A-Share Pre-Market Brief | 2026-08-04 (Tuesday)

⚠️ Risk disclaimer (up front): This list is pre-market information gathering and observation only. It does not constitute investment advice, and contains no ratings, target prices, position sizing, or buy/sell recommendations. The "catalyst grade," "score," and "attention ranking" used below are a research ordering of how hard the public information is, and do not represent any judgment about future returns.

Data-basis disclosure first (please read this section before anything else)

1. Time window: 2026-08-03 15:00 (previous session's close) → 2026-08-04 08:10 (cut-off). Each item inside the window is tagged with its release time.
2. Market data source: Single-stock price change, turnover value, turnover rate, PE (TTM), and free-float market cap are all on a 2026-08-03 close basis, taken from public market-data vendors; limit-up pool / failed-limit pool / dragon-tiger list, and industry and concept fund flows are taken from East Money public data (date 20260803). Financial data was reconciled item by item against the original CNINFO periodic reports.
3. Policy-document basis (the single most important methodological statement in this issue): the 15th Five-Year Plan for Building the New Power System (NDRC Energy [2026] No. 942, dated 2026-07-21, publicly released on 2026-08-03) — this report did not stop at media paraphrase. It is based on local full-text parsing and word-frequency counting of the PDF attachment (18 pages) on the NDRC official website, and was re-downloaded and independently re-verified by a separate quality-control step (page-by-page verification that the whole document is an extractable text layer with no image-based text; the word frequencies reproduced exactly). This step completely changed the direction of this issue's conclusions — see Section 2.1.
4. This report introduces an "evidence grade" system (please note): not every stock went through equally strong verification. Sections 3 and 9 are tagged with an evidence grade: A = reconciled item by item against the original annual report / CNINFO filings; B = cross-checked against third-party financial databases; C = only price and valuation pre-screening completed. Confidence in the fundamental details of grade-C names is markedly lower than for grade A; please weight your reading accordingly.
5. Clarification of label meanings (compliance): "priority deep-dive / watch closely / watch only / Pass" are this report's fixed research-priority terms, indicating the order in which research resources are allocated and the attention ranking. They are not ratings, not buy/sell advice, and carry no target price or position size.
6. One basis difference that was checked but remains unresolved: for 601567, the securities short name is disclosed by the exchange as 三星医疗 (Sanxing Medical), while market-data vendors show 三星电气 (Sanxing Electric). This report uses the exchange's version, 三星医疗 (Sanxing Medical), by source priority, and keeps the alias note so clients can search either way.
7. Items not obtained, written as "no reliable data":
   • August 4 pre-market auction data (cut-off 08:10; the call auction does not start until 09:15) — Section 8 gives only judgment criteria and does not predict auction numbers.
   • Northbound single-day net flow: the exchanges stopped disclosing single-day net buying from August 2024, so no direction is given.
   • Segment product revenue for several companies (China XD Electric's UHV share, XJ Electric's pure charging-pile revenue, Dongfang Electric's full-year 2025 nuclear revenue, NARI Technology's converter-valve volume) was not separately disclosed by any of the companies, and is uniformly marked "no reliable data." We do not back out a number from winning-bid amounts and pass it off as a disclosed figure.
   • The full text of this plan discloses no total investment amount whatsoever ("投资" / investment appears only 3 times, all in mechanism-type wording such as "zero-investment" grid connection and the transmission-and-distribution tariff incentive-and-constraint mechanism). The widely circulated "RMB 5 trillion of grid investment in the 15th Five-Year Plan" does not come from this document, and this report does not cite that number to support any single-stock conclusion.
8. One relationship correction verified back to primary sources: the controlling shareholder of Pinggao Electric, China XD Electric, and XJ Electric is in all three cases China Electric Equipment Group (ultimate controller: SASAC of the State Council), not the State Grid system. Among the names covered in this report, only NARI Technology is in the State Grid system (State Grid Electric Power Research Institute holds 56.91%). The three companies belong to one group, compete against each other in the same State Grid tenders, and each company's share is allocated by the group — this is critical to understanding order allocation. For the first three, State Grid is the largest customer, not a related party.

⚠️ Data-fetch failures and implementation details this run (internal only):

• East Money daily-bar interface returned RemoteDisconnected across the board; all 11 candidate stocks failed, so we switched to Tencent forward-adjusted daily K-lines. This is the second consecutive trading day this interface has malfunctioned (recorded in the 8/3 recap).
• news_report_time_baidu returned HTTP 403; stock_a_indicator_lg no longer exists in akshare 1.18.79, so valuation percentiles were computed in-house from the East Money stock_value_em series. The East Money push2 single-stock interface returned 502 this run.
• The main agent's first attempt to fetch the NDRC plan PDF via WebFetch returned a compressed binary stream that the small model could not parse, and it temporarily fell back to second-hand media paraphrase. Three sub-agents then each successfully parsed the full text locally with pdfplumber, which directly overturned the main agent's original judgment that "UHV is the first branch"; the risk-control step independently re-downloaded the PDF and re-verified the word frequencies (18 pages, all a text layer, fully reproduced). Lesson: policy PDFs must be parsed locally — never rely on a small model's WebFetch summary, and certainly never let media paraphrase set the branch ranking.
• The first WebSearch ("A 股 8 月 4 日 盘前 财联社 早知道") returned a large volume of stale 2017 content (Anbang, Eastech, Zhongyin), all of which was discarded in favour of the real-time newswire interface. Time contamination in search engines for "Cailian Press morning brief"-type queries is extremely severe.
• Three hard errors caught by risk control (fixed): ① the original table header said pumped storage 65.94 million kW and nuclear 62.48 million kW were "taken from the original Box 1 table in the document" — in fact Box 1 has only 16 items and does not include pumped storage or nuclear; these two base-year figures are external statistics, a case of "using a fake source to support a real conclusion," so a "base-year source" column was added; ② the 2021 pumped-storage plan text says "around 120 million kW" on a "total commissioned scale" basis, whereas the report originally wrote "or more" and compared it directly with "installed capacity" — 8 instances fixed; ③ the "RMB 5 trillion of grid investment" is not in this document ("万亿元" / trillion yuan appears 0 times in the full text); it originally sat in the NARI Technology logic paragraph and has been deleted.
• An omission caught by risk control (added back): item (23) of the document contains a further brand-new granular hard target — "grid-side standalone new-type energy storage of 140 million kW by 2030, with roughly 80 million kW added during the 15th Five-Year Plan, and an equivalent peak-shaving duration targeted at 4 hours" — which the main agent missed during full-text parsing. On that basis it had originally pushed storage down to C+ and called the "expectation gap negative"; this has been changed to pending assessment.
• Two things the main agent nearly got wrong: ① one search result reported State Grid's third batch of UHV winning bids as "RMB 108.72 billion" and the first three batches as "RMB 292.61 billion" (inflated 10×); checked back against the original text, the figures are RMB 10.872 billion / RMB 29.261 billion; ② the draft described Pinggao and XJ as "State Grid system," and a sub-agent checking the shareholder register corrected this to China Electric Equipment Group. Tender amounts and equity relationships must be verified back to primary sources.
• Self-check correction: the previous version of this internal block said "TBEA's coal-price sensitivity calculation was not used in any single-stock conclusion," but Section 5.8 of the main text did in fact use the "coal price ±10% → net profit attributable to parent ±20%" elasticity and drew a qualitative conclusion from it. A disclosure has been added to the main text noting that this calculation depends on an average-price basis with a nearly 2× discrepancy. The fact that the internal record and the main text diverged is itself worth remembering — once the risk-control record becomes inaccurate, nobody checks that point again on the next round.

---

0. Today in one line

1. The strongest catalyst is the only S-grade item: the 15th Five-Year Plan for Building the New Power System (NDRC Energy [2026] No. 942), first reported by Cailian Press at 16:47 on August 3 and carried by Xinhua at 20:55, meaning the public release fell after the close.
2. But "today is its first pricing day" is only an assumption, not a fact — the document is dated July 21, and during the 8/3 session the nuclear concept saw net inflows of RMB 3.343 billion, the highest of any concept market-wide (this document mentions nuclear only once, at 110 million kW), the pumped-storage concept saw net inflows of RMB 1.760 billion (the one target in the document that was revised upward), and second- and third-tier grid-equipment names have been rising for two weeks since 7/21–7/23. "The content leaked early or has already been partly front-run" is an explanation that must be held in parallel, and needs to be tested against today's actual follow-through.
3. What this document actually says and the label the market will most likely stick on it are two different things. Full-text word frequency (independently re-verified): "特高压" (UHV) appears only 1 time, and it is in "Box 7, Power Safety Governance," about construction safety and quality supervision; "变压器" (transformer) 0 times; "换流" (converter) 0 times; while "配电网" (distribution grid) appears 19 times. The substantive wording in the main-grid section is "promote the retrofit and upgrade of existing DC transmission corridors" and "accelerate the completion and commissioning of transmission corridors already included in the plan during the 14th Five-Year Plan period." This is a document centred on the distribution grid, storage, virtual power plants, and power-market mechanisms.
4. But please read a self-imposed limit at the same time: word frequency reflects where the document places its emphasis; it is not equivalent to how investment is allocated. The document discloses no investment amount; "corridors already in the plan to be completed as early as possible" still corresponds to real main-grid equipment orders, and "distribution-grid quality upgrades / rural grid retrofits" will inevitably consume distribution transformers. This judgment is used to correct narrative weighting; it is not sufficient on its own to derive the order elasticity of each link.
5. The biggest perception gap today: yesterday's limit-ups were second- and third-tier grid equipment names (8 of them). If the market keeps reading this document as a "big UHV positive" and chases them higher, the stated reason does not match the document's content. The UHV orders are real (the first three batches' winning bids of RMB 29.261 billion already exceed the RMB 22.062 billion for all of 2025), but they come from State Grid's annual tenders, were made public on July 27, and have no causal relationship with this document.
6. Possible direction of flows: from "misread UHV" toward "the distribution grid and pumped storage the document actually emphasises." The 2030 pumped-storage target of 160 million kW is roughly a third higher than the 2021 plan, and is the most positive expectation gap in the document (though this report was unable to verify its under-construction pipeline coverage — see the footnote in Section 2.1).
7. Driver types: policy (S grade, the only one) + overseas industry (A grade, AI cloud and memory reversed overnight) + price (B+ grade, day 2 of the solar anti-involution trade).
8. Pre-market read: leaning positive, but structure matters more than direction; do not chase. Overnight the Nasdaq rose +2.13% and the FTSE China A50 night session +0.42%. Note: the overnight US gains on August 3 did not produce a higher A-share open — all four major indices opened lower, so external influence should be read structurally rather than at the index level. At the same time, CME data shows a 67.2% probability of a 25bp Fed hike in September (data taken at 06:35 on August 4; three financial data sources agree), and offshore RMB fell 57 pips — both headwinds that compress valuations.

---

1. News overview

Impact grade: S = changes an industry trend or directly drives order elasticity; A = clearly positive for a branch with multiple beneficiaries; B = positive but with a long chain requiring verification; C = sentiment stimulus with weak persistence. Items marked ⏰ already played out during yesterday's session rather than being new after the close, and are weighted down per iron rule 4.

1.1 Older news kept as background but not counted in this window (to avoid treating it as new)

---

2. Strongest catalyst branches, in descending order

2.1 [Core of this issue] Empirical evidence from the full plan PDF: what it actually says

(a) Word frequency (based on the 18-page full PDF from the NDRC website, parsed locally, then re-downloaded and re-verified by an independent QC step with identical results)

⚠️ How to use this table correctly, and its boundaries (read together with the table): word frequency reflects where the document places its emphasis, and is not equivalent to how investment is allocated. The document discloses no investment amount; "corridors already in the plan to be completed as early as possible" still corresponds to real main-grid equipment orders, and "distribution-grid quality upgrades / rural grid retrofits" will inevitably consume distribution transformers. This table is for correcting narrative weighting; it is not sufficient on its own to derive the order elasticity of each link. The one grid box that comes with a project list (Box 6) lists exclusively back-to-back DC projects: 4 approved for construction (Hebei–Liaoning second circuit, Shaanxi–Chongqing, Fujian–Guangdong second circuit, Sichuan–Guizhou) + 3 held in advance reserve (Hunan–Guangxi, Jiangxi–Zhejiang, Gansu–Sichuan) + 3 under research and feasibility study (Jiangxi–Guangdong, Yunnan–Sichuan, Xinjiang–Tibet).

(b) Item-by-item comparison of the plan's quantified targets (base-year sources listed separately)

Footnote ① (basis comparability): the 2021 version used "total commissioned scale," this one uses "installed capacity"; strict comparability between the two is pending verification. "Roughly a third higher" is the arithmetic comparison of 1.6 against 1.2 — please read it within that boundary. Footnote ② (a question this report has not yet answered): this report used "already fully covered by the approved pipeline of 125 million kW" to downgrade nuclear, but did not run the same pipeline-coverage test on pumped storage — as of mid-2026, how much pumped storage is under construction plus approved nationwide is no reliable data in this report. If under construction + approved is already close to 160 million kW, the "most positive expectation gap" characterisation of branch 2 would need to be lowered. This is one of the most important data gaps to fill in this issue. Also worth recording: the 2021 version's 2025 target was "more than 62 million kW," and actual completion was 65.94 million kW — meeting the target, not beating it. Footnote ③: Box 1 contains 16 indicators in total and does not include pumped storage or nuclear; the targets for those two are in the main text, Chapter 5, item (17) and Chapter 2, item (3). "About 40 million kW of inter-provincial mutual support added during the 15th Five-Year Plan" is derived by subtracting Box 1's 80 million from 120 million kW — not stated in the document.

Conclusion: the real increments in this plan are in [distribution grid], [pumped storage], [vehicle-grid interaction], [back-to-back DC], and [grid-side standalone storage], not in UHV, nuclear, or virtual power plants. If UHV, nuclear, and virtual power plant concepts gap up and run today, that is closer to sentiment pricing than to pricing the document's content.

2.2 Branch strength, in descending order

---

3. Overall single-stock catalyst-strength ranking (descending by stock)

Main-board tradable: ✅ = Shanghai/Shenzhen main board; ⚠️ = ChiNext / STAR Market (observation only, not a primary pick). Evidence grade: A = reconciled item by item against the original annual report / CNINFO filings; B = cross-checked against third-party databases; C = only price and valuation pre-screening completed (markedly lower confidence). Prices as of the 2026-08-03 close.

---

4. Single-stock scoring model (100 points total)

⚠️ An important limitation of the scoring: this table scores names of different evidence grades with the same ruler, so the score itself cannot express differences in confidence. Please read it together with the "evidence grade" column in Section 3 — at the same 60+ score, the reliability of a grade-A name and a grade-C name are not in the same league.

Representative calculation (NARI Technology 600406 = 82 points, evidence grade A) News source 15 (verified in the original PDF that "main-distribution-micro coordination" is the Chapter 3 title) + directness 15 (direction named in the document + 18.1% relay-protection tender share) + elasticity 8 (within-grid growth of only +6.12% limits elasticity) + fundamentals 15 (No. 1 in relay protection, OCF/net profit ≥1.27x for 4 years, net cash RMB 20.5 billion, goodwill RMB 1 million) + expectation gap 10 (PE at the 19% percentile of 5 years, stock up only 5.2% in 3 years) + persistence 10 + trading attributes 10 (turnover of 1.24%, the least stretched) + risk deduction −1 (large cap limits elasticity) = 82

Comparison (Tongwei 600438 = 42 points, evidence grade A) News source 10 + directness 10 + elasticity 2 (interim guidance already confirmed a wider loss) + fundamentals 2 (gearing 74.23%, net interest-bearing debt about RMB 62.5 billion, gross margin −4.03%) + expectation gap 2 + persistence 4 + trading attributes 10 (main board, low first-board position) + risk deduction −8 = 42

Comparison (Shunna Electric 000533 = 35 points, evidence grade C) News source 15 + directness 2 + elasticity 2 + fundamentals 3 + expectation gap 2 + persistence 1 + trading attributes 2 (5 limit-ups in 8 days) + risk deduction −15 (elevated second wave −10, turnover 16.88% −5) = 35

---

5. Detailed analysis of the top 10 stocks

51 国电南瑞 NARI Technology600406[Evidence grade A] — the direction the document actually points to, with valuation not stretched

• Related news: the plan's Chapter 3 title is precisely "Focus on building a new power grid with coordinated main, distribution and micro grids"; distribution grid appears 19 times in the full text and Box 4 lists 8 tasks; distribution-grid hosting of distributed renewables goes 500 → 900 million kW.
• Catalyst logic: direct order benefit. It is the top-share supplier in State Grid's batch tenders, and a scale-up in grid investment transmits to orders through tender share — but note: share fluctuates with each batch and with group allocation, so the transmission is not mechanical. Virtual power plants, demand response, and spot-power-market systems are all products it already sells (in 2025 it delivered next-generation dispatch systems in Shandong/Liaoning/Guizhou, spot power markets in Shanxi/Gansu, and distribution automation in Jiangsu/Beijing).
• Fundamental verification (reconciled item by item against the original annual report and CNINFO filings):
  • Relay protection: across State Grid's first 5 tender batches of 2025, NR Electric led with an 18.1% share, ahead of second-placed Sifang Automation (14.7%).
  • Top-tier financial quality: FY2025 revenue RMB 66.229 billion (+14.53%), net profit attributable to parent RMB 8.279 billion (+8.79%); operating cash flow / net profit ≥1.27x for 4 consecutive years (1.44x in 2025); interest-bearing debt of only RMB 528 million, net cash about RMB 20.5 billion, goodwill RMB 1 million; gearing 40.20% and almost entirely operating liabilities; payout ratio 60.3%, dividend yield about 2.48%.
  • Order backlog RMB 52.031 billion, FY2025 new orders RMB 75.868 billion (+14.40%), contract liabilities +21% YoY.
  • The only State Grid-system name among the stocks covered in this report: State Grid Electric Power Research Institute holds 56.91%.
• Two negatives that must be stated plainly: ① within-grid business grew only 6.12% in FY2025, with all growth coming from outside the grid (+33.09%) and overseas (+84%) — the part the policy points to is exactly its slowest-growing part; ② the shareholder count rose from 77,900 to 302,900 over 5 quarters (3.9×), meaning the float has become retail-heavy and short-term crowding is high.
• Expectation gap: PE (TTM) 24.20 is at the 19% percentile of the past 5 years and the 32% percentile of the past 3 years; the stock is up only 5.2% over 3 years and sits about 20.5% below the one-year high of RMB 31.52. It carries no leader premium — its PE is below Sifang Automation's (37.18) and only slightly above XJ's (22.49).
• Technicals and sentiment: main board; 8/3 +2.96%, turnover value RMB 2.465 billion, turnover rate only 1.24% (the lowest on the list).
• Final judgment: priority deep-dive. The direction the document actually points to + the top tender share + top-tier financials + valuation at a historically low percentile. The cost is elasticity: a 10% move on a RMB 200 billion market cap needs RMB 20 billion of incremental money. It is better used as a confirmation signal for whether the branch is real.

52 许继电气 XJ Electric000400[Evidence grade A] — converter valves underrated, charging piles overrated

• Related news: Box 6 of the plan lists back-to-back DC mutual-support projects (4 approved for construction + 3 held in advance reserve + 3 under research and feasibility study), with inter-provincial mutual support going 80 → 120 million kW; aggregated dispatchable charging capacity from vehicle-grid interaction goes 10 → 50 million kW (+400%).
• Catalyst logic (the two legs differ enormously in hardness):
  • Converter valves = direct order benefit. Checked filing by filing against the original CNINFO texts: as of July 23, 2026, converter valve awards alone reached RMB 2.299 billion, 2.26× the full-year 2025 revenue of the DC transmission segment (RMB 1.018 billion). That segment's gross margin of 32.71% is the company's highest.
  • Charging piles = pure concept read-across (a key correction in this report): ① "charging/swapping equipment and other manufacturing services" at RMB 1.351 billion in 2025 is a mixed line item (it includes processing and manufacturing services), so it is the upper bound of pure charging-pile revenue, at <9.0% of revenue; ② the last time it was disclosed separately (2022) it was only RMB 735 million, 4.92% of revenue, gross margin 8.86%; ③ the business went from RMB 1.336 billion in 2019 to <RMB 1.351 billion in 2025 — five years of zero growth; ④ the plan's charging-facility target is itself conservative — by end-June 2026 the national total already reached 23.057 million units (+43.2%), of which 78.3% are private units installed with vehicles (not XJ's market), and if the current growth rate holds it may be met early; ⑤ this business's gross margin is below the company average, so a rising share actually dilutes the blended gross margin.
• Fundamental verification: six segments — smart transmission and distribution 27.4% (gross margin 29.52%, contributing 34.6% of gross profit), smart meters 26.3%, smart medium voltage 21.0%, DC transmission 6.8% (gross margin 32.71%, contributing 9.5% of gross profit), charging/swapping 9.0% (gross margin 17.06%), renewables integration 9.5%. Net cash about RMB 7 billion; contract liabilities RMB 2.579 billion (+19.0%). Controlling shareholder China Electric Equipment Group at 37.92%; State Grid is the largest customer but not a related party, at 51.89% of sales.
• Biggest risk: 2026Q1 net profit attributable to parent of RMB 111 million, −46.50% YoY, which the company attributes to "gross margin decline caused by falling market prices for some products" (23.47% → 19.01%). This is a price war, not a demand problem.
• Technicals and sentiment: main board; 8/3 +3.68%, but up 20.1% from RMB 19.68 on 7/13, so order expectations have been partly traded already.
• Final judgment: priority deep-dive, but please understand it as a "converter valve name," not a "charging pile name." The bull/bear disagreement is not about charging piles but about whether the ramp in high-margin converter-valve orders can outrun price cuts in meters and distribution.

53 东方电气 Dongfang Electric600875[Evidence grade A] — riding the target with the most positive expectation gap, but the order book is shrinking

• Related news: the plan sets pumped storage at around 160 million kW by 2030, roughly a third above the 2021 Medium- and Long-Term Development Plan for Pumped Storage (2021—2035), which set "total commissioned scale of around 120 million kW" by 2030 (note: the two versions use "total commissioned scale" and "installed capacity" respectively; strict comparability is pending verification).
• Catalyst logic: industry-trend benefit. Hard share figures taken from the original 2024 annual report: "overall pumped-storage market share 41.6%, conventional hydro market share 45%." Physical output corroborates: 2025 hydro generator unit output of 8,056.6 MW, +111.96% YoY. The private-placement-funded "pumped storage manufacturing capability enhancement project" was completed in March 2026 and, at full capacity, adds 19 pumped-storage unit sets per year.
• ⚠️ The most important negative this report found (rarely discussed by the market): year-end order backlog fell from RMB 152.606 billion at end-2024 to RMB 140.31 billion at end-2025, −8.1% YoY; 2026Q1 newly effective order growth collapsed from double digits to +2.34%. New orders are growing while the backlog shrinks, meaning revenue recognition on delivery is now running faster than new order intake — the order book is being consumed.
• The nuclear leg needs demystifying: 2025H1 nuclear revenue was only RMB 2.685 billion, 7.04% of the total, and gross margin fell from 23.35% in 2024H1 to 14.87%, below coal power's 19.13% over the same period — contradicting the consensus that "nuclear is a high-margin business." The 2025 annual report no longer breaks out nuclear revenue separately (segments were reduced from 5 to 3), so this is no reliable data.
• Fundamental verification: FY2025 revenue RMB 78.615 billion (+12.80%), net profit attributable to parent RMB 3.831 billion (+31.11%); but 2026Q1 revenue growth slowed to +5.57% and ex-non-recurring profit rose only +11.49%, while the +37.41% headline net profit includes RMB 325 million of fair-value changes on equities. 2025 operating cash flow / net profit was only 0.53x (−79.98% YoY). PE 21.73, but PB 1.98 sits at the 90.6% percentile of ten years.
• Technicals and sentiment: main board; 8/3 +6.69% on turnover of RMB 2.548 billion (double the RMB 1.283 billion of 7/31). It is still −28.0% below the interim high of RMB 37.19 on 2026-05-08 — this is neither a bottom breakout nor chasing at the highs; it is a policy bounce after a deep correction.
• Final judgment: watch closely. The share is the hardest, but the −8.1% order backlog and yesterday's high-volume surge together argue against chasing. The verification window is the new-order and backlog figures in the interim report in late August.

54 平高电气 Pinggao Electric600312[Evidence grade A] — the most solid orders, but reported revenue has not reflected them and cash flow is deteriorating

• Related news: four State Grid award announcements during 2026 — RMB 1.223 billion on 4/01, RMB 1.327 billion on 5/09, RMB 2.092 billion on 6/12 (16.71% of 2025 revenue), and RMB 1.818 billion on 7/21 (14.53% of 2025 revenue), totalling RMB 6.460 billion ≈ 51.6% of 2025 revenue. In State Grid's second 2026 UHV batch of RMB 9.415 billion, Pinggao took RMB 2.092 billion, 22.2%, the largest single share.
  • ⚠️ Basis warning: all four filings are on a combined "company, subsidiaries and joint ventures" basis, and the joint-venture portion is not consolidated into revenue, so RMB 6.460 billion ≠ RMB 6.460 billion of future consolidated revenue.
• ⚠️ Putting the catalyst back where it belongs: these awards all happened before the plan was released and come from State Grid's annual tender cadence. For Pinggao, this plan "extends the visibility of the order cycle" (a valuation-duration matter), not "a new order event" (an EPS matter).
• Fundamental verification (reconciled against the CNINFO annual report):
  • High-voltage segment RMB 7.747 billion, 61.89% of revenue, gross margin 27.76% (+2.23pct), contributing 71.8% of gross profit; the international segment is 2.06% of revenue with a gross margin of −24.39%, a gross loss for two straight years.
  • An extremely clean balance sheet: total interest-bearing debt of about RMB 75 million, cash of RMB 6.060 billion, net cash of about RMB 5.99 billion ≈ 21.2% of total market cap. (Note: how much of the cash is restricted margin for notes payable is not broken out in the quarterly report — pending verification.)
  • Contract liabilities of RMB 1.824 billion in 2026Q1, +36.6% YoY, independently confirming that orders are improving.
  • Controlling shareholder is China Electric Equipment Group at 41.42%, not State Grid; that group also controls China XD Electric and XJ Electric, so the three compete against each other in the same tender batch, with shares allocated by the group (Pinggao led batch 2 with 22.2%; in batch 3, XD's RMB 4.129 billion overtook Pinggao's RMB 1.818 billion).
• ⚠️ Three negatives that must be stated: ① 2025 revenue grew only +0.93% and 2026Q1 was −2.41% — the reported revenue line still shows nothing of the "big tender year"; ② operating cash flow / net profit fell from 2.94x in 2024 to 0.72x in 2025, and 2026Q1 was RMB −1.035 billion (accounts payable took RMB 1.55 billion less from upstream over the year); ③ PB 2.42 is at the 92% percentile of the past 10 years (while PE is at the 31.9% percentile — the result of ROE recovering from 0.78% to 10.28%, meaning "the earnings recovery is complete" has already been priced in).
• Another overlooked risk: SHFE copper closed at RMB 106,120/tonne on 8/3, +43.8% from end-2024. Bid prices are locked at award, while raw materials are only purchased before delivery — a big tender year + copper at historic highs = gross margin pressure that will become visible in 2027. Whether State Grid tenders include price-adjustment clauses is no reliable data.
• Technicals and sentiment: main board; 8/3 +5.43%, turnover rate 3.24%, no limit-up in the past 10 days; but already +17.5% over the past month.
• Final judgment: watch closely. The order evidence is the most solid, but unrealised revenue + deteriorating cash flow + PB at the 92% percentile stack up. The August 20 interim report is a verification point available within 16 days.

55 四方股份 Sifang Automation601126[Evidence grade B] — No. 2 in relay protection, growing faster than the leader

• Related news: the plan's distribution-grid hosting of distributed renewables 500 → 900 million kW, and "main-distribution-micro coordination."
• Fundamentals: 14.7% share, No. 2 across State Grid's first 5 relay-protection tender batches of 2025; FY2025 revenue RMB 8.193 billion (+17.87%), net profit attributable to parent RMB 829 million (+15.84%); gross margin 30.22%, ROE 17.75%, both better than NARI Technology; 2026Q1 revenue +18.05%, net profit attributable to parent +12.08%.
• Risk: PE 37.18, PB 6.98, the most expensive valuation in the distribution-grid chain; gearing of 59.82% is above peers.
• Technicals and sentiment: main board; 8/3 +0.84%, turnover rate 2.29%, has not started moving.
• Final judgment: watch closely. The "fastest-growing but most expensive" name in the distribution-grid chain.

56 中国西电 China XD Electric601179[Evidence grade A] — the largest award in the industry, but the widest gap between valuation and ROE

• Related news: the third batch award on 7/27 of RMB 4.129 billion (UHV RMB 2.632 billion + transmission & transformation RMB 1.497 billion), the highest in the industry.
• Fundamental verification: FY2025 revenue RMB 23.813 billion (+6.88%), net profit attributable to parent RMB 1.270 billion (+20.45%); gross margin up four periods in a row from 16.43% to 23.35%, cash flow 2.16x, net cash about RMB 8.3 billion, contract liabilities +29.3% YoY.
• ⚠️ The gap between valuation and profitability is the widest on the list:
  • ROE of 5.64% is the lowest among Pinggao (10.28%), XJ (9.94%), Sieyuan (22.63%) and TBEA (8.75%);
  • yet PE 53.96 is the highest of the five; PB 3.05 is at the 94.4% percentile of 8 years;
  • PB/ROE = 54.1, the highest among peers (TBEA 17.6, XJ 20.1, Pinggao 23.9, Sieyuan 33.0);
  • +114.0% over the past year, +52.9% YTD.
• ⚠️ A counter-intuitive falsification point: the "power electronic devices" segment that carries the converter valves (16.7% of revenue) has a gross margin of only 16.49%, the lowest of the three main businesses. That means a volume ramp in Box 6's back-to-back DC projects dilutes rather than lifts China XD Electric's gross margin.
• Final judgment: watch only. For the same order logic, Pinggao Electric's valuation is more than half cheaper. The fundamentals genuinely are improving, but there is a clear gap between the pace of improvement (revenue +5%, ROE +0.67pct a year) and what the valuation implies (3× PB requires ROE reaching 12-15%).

57 通威股份 Tongwei600438[Evidence grade A] — a futures limit-up ≠ earnings delivery, and it has already been falsified once

• Related news: the polysilicon futures front-month contract PS2609 hit limit-up at RMB 35,890/tonne (+8.99%).
• ⚠️ Five independent pieces of evidence that this logic does not hold right now:
  1. The interim guidance has already locked in the answer. On 2026-07-15 the company issued earnings guidance: an H1 net loss attributable to parent of RMB 4.80–5.40 billion, guidance type "continued loss," against RMB −4.955 billion in the same period last year — a midpoint of RMB −5.1 billion, about 2.9% wider YoY, not narrowing; Q1 already lost RMB 2.444 billion, so the Q2 single-quarter midpoint is about RMB −2.66 billion, about 9% wider QoQ.
  2. The futures limit-up came on falling open interest. Main-contract positions of 103,572 lots were down 9.7% from the prior day's 114,761 — driven by short covering, not new long positions.
  3. Futures and spot diverged sharply. On 8/3 futures rose +8.99% while spot N-type recharge material rose only +1.9% to RMB 32,500/tonne — the futures gain was 4.7× the spot move.
  4. The cost line is still far away. Backed out from segment data, 2025 all-in unit cost for polysilicon was on the order of RMB 44,000/tonne (the segment covers multiple businesses, so the back-out has a large error margin and is only an order-of-magnitude reference); spot at RMB 32,500/tonne is roughly equal to the estimated cash cost and still clearly below the estimated all-in cost.
  5. We just watched this play out 11 months ago. In December 2025 spot genuinely reached RMB 52,000/tonne — 45% above today's futures price, and a real spot move — and the company still lost RMB 4.8–5.4 billion in H1 2026, with the stock falling from RMB 27.75 to RMB 10.32.
• Order-of-magnitude test: even if the entire futures-spot spread (RMB 3,390/tonne) were realised, on 2025 sales volume of 384,800 tonnes that is an annualised gross profit improvement of about RMB 1.3 billion — less than a quarter of the half-year loss.
• And it cuts both ways for Tongwei: 2025 cell shipments were 103 GW and module shipments 43 GW (47% of revenue), and this part needs to buy in wafers, so higher polysilicon prices push up its costs.
• Financial risk: gearing 74.23%, net interest-bearing debt about RMB 62.5 billion; 2026Q1 gross margin −4.03%, operating cash flow RMB −2.656 billion; BVPS attributable to parent fell from RMB 10.33 to RMB 7.77 (−24.8%) in a year and a quarter. The feed business contributed RMB 2.771 billion of gross profit in 2025 and is the only tourniquet.
• In fairness, the other side: the substantive difference between this round and the last is the addition of three mandatory national standards, a legally binding hard constraint — but the effective date is 2027-01-01, and estimates of the capacity elimination range anywhere from 20% to 60%.
• Final judgment: watch only. The branch is real and the stock's first-board position is low, but "the interim guidance has already written out the answer" carries more weight than one futures limit-up.

58 特变电工 TBEA600089[Evidence grade A] — neither main line is pure; the real profit base is coal

• Related news: the plan (transformers) + the polysilicon futures limit-up.
• ⚠️ The verification does not support a "dual main-line resonance":
  • Of 2025 net profit attributable to parent of RMB 5.954 billion, coal subsidiary Tianchi Energy contributes roughly RMB 2.91 billion on a crude stake-proportion basis (net profit RMB 3.393 billion × 85.78% stake), about 49%. ⚠️This figure is a simple stake-proportion conversion compared against consolidated net profit attributable to parent, without adjustments for intra-group elimination, minority interest, or parent-company expense allocation; it is an order-of-magnitude estimate, not a disclosed figure.
  • The entire renewables segment (polysilicon + EPC) had 2025 revenue of RMB 13.555 billion but gross profit of only RMB 80 million, 0.44% of company gross profit.
  • Elasticity calculation: every RMB 1/kg rise in polysilicon adds about RMB 53 million to TBEA's net profit attributable to parent ≈ +0.85% of TTM net profit; a 10% rise is about +2.7%. By contrast, every ±10% move in coal prices moves net profit attributable to parent by about ±20% — coal's profit leverage is roughly 7-8× that of polysilicon. ⚠️This coal-price elasticity depends on the realised average coal price, and the company's disclosures contain two versions — RMB 117.42/tonne and RMB 229.2/tonne, nearly 2× apart — so the range of the elasticity conclusion is pending verification.
  • A reverse hedge: rising polysilicon prices = higher costs for its renewables power-station EPC segment, so the same policy makes money in one hand and loses it in the other within the same body.
  • The market has already priced this: on 8/3 TBEA rose only +2.10%, while Tongwei rose +10.02% and Daqo +8.00%.
• Valuation correction: PE (TTM) 17.51 is cheap because it includes non-recurring items; on an ex-non-recurring basis it is 24.02x. In 2026Q1 ex-non-recurring profit fell −3.77% YoY, so the core business is actually shrinking.
• The grid leg is real: electrical equipment 2025 revenue of RMB 26.760 billion (+19.66%), gross margin 19.81% (+2.23pct) — the only one of the three segments with both volume and price rising; in State Grid's second 2026 transformer package it took the top share at 16.7%.
• Final judgment: watch only. It is a "defensive hybrid asset" — resilient when polysilicon collapses (2026 YTD −3.56% vs Tongwei −37.98%), at the cost of not rallying when polysilicon rebounds.

59 中国电建 PowerChina601669[⚠️Evidence grade C, pre-screening only] — the intersection of the most positive expectation gap × the lowest attention

• Related news: pumped storage at 160 million kW by 2030, roughly a third above the 2021 plan.
• Catalyst logic: industry-trend benefit — pumped-storage plant construction → EPC orders.
• The verifiable part: PE (TTM) 9.11, the lowest on the list; free-float market cap RMB 64.573 billion. On 8/3 it closed at RMB 4.94, +1.02%, turnover value RMB 612 million, turnover rate only 0.95%.
• ⚠️ Confidence limits that must be stated: this report has not completed primary-source financial verification on it. Its status as "the main force in pumped-storage engineering" is currently an industry label rather than data this report has verified — the amount of pumped-storage-related order backlog / new orders and its share of revenue is no reliable data. Engineering has low gross margins and long collection cycles, and pumped storage takes years from approval to commissioning, so earnings realisation is extremely slow.
• Technicals and sentiment: main board; essentially has not moved at all.
• Final judgment: watch closely (treat as grade-C evidence). Its value lies in "the branch with the most positive expectation gap × the stock with the lowest attention," and if it turns from zero attention to high volume today, that is a leading signal that the pumped-storage line is genuinely broadening — but please treat it as an observation indicator, not a name on which due diligence is complete.

510 中国核电 CNNP601985[Evidence grade C] — we got it wrong in yesterday's pre-market list; today we explain why we still do not list it

• An honest accounting is required: yesterday (8/3) the pre-market list ranked CNNP No. 2 on attention; with hindsight it closed −0.33% that day, the only decliner on the list. We are not listing it again today just because the plan mentions nuclear.
• Why it is still not listed: ① 110 million kW is not a new number — the April 2026 China Nuclear Energy Development Report (2026) already proposed the same target; ② that target is already fully covered by the existing pipeline — as of April 2026, operating + under construction + approved totalled 112 units / 125 million kW, so the plan target implies no new approvals; ③ the operating end does not directly benefit from equipment tenders; ④ the market has already voted with its feet — the stock has been nearly flat for the past 10 trading days, and on 8/3, with the nuclear concept drawing RMB 3.343 billion of net inflows (the largest market-wide) and both Jiangsu Shentong and China Nuclear Engineering hitting limit-up, it still closed in the red.
• Fundamentals: nuclear operations leader, PE 22.56, free-float market cap RMB 170.516 billion. On 8/3 it closed at RMB 9.03, −0.33%, turnover rate 1.07%.
• Final judgment: Pass. The fundamentals are flawless, but this news is not incremental for it.

---

6. Pass list

---

7. Intra-branch rankings

7.1 Distribution grid / main-distribution-micro coordination (S grade, branch 1)

• Hardest name: NARI Technology (600406). Most suitable for main-board trading and not stretched: NARI Technology (600406), Sifang Automation (601126).
• Who gets a Pass: Beijing Creative (002350) — PE 382 with no fundamental support.

7.2 Pumped storage (A+ grade, branch 2)

• Note: the pumped-storage concept already drew RMB 1.760 billion of net inflows on 8/3 and rose 2.78%, so part of the first move is out; and the nationwide under-construction + approved pumped-storage scale could not be verified — if it is already close to 160 million kW, this branch's characterisation would need to be lowered.

7.3 UHV / main grid equipment (A grade, branch 3, ⚠️catalyst mismatched)

• Hardest name: Pinggao Electric (600312). Who gets a Pass: Shunna Electric (000533), Changlan Technology (002879).
• ⚠️ The biggest risk for this entire branch is a mismatched rationale: if the market pushes it on the grounds that "the plan is positive for UHV," while "UHV" appears only once in the document and in the safety-supervision chapter, that rationale will not survive scrutiny the next day.

---

8. Opening verification signals for today

Cut-off is 08:10 and the call auction does not start until 09:15, so this section gives only judgment criteria and does not predict auction numbers. Everything below is pattern description and historical experience, not operating guidance.

8.1 Auction signals

• Distribution-grid core names (NARI Technology 600406, XJ Electric 000400, Sifang Automation 601126, Dongfang Electronics 000682): this is the group to watch most closely today. If they gap up more than 3% on volume, the money has understood the document's content; if they stay flat while second- and third-tier UHV names lock limit-up at the open, the market is still trading the headline.
• Pumped-storage chain (Dongfang Electric 600875, PowerChina 601669): Dongfang Electric already surged on volume yesterday, so a gap-up-then-fade must be guarded against; if PowerChina moves from "zero attention" (turnover 0.95%) to a high-volume gap-up, that confirms the branch is broadening.
• Second- and third-tier UHV names (Shunna Electric 000533, Changlan Technology 002879): a locked limit at the open or a gap-up of more than 7% is a typical pattern of concentrated liquidity release, and historically subsequent volatility increases markedly — these are already an elevated second wave, with turnover of 16-18% yesterday.
• Nuclear (CNNP 601985, China Nuclear Engineering 601611): if they gap up, note that this is sentiment pricing of an old target.

8.2 Sector signals

• ✅ Healthy: 3 or more new faces hitting limit-up in the distribution grid / pumped storage, with NARI Technology's turnover expanding markedly from 1.24% (= institutional money entering).
• ✅ Broadening: names that "did not move yesterday" — PowerChina, Zhefu Holding, Dongfang Electronics, Sifang Automation — start trading on volume.
• ⚠️ Dangerous: limit-ups concentrated in the same batch of second- and third-tier names that already limited up yesterday (Shunna, Changlan, Baili, Leshan Electric Power), with no new core names joining — the classic pattern of a branch topping out.
• ⚠️ Dangerous: Zhongchao Holding (002471) already failed its limit yesterday; if the number of failed limits among second- and third-tier grid names rises noticeably today, willingness to relay is exhausted.

8.3 Single-stock signals

• Follow-through in the top 5: if NARI Technology and Pinggao Electric gap up and then keep rising on volume without breaking the opening price, that is incremental money.
• Quality of the seal: sealing orders on yesterday's grid limit-ups were generally only RMB 30 million–110 million, i.e. thin; only a marked increase in sealing amounts on new faces today would be a qualitative change.
• Large-order net inflows: yesterday's dragon-tiger list showed net buying of RMB 54.93 million in Baili Electric, RMB 45.40 million in Leshan Electric Power, RMB 43.79 million in Shunna Electric, and RMB 22.33 million in Jiusheng Electric — institutional participation is low and hot money dominates. Only large net buying from institutional-only seats today would mark a change in the character of the money.
• Being outcompeted by a stronger stock: if the memory oversold bounce (GigaDevice) trades stronger than power equipment, the money is rotating to a different main line.

8.4 Risk signals

• 🚩 Gap up then dive: grid core names gapping up more than 5% and quickly falling back — historically this pattern corresponds to a one-off digestion of the positive.
• 🚩 Core names not following: NARI Technology, Pinggao Electric and XJ Electric all up less than 1% while second and third tiers keep pushing higher — the classic late-stage structure of a branch.
• 🚩 Lone limit-up: only 1-2 power-equipment names limit up, and all are yesterday's old faces.
• 🚩 Memory sells off again: if GigaDevice (603986) and Demingli (001309) keep falling despite the overnight overseas reversal, the domestically driven clear-out is not over, and the STAR 50 may drag the index again.
• 🚩 Macro headwinds materialise: the 67.2% probability of a September Fed hike and the 57-pip drop in offshore RMB — if the currency weakens further intraday, overall valuations get compressed.
• 🚩 Yesterday's strong themes fade: if the 8/3 consecutive-limit-up ladder (Chuanzhi Education 6 limit-ups, Yiming Food 5, Gaozheng Explosive 4, Leo Group 2 with RMB 7 billion of turnover) fails its limits en masse, the sentiment score will fall quickly.
• 🚩 The "front-running" hypothesis is confirmed: if the distribution-grid core names stay flat all day while the names that already rallied yesterday gap up and cash out, the document's content was digested long ago, and this report's premise that "today is the first pricing day" does not hold.

---

9. Final conclusions

This section is an attention ranking, not a rating and not buy/sell advice. Please read it together with the "evidence grade."

9.1 The 5 main-board stocks most worth watching today (including 1 with pre-screening only — see evidence grade)

9.2 The 3 strongest branches today

9.3 Directions not worth chasing today, and why

1. Anything chased on the basis that this plan is a "big UHV positive" — this is today's biggest pricing trap. "UHV" appears only 1 time in the full text and in the Box 7 safety-supervision chapter, "transformer" and "converter" appear 0 times each, and the main-grid wording is "retrofit and upgrade of existing corridors" and "corridors already in the plan during the 14th Five-Year Plan to be completed as early as possible." The UHV orders are real, but they come from State Grid's annual tenders and were public on July 27. (Also remember the self-imposed limit in 2.1(a): word frequency reflects emphasis, not investment allocation.)
2. Elevated second- and third-tier grid equipment names (Shunna Electric 000533, Changlan Technology 002879, Huadian Liaoning Energy 600396, Leshan Electric Power 600644) — Shunna has "5 limit-ups in 8 days," Changlan "5 consecutive limit-ups, then a limit-down shakeout, then limit-up again," with turnover of 16-18% yesterday, and none of them has disclosed any winning-bid amount.
3. Nuclear (CNNP 601985, China Nuclear Engineering 601611, Jiangsu Shentong 002438) — 110 million kW is not a new number, and it is already fully covered by the 125 million kW of operating + under construction + approved capacity; the nuclear concept drew RMB 3.343 billion of net inflows yesterday, the largest market-wide, so the first move is fully played out.
4. Virtual power plants (Fuling Electric Power 600452 and others) — the 2030 figure of 50 million kW is exactly the same as the March 2025 guiding opinions; and yesterday the concept rose 2.64% while seeing RMB 948 million of net outflows.
5. Solar anti-involution (Tongwei 600438) — Tongwei's July 15 interim guidance already wrote out the answer: an H1 loss of RMB 4.8–5.4 billion, 2.9% wider YoY, with Q2 9% worse QoQ. The futures limit-up came on falling open interest, and spot rose only 1.9%. In December 2025 spot reached RMB 52,000/tonne (45% above today's futures price) and H1 was still a huge loss.
6. Memory oversold bounce (GigaDevice 603986, Demingli 001309) — the overnight overseas reversal is sentiment repair, not a new positive. Yesterday GigaDevice was locked limit-down on RMB 23.033 billion of turnover, and semiconductors saw RMB 9.316 billion of net outflows, the largest market-wide. Also note: Shengyi Technology (600183) is copper-clad laminates / electronic materials, not memory — do not associate it as a memory name.
7. Oil & gas E&P (CNOOC 600938, COSL 601808) — WTI −7% plus a possible Hormuz reopening; directionally negative, and this report offers no positive assessment.

9.4 Final one-line judgment

Today is the first trading day after an S-grade policy document was made public, and what that document says is very likely not the same thing as the label it is about to be given — pull the full text and count it out: "distribution grid" 19 times, "UHV" 1 time (and in the safety-supervision chapter), "transformer" and "converter" 0 times each, which shows the document's emphasis lies on the distribution-grid side (though that only speaks to narrative weighting, not to investment allocation). So what deserves attention today is not the misread second- and third-tier UHV names but the distribution grid the document actually emphasises (NARI Technology, XJ Electric) and pumped storage (Dongfang Electric); as for nuclear and virtual power plants, the two numbers that will be treated as positives, checking them item by item against base years and prior figures shows they are respectively an old number and a restatement of an old target. One more reminder: the document is dated July 21, and nuclear and pumped-storage flows already moved noticeably during yesterday's session — "today is the first pricing day" is this report's assumption, not a fact, so please verify it against today's actual follow-through.

---

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