aluminium ｜ B+ ｜ 66 pts ｜ watch only · Shenhuo Coal & Power

• Related news (in-window): the H1 report was disclosed at 16:43 on 7/27 (PDF⁴).
• Highest profit quality on the list [★]: ex-non-recurring RMB 4.785 billion (+138.07%) is higher than net profit attributable to parent of RMB 4.781 billion (non-recurring items total -RMB 3.93 million); operating cash flow RMB 7.200 billion = 1.51x net profit attributable to parent; cash rose from RMB 3.017 billion at the start of the year to RMB 11.037 billion.
• Growth breakdown: electrolytic aluminium revenue RMB 17.051 billion (+20.28%) while operating costs fell 12.91% (alumina prices dropped sharply YoY); the electrolytic aluminium gross margin was 45.33%, contributing about 87% of the company's gross-profit increase; coal about 8%. The company disclosed that aluminium products were "produced and sold at full capacity", i.e. revenue growth came almost entirely from selling prices.
• Negative expectation gap (decisive): on the evening of 7/9 it already pre-announced an "exact figure" of RMB 4.800 billion of net profit attributable to parent, the stock hit limit-up +10.02% on 7/10, and July's cumulative gain is +20.09%. The actual figure on 7/27 was RMB 4.781 billion, 0.39% below the pre-announcement; ex-non-recurring and EPS were likewise marginally lower; no interim distribution or bonus issue; and the wording on drivers matches the pre-announcement word for word — zero positive increment.
• Risks: the company's own text acknowledges that "in the second quarter … electrolytic aluminium prices oscillated lower"; the share-reduction window for shareholder Shangqiu Putian Industry & Trade runs to 8/13, with about 0.62% of the quota remaining.
• Technical sentiment: RMB 25.35 (+1.48%), +18.4% over the past month, -31.9% from the April high; PE (TTM) 8.28x, PB 2.06x (as at the 07-27 close).
• Final judgement: the most solid fundamentals on the list, but as a "trade for tomorrow" it is a textbook case of the good news already being realised — this move finished with that limit-up candle on 7/10.

3红板科技603459A-PCB ｜ A- ｜ 62 pts ｜ watch closely · Redboard Technology

• Related news: the PCB price-hike chain (out-of-window carry-over); on 7/27 it hit limit-up together with 国际复材 (301526) and 宏和科技 (603256) (Sina¹).
• Why it was added back: the first draft wrote only "红板科技" without giving a code and did not put it on the ranking — a violation of iron rule 5. On verification, 红板科技 is a Shanghai main-board name, 603459, with a pure-play PCB business (HDI boards, rigid boards, flexible boards, rigid-flex boards, substrate-like PCBs, IC substrates); in 2024 it supplied 154 million handset HDI mainboards to the world's top-ten handset brands, a share of about 13%, and it has 1.6T optical-module PCB mass-production capability. Among today's PCB limit-ups it is the main-board name with the best business match, clearly stronger than 贤丰控股 (002141), which hit limit-up in the same batch.
• Fundamentals [△ secondary-verified, financial interfaces not pulled]: 2025 net profit growth of about 152% (media basis). Revenue / gross margin / cash flow / gearing / current valuation and share-price position were all unverified by this report.
• A deduction that must be counted: a recent IPO listed on 2026-04-08 (offer price RMB 17.70, debut +208.25%). A recent IPO's volatility, ownership structure and lock-up-expiry rhythm are all completely different from a seasoned stock, and this report did not verify its lock-up structure. The risk deduction is set at -5.
• Final judgement: the purest main-board business match in the PCB branch, but the recent-IPO characteristics and the unverified valuation position are two substantive gaps.

4宏和科技603256A-electronic cloth ｜ A- ｜ 61 pts ｜ watch closely · Honghe Technology

• Related news: electronic cloth's fifth price increase of the year, with the average price for common specifications at RMB 7.4/metre (+100% from the 2025Q3 low); leading producers' utilisation exceeds 100% and low-dielectric cloth is booked out more than 4 months ahead (Sina², Huxiu²¹). Both are out-of-window carry-over.
• Positive logic: it directly affects revenue and gross margin — electronic cloth is the key upstream material for AI CCL, and a doubling in price means unit prices and gross margins rise together. This is one of the shortest transmission chains on the board.
• Fundamentals: ○ unverified. Revenue, gross margin, capacity and order book are all data this report has not confirmed. Under the hard scoring constraints, earnings elasticity is capped at 8 pts, fundamentals at 8 pts, and expectation gap scores 0.
• Branch stage: mid-fermentation but lacking a core leader. Four PCB / computing-hardware names hit limit-up today (贤丰控股 002141, 国际复材 301526, 宏和科技 603256, 红板科技 603459), of which 3 are main board and one has a doubtful main business; core leaders such as 生益科技 (600183), 中国巨石 (600176) and 沪电股份 (002463) did not hit limit-up on matching volume.
• Final judgement: the branch logic is the hardest on the board; the stock's certainty is the lowest among the top names. Its value is "track this branch", not "this ticket has been validated".

5通富微电002156B+advanced packaging ｜ B+ ｜ 59 pts ｜ watch only (high position) · Tongfu Microelectronics

• Related news: it appears on the strategic-placement list in ChangXin Technology's offering announcement, as "a listed semiconductor supply-chain company with a strategic cooperation relationship with the issuer", allotted 18.2448 million shares / RMB 158 million with a lock-up from 12 months; on 7/14 it pre-announced H1 net profit attributable to parent of RMB 1.6–1.8 billion (+288.26%~336.80%). Both are out-of-window carry-over.
• Tier: an industry-trend positive plus an indirect supply-chain positive one layer removed from memory; it does not constitute a "direct order positive". The company has disclosed no ChangXin orders and no memory revenue share, and its five investor-relations records (6/9, 6/15, 6/23, 7/7, 7/17) never name ChangXin Technology or Yangtze Memory. The circulating claims of "45% of ChangXin's packaging orders" and "annual orders of RMB 10–15 billion" all originate from self-media.
• Profit structure [★]: H1 pre-announced net profit attributable to parent midpoint RMB 1.7 billion − ex-non-recurring midpoint RMB 750 million = non-recurring items of about RMB 950 million, 56% of net profit attributable to parent, which the filing attributes to "industrial investments around the supply chain and up- and downstream, which produced good investment returns". Extrapolating RMB 1.7 billion ×2 to a full-year RMB 3.4 billion and saying "the PE is only 34x" applies an operating valuation multiple to one-off gains. Stripping those out, the real operating elasticity is still decent: single-quarter Q2 ex-non-recurring was about RMB 578 million, +83% YoY and +237% QoQ.
• The revenue backbone is not memory: the 2025 annual report discloses only "IC packaging and testing RMB 27.248 billion, 97.59% of revenue", with no memory breakout; its largest customer, AMD, contributes more than 50% of sales (private-placement prospectus). Treating it as a "memory elasticity play" misreads its revenue structure.
• Technical sentiment: RMB 76.67 (+0.04%), +103.8% year to date and +189.4% over the past year, only -9.5% from the all-time high of 84.70 (7/15); on 7/27 the intraday range was 7.44%, turnover ratio 15.30% and turnover RMB 17.28 billion; within July there have been violent swings such as -9.99% on 7/17, -7.84% on 7/20 and +10.00% on 7/21. PE (TTM, ex-non-recurring) 128.1x, PB 7.43x (as at the 07-27 close). The shareholder count was +24.46% QoQ in 2026Q1.
• Additional risks: the Big Fund Phase I has already cut from 11.72% to 5.00% (2026-06-24), and once below 5% further reductions need no pre-disclosure; the RMB 4.4 billion private placement has been registered with the CSRC (7/4) and actual issuance will bring about 3.7% dilution. Risk deduction -7.
• Final judgement: the logic is there and the earnings are there, but the price is near the all-time high, half the profit is one-off, and there is dual pressure from shareholders and share capital. "Follow the branch; this ticket's position is expensive."

6东山精密002384B+optical modules / PCB ｜ B+ ｜ 59 pts ｜ watch only · Dongshan Precision

• Related news (in-window): at 16:44 on 7/27 it filed its first buyback of 115,000 shares for RMB 22.94 million (PDF⁵); on 7/15 it pre-announced H1 net profit attributable to parent of RMB 2.9–3.0 billion (+282.58%~295.78%) and ex-non-recurring of RMB 2.4–2.5 billion (out-of-window).
• A logic mismatch that must be called out [★]: the total buyback of RMB 200–300 million is 0.077% of the RMB 388.1 billion market cap; the first actual execution of RMB 22.94 million is 0.0006% of market cap; the purpose is employee shareholding / equity incentives with no cancellation, so there is no EPS accretion at all, and share-based payment expenses will be recognised in future (a deduction from profit at the margin). This is not the same thing as 宁德时代 (300750)'s RMB 20–40 billion buyback for cancellation.
• The real elasticity: Source Photonics has been consolidated since October 2025, and the company's filings disclose that it contributed 16.02% of revenue and 52.92% of profit in 2026Q1; 100G/200G EML optical chips are shipping in volume; on 6/16 the board approved a US$1.2 billion expansion project. This is the part that reaches "industry-trend positive" level.
• Deductions: 2026Q1 capex of RMB 2.160 billion > operating cash flow of RMB 1.127 billion, with the gap filled by RMB 2.4 billion of financing, so free cash flow is negative, and gearing is 63.69%; PB 17.14; backing out from the pre-announcement, Q2 non-recurring items are about RMB 450 million (15–17% of H1 profit); the H-share listing application resubmitted on 5/21 carries dilution risk; on the rumour of "RMB 40 billion of profit in 2027" the company replied explicitly that "the company does not publish earnings forecasts". Risk deduction -9.
• Technical sentiment: RMB 211.90 (+6.39%), turnover RMB 16.63 billion, -23.2% from the 6/25 high, +150.3% year to date.
• Final judgement: the fundamentals are strong, but tonight's "buyback" filing is not a reason to buy — chasing the price higher because of the buyback means buying into the wrong narrative.

7凯盛科技600552BTGV glass substrate ｜ B ｜ 52 pts ｜ watch only · Kaisheng Technology

• Related news (in-window; the only new main-board industrial filing in this window): the board resolution at 15:32 on 7/27 (No. 2026-022, PDF³) to invest about RMB 150 million in a TGV advanced-packaging glass pilot line within the Bengbu UTG Phase II plant, "to carry out key-process optimisation R&D and advance product trial production and customer validation"; the filing simultaneously warns that "there is uncertainty in the project implementation timeline".
• The company's own counter-evidence (most critical): on 2026-06-18 (filing 2026-019) and 2026-07-01 (filing 2026-020) Kaisheng twice stated in the original text — "the company's TGV technology is still at the early R&D stage and has not achieved any commercial mass production, nor generated any operating revenue"; it also stated that as at the 6/30 close its PE was 239.56 versus a 78.46 average for its C39 industry, and proactively warned of "overheated market sentiment and the risk of irrational speculation".
• Order-of-magnitude estimate: RMB 150 million = 2.6% of 2025 revenue of RMB 5.880 billion; the pilot line itself generates no revenue at scale, and once capitalised and depreciated over 10 years it adds about RMB 15 million of annual depreciation, so in the near term it actually weighs on profit first. In the base case the actual elasticity for 2026–2027 revenue and profit is close to 0.
• Fundamentals [★]: 2025 revenue RMB 5.880 billion (+20.15%) but net profit attributable to parent RMB 128.7 million (-8.20%), net margin from 3.98% to 2.19%, and single-quarter 2025Q4 net profit attributable to parent of only -RMB 10,200; 2026Q1 operating cash flow only RMB 6.72 million, covering net profit attributable to parent 0.24x; current ratio 0.96 (<1), quick ratio 0.44, gearing 61.93%; full-year diluted ROE only 3.04%. Main business: display materials 78.67%, application materials 19.37%.
• Orders / customers / certifications: all without public evidence. The filing itself says "advance product trial production and customer validation", i.e. customer validation is still in progress.
• Technical sentiment: RMB 16.39 (+5.40%), turnover ratio 6.35%; -49.7% from the 6/30 high of RMB 32.58 (a low position is its only plus); PE (TTM) 116.6x, PB 3.63; four unusual share-price movement filings in the past two months (6/9, 6/18, 7/1, 7/4 — two of them in July).
• Final judgement: "the story's progress bar moved forward one notch", not "the revenue statement is about to change". It is stronger than a pure concept (a board resolution, a real budget, R&D accumulated since 2024), but it is very far from earnings. If tomorrow's open is clearly more than 5% higher, that means the market has read the pilot line as mass production — which directly conflicts with the company's own public statements and is a pricing contradiction that needs re-assessment (an observation item).

---

6. Pass List

---

7. Intra-Branch Ranking

Branch 1: the PCB / CCL / electronic-cloth price-hike chain

• Hardest stock: 生益科技 (600183) — the shortest price-increase transmission chain.
• The limit-up main-board name with the purest business: 红板科技 (603459).
• Who is following: 中材科技 (002080), 贤丰控股 (002141).
• What this branch lacked today: a core leader. Everything that hit limit-up was a high-beta or back-row name (Redboard, Honghe, Xianfeng, International Composite Materials), while core leaders such as Shengyi, Jushi and WUS did not hit limit-up on matching volume. When the core leaders do not move in step, the ladder usually does not survive 2 trading days.

Branch 2: memory / advanced packaging

• Hardest stock: 深科技 (000021) — the only main-board name with a formally disclosed memory revenue share.
• Shared evidence gap: the quantified business relationship between every A-share company and ChangXin Technology (order value, share, revenue contribution) is entirely undisclosed, and figures circulating online are all uncreditable.
• Structural weakness: the core leader is on the STAR Market and the main board has only satellites — this is the fundamental reason it ranks behind the PCB chain.
• Additional note: TrendForce explicitly states that the 26Q3 increase has narrowed from about 60% in Q2, so this branch's price momentum is weakening, not strengthening.

Branch 3: innovative drugs

• Structural problem: the core leader is on ChiNext and the main-board names are all followers — this branch has no hard main-board name to take the baton, and once CSPC Innovation pulls back the main-board followers lack support. This is also why, despite a branch strength of B+, its main-board participability is the lowest.

---

8. Opening Verification Signals for Tomorrow (2026-07-28)

8.0 The 7/28 trading calendar and overnight variables (missing from the first draft; completed here)

8.1 Auction signals

• 长鑫科技 (688825): the master sentiment valve for the whole board. Its debut turnover ratio already exceeded 66% with turnover of RMB 141.1 billion, so next-day selling pressure is heavy. Treat the size of its opening gap and the bid absorption as a reading of branch strength, not as any basis for action. If it gaps up and fades, or simply gaps down, today's entire memory / semiconductor premium will be given back in step.
• 长城军工 (601606): 3 consecutive limit-ups + an H1 pre-announced loss of RMB 96 million + a bid queue of only RMB 164 million (6.7% of turnover) + cooling geopolitics. A one-word board or a large gap up would diverge severely from its fundamentals — a structure to treat with great caution; a gap down means the defence line most likely retreats that day.
• 立新能源 (001258): 7 limit-ups in 8 days + a self-initiated risk warning. It is the leading indicator for sentiment in high-level names on the whole board — if it breaks its limit-up seal or gaps down, every consecutive-limit-up name that day (珍宝岛 603567, 哈三联 002900, 创新医疗 002173, 银星能源 000862) needs caution.
• 凯盛科技 (600552): a gap up within 5% = a rational market; clearly more than 5% (especially near limit-up) = the market has read the pilot line as mass production, which directly conflicts with the company's own "zero revenue" filing and means the pricing contradiction is widening (an observation item).
• The PCB chain: if 红板科技 (603459), 宏和科技 (603256) and 贤丰控股 (002141) gap up while 生益科技 (600183), 中国巨石 (600176) and 沪电股份 (002463) do not follow on volume, that means there is only sentiment and no core leader, and the branch does not hold.

8.2 Sector signals

• Confirmation signal: 3 or more fast limit-ups in the PCB / CCL chain, with a core leader such as 生益科技 (600183) or 沪电股份 (002463) moving on matching volume → the branch upgrades from "sentiment" to "main line".
• Confirmation signal: the memory chain forms a second- and third-board ladder (at least two of 深科技 000021, 通富微电 002156, 太极实业 600667 moving together on volume), rather than ChangXin standing alone.
• Falsification signal: the innovative-drug ladder (4 limit-ups in 8 days → 3 in 4 days → 2 in 4 days) collectively breaks its seals, meaning the return of divergence has failed.
• Calendar constraint: the brain-computer-interface time window only runs to 8/1, leaving just 4 trading days this week.

8.3 Stock-level signals

• Front-row absorption: watch whether 2 or more of the top 5 in the limit-up ladder (立新能源 001258, 珍宝岛 603567, 哈三联 002900, 长城军工 601606, 创新医疗 002173) seal quickly before 9:45. If none do, the ladder has broken (a signal, not a trading recommendation).
• Volume without breaking the seal: if 红板科技 (603459), 宏和科技 (603256) and 南京熊猫 (600775) can hold their seals on volume, capital is willing to bid on the new branch.
• Flow direction: the direction of large-order flows in 深科技 (000021) and 通富微电 (002156) is the most direct evidence of whether the ChangXin premium can spill over. Note: this report could not obtain 7/27 large-order flow data (the interface was unreachable) and used no flow figures as a scoring input.
• Positioning risk: if a more recognisable name appears in the PCB chain (for example a core leader going straight to a one-word board), the high-beta names that moved first get pushed to the back row.

8.4 Risk signals (the highest weight today)

1. Geopolitics cuts both ways: if the Iranian de-escalation is falsified (new strikes / a new blockade statement), the gains after 13:30 on 7/27 will be given back quickly and defence / oil & gas will rally inversely; if de-escalation continues, defence (长城军工 601606) and oil & gas (中曼石油 603619) keep falling. This is two-way — do not bet one side.
2. Suppressed risk appetite ahead of the FOMC (7/28–29) — the opposite direction to this report's core "rotation back into tech growth" thesis.
3. Overnight US equities are entirely unknown — the US market had not opened when this report was generated, the biggest blind spot.
4. Incremental capital is in doubt: today's turnover of RMB 2.08 trillion looks like an increase of RMB 145.5 billion, but ChangXin alone accounted for RMB 141.1 billion, and stripping it out the rest of the market added almost nothing. If two-market turnover falls back below RMB 1.9 trillion on 7/28, incremental capital has not followed.
5. An extremely thin ladder: 121 limit-ups but only 8 stocks with consecutive limit-ups. If the consecutive count drops below 5 tomorrow, there is no basis for a relay even if the index closes green.
6. High-level names have started to collapse: 立新能源 (001258) has been risk-warned by the company itself, and 中曼石油 (603619) already hit limit-down today. High-level names topping out usually leads a broader weakening of sentiment by 1–2 trading days.
7. A tightening regulatory environment: the Supreme People's Court / Supreme People's Procuratorate judicial interpretation on insider trading takes effect from 7/27; regulators are focused on real-time fund valuations fuelling short-term speculation; a former CSRC vice-chairman is under investigation — a persistently negative environmental variable for high-level thematic speculation.
8. Peak interim-reporting season: "every positive that was pre-announced has already been digested; what was not pre-announced could be a landmine" — 深科技 (000021) not pre-announcing is one such case.

---

9. Final Conclusions

① The 5 main-board stocks most worth watching today

Important: this section is a watch list, not a buy list. Each name carries its conclusion label directly — please read the label along with the name.

An evidence gap that must be stated: names 1 and 2 are ★ primary-verified, name 3 is △ secondary-verified, and names 4 and 5 are ○ entirely unverified. The ranking reflects a composite of "branch hardness + verifiability + position", not an ordering by certainty, and it is certainly not investment advice. Ranked purely by evidence strength, only 深科技 (000021) and 神火股份 (000933) are reliable.

② The 3 strongest branches today (consistent with the branch table in Section 2)

The 4th-placed brain-computer interface (B) has been removed from the "3 strongest branches" this time because the financing figure was corrected (RMB 44 billion → about RMB 4.4 billion) and because "recommended GB/T standards lower the device registration threshold" lacks any regulatory text to support it; only the calendar value of 8/1 is retained.

③ Directions not worth chasing today

1. Defence / oil shipping (the geopolitical line) — the driver itself is disappearing. Brent fell from 96.78 on 7/24 to 89.41 on 7/27 (-7.6%), and 中曼石油 (603619) already hit limit-down today. 长城军工 (601606) with 3 consecutive limit-ups on top of an H1 pre-announced loss of RMB 96 million, negative ex-non-recurring earnings and zero order filings is an extreme case of pure sentiment.
2. High-level power / green-power names — after 7 limit-ups in 8 days the leader 立新能源 (001258) issued its own risk warning (PE 479.5 vs industry 20.17), and the follower 银星能源 (000862) is riskier still.
3. The accounting trap in "earnings explosions" — of 东方精工 (002611)'s +867.75% net profit attributable to parent, 97.76% is a one-off gain from selling an asset, with ex-non-recurring -66.48% and revenue -21.68%. Be wary of any list that reports only net-profit growth without looking at the ex-non-recurring figure.
4. "Progress bar" stocks — 创新医疗 (002173)'s filing says AM5 "has not generated revenue so far"; 凯盛科技 (600552)'s filing says TGV "has not generated any operating revenue"; 蓝思科技 (300433) has only an MOU with Intel, no orders and no amount. What they share: the company itself has already denied any near-term earnings contribution in its filings, yet the market is pricing them on earnings.
5. The "buyback ≠ positive" mismatch — 东山精密 (002384)'s buyback is 0.077% of market cap, is not for cancellation and does not boost EPS; 科大讯飞 (002230) and 工业富联 (601138) are the same. A buyback for cancellation (宁德时代 300750, RMB 20–40 billion) and an incentive buyback (东山精密) are two completely different things.
6. Label-mismatch stocks — 首开股份 (600376), a property developer, is classified under humanoid robots; 龙蟠科技 (603906), whose main business is LFP cathode material, is classified under the VC price increase. The branch being right does not mean the stock is right.

④ The final one-sentence judgement

This window (after the 7/27 close) produced no new strong catalyst and the filings skewed negative; the real conflict on 7/28 is "whether the money released by the geopolitical retreat gets positioned before the FOMC". The geopolitical premium was knocked out in one go during the 7/27 session, and ChangXin's listing pushed the memory pricing anchor to an extreme, but stripping out ChangXin the whole market added almost nothing, and of 121 limit-ups only 8 were consecutive — this is a position suited only to screening branches and very poorly suited to chasing high-level relays; and every name with 7 limit-ups in 8 days, 3 consecutive limit-ups, or a self-disclosed zero revenue belongs to the category of buying the wrong ticket in the right branch.

---

Appendix: Known Uncertainties in This Report and This Revision Log

A. Remaining uncertainties

1. The 7/27 US session is unknown — this report was generated at 19:30 Beijing time and the US market only opens at 21:30. This is this report's single biggest blind spot, especially for the overnight direction of the semiconductor and memory chains.
2. The PBoC reverse-repo size conflicts between two sources (RMB 2.1 trillion vs RMB 1.8 trillion); the PBoC announcement prevails.
3. For the ○ names (宏和科技 603256, 生益科技 600183, 中国巨石 600176, 中材科技 002080, 沪电股份 002463, 深南电路 002916, 贤丰控股 002141, 南京熊猫 600775, 哈三联 002900, 珍宝岛 603567, 益佰制药 600594, 海南海药 000566, 明新旭腾 605068, 北自科技 603082, 智度股份 000676, 圣阳股份 002580, 龙蟠科技 603906, 科大讯飞 002230, 工业富联 601138, 太极实业 600667, 银星能源 000862, 立新能源 001258, 创新医疗 002173, 首开股份 600376), financials, share-price positions and business match were all unverified; 红板科技 (603459) is secondary-verified.
4. 7/27 large-order flow data could not be obtained (the interface was unreachable), and no flow figures were used as a scoring input anywhere in the text.
5. Whether the end-July Politburo meeting will be held and on what date, and the 7/31 official PMI — could not be verified.
6. The inference that 深科技 "issued no pre-announcement" has not yet ruled out the "year-earlier EPS ≤ RMB 0.03" exemption.
7. The exact date on Kingboard's sixth price-increase letter could not be confirmed, so the release time for news #13 is labelled "ongoing through July" and has been down-weighted.
8. 石药创新 (300765)'s H1 pre-announcement was not traced back to the company's original filing, is accepted at the media tier, and the level has been cut from A to B+.

B. Substantive revisions after this round of risk-auditor quality control

---

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