A-Share Pre-Market Brief | 2026-08-21 Friday

0. Today in One Sentence

The strongest catalyst is a policy-grade one: on 8/20 at 14:59 (one minute before the close), six Shanghai departments jointly issued the "Notice on Optimizing the City's Real Estate Policy Measures" (the "Shanghai Eight Measures", 5 aspects and 8 policy items), effective today, 8/21. But this brief must simultaneously deliver an equally important and counter-intuitive conclusion: this policy and Shanghai's local property stocks are mismatched — the three demand-side tools of the Shanghai Eight Measures (second homes outside the Outer Ring with down payments cut to 15%, housing-provident-fund withdrawals allowed to pay down payments extended to completed new homes, and trade-in subsidies of up to RMB 80,000 against a total funding pool of only RMB 200 million) target rigid-demand and trade-up buyers outside the Outer Ring; whereas 城投控股 Shanghai Chengtou Holding (600649) contracted at an average price of about RMB 126,000/sqm in H1 and 陆家嘴 Lujiazui (600663) at about RMB 168,000/sqm for residential, with the bulk of saleable value in central-district luxury housing — they are not in the same market as the policy tools. More critically, all three main names on this chain have already guided their 2026 interim results to losses back in July (城投控股 (600649) −50 million to −75 million, a first-ever loss; 中华企业 China Enterprise (600675) −212 million to −352 million, a first-ever loss; 光明地产 Bright Real Estate (600708) −400 million to −600 million, a widening loss).

Hence today's branch ranking: catalyst strength points to the Shanghai Eight Measures, but evidence strength points to precious metals. The only branch that combines all three of "a fresh overnight driver + interim results already disclosed and sharply profitable + business purity verified against a primary source" is silver: the US Treasury doubled the per-operation size of its long-bond buybacks to $4 billion, and on the 8/20 overnight session SLV +2.75% materially outperformed GLD +0.34%; 盛达资源 Shengda Resources (000603) disclosed its interim report on 8/1 with net profit attributable to parent +456.46%, 100% self-mined ore, silver accounting for 71.51% of mining-and-processing profit, and a mining-and-processing segment gross margin of 78.37%.

Likely direction of flows: from yesterday's single track (the pharma chain contributed 37 of the 79 limit-ups) toward the low-priced property chain that benefits from policy — but the property chain can only be traded as a sentiment line. The characterization of yesterday's main line must be precise — it is not that "the catalyst has been falsified", but that "a one-off repricing was completed within a single US session and the next day enters decay": Moderna fell −23.55% on 8/20, yet is still +111.7% versus 62.96 on 8/18 before the catalyst was released. What actually constitutes a negative today is three other things: ① the A-share market only begins pricing this on day 1 after the US repricing was already complete; ② last night the A-share vaccine names' own interim results were losses; ③ the VAT rate on vaccine products has already been raised from 3% to 13%, an industry-wide gross-margin suppressor that was entirely unpriced in yesterday's rally narrative.

Pre-market state call: yesterday's main line most likely gaps up and fades, or splits outright; today is a day of "rotation into a new direction", not "continuation".

---

1. News Overview

Scan window: 2026-08-20 15:00 (previous session close) → 2026-08-21 07:30 (time this brief was generated). Every item is tagged with its publication time; cross-market data is tagged with its read timestamp.

⚠️ Carry-over information that falls outside this window but must be read alongside it today (see §2 Branch One): the H1 pre-announcements of the three main Shanghai property names were released on 7/11 (城投控股 600649, first-ever loss), 7/15 (中华企业 600675, first-ever loss; 上实发展 SIIC Development 600748, narrower loss; 光明地产 600708, widening loss) and 7/14 (志邦家居 Zbom Home 603801, first-ever loss) respectively. These are not new news, but they determine whether today's policy line can be called "earnings delivery".

---

2. Strongest Positive Branches, Ranked

Branch One: Shanghai Real Estate, the "Shanghai Eight Measures" — Strongest Catalyst, But It Must Be Rated Separately From Fundamentals

Policy content (six departments: Municipal Housing and Urban-Rural Construction Management Commission, Municipal Housing Administration, Municipal Finance Bureau, PBoC Shanghai Branch, Shanghai Financial Regulatory Bureau, Municipal Housing Provident Fund Center; issued 2026-08-20, effective 2026-08-21. All clauses below are quoted from the original text on the Shanghai Municipal Housing Administration website¹⁸):

1. Provident fund: "A contributor purchasing newly built completed commodity housing in this city may apply to withdraw the housing provident fund of the contributor and spouse to pay the down payment"; "one withdrawal per year for 5 years after obtaining the property ownership certificate, with total withdrawals not exceeding the purchase amount paid with own funds".
2. Commercial mortgage down payment: "For the purchase of housing outside the Outer Ring, the minimum down-payment ratio for second-home commercial personal housing loans is adjusted to no less than 15%".

   ⚠️ Accounting-basis note: the only operative change in this notice is the "second home outside the Outer Ring" tier. The "first homes uniformly no less than 15%, second homes inside the Outer Ring no less than 25%" reported in media summaries describes the state of affairs after implementation, most of which is pre-existing (Shanghai already issued a round of "Shanghai Seven Measures" on 2026-02-25) — it is not new in this round. Listing the three tiers side by side as a "three-pronged push" overstates the marginal force of this policy.

3. "Trade-in" home-purchase subsidy: for buying newly built commodity housing, "calculated as 1% of the total loan amount for the newly built commodity housing purchased", capped at RMB 50,000 per unit; for buying second-hand housing inside the Outer Ring, "a subsidy of RMB 30,000 per unit". Total funding of RMB 200 million, "first come first served until exhausted", running through March 31, 2027.
4. Housing-voucher resettlement: "Add housing-voucher resettlement as a method in expropriation and demolition compensation and resettlement, with priority focus on urban-village redevelopment and old-city renovation projects".
5. Acquisition of second-hand housing for stock: "Each central district shall advance the acquisition of second-hand housing and accelerate conversion to rental use, effectively increasing the supply of affordable rental housing in central districts".

① The Sector Barely Reacted Yesterday, But the Reason Is Not "Too Late" — It Is That "The Limit-Ups Happened Before the Policy"

The policy was released at 14:59, one minute from the close. At the 8/20 close, Real Estate Development (BK0451) was up only +2.05%, with net main-force inflow of RMB 130 million and 80 advancers vs 10 decliners; Shanghai local names generally rose only 0.5%–2.7% (陆家嘴 600663 +1.27%, 浦东金桥 Pudong Jinqiao 600639 +0.74%, 外高桥 Waigaoqiao 600648 +1.56%, 上实发展 600748 +2.37%, 光明地产 600708 +2.67%, 中华企业 600675 +2.68%, 市北高新 Shibei Hi-Tech 600604 +2.70%).

But after a minute-by-minute check of the 8/20 intraday tape, one common ex-post attribution must be corrected:

Both limit-up names sealed far earlier than 14:59. Explaining the 8/20 limit-ups with "the Shanghai Eight Measures clauses" is ex-post attribution — the 8/20 price reaction was to expectations/rumors, not to the published text. Note also that 城投控股 (600649)'s seal that day was not solid: it slipped back to 4.12 at 13:45 and to 4.11–4.12 between 14:15–14:30, only re-sealing into the close.

The implication for today runs both ways: on one hand, the policy text itself genuinely has not been priced (which supports "there is still room today"); on the other hand, the leaders limit-up before the text came out, meaning informed money is already in, so today is more likely to be a cash-out than a launch. The dividing line between the two scenarios is at the auction (see §8-1).

② The Financials of This Chain: Three Main Names Already Guided to Losses in July

This is the most important self-correction in this brief. The first draft described the uncertainty at 城投控股 (600649) and 中华企业 (600675) as "interim reports not yet disclosed" — wrong: both companies' H1 pre-announcements were released back in July, and both are first-ever losses.

③ Saleable-Value Mismatch: Of the Five Tools, Only Two Reach These Companies (The Single Most Important Judgment in This Brief)

They must be looked at separately — the five measures of the Shanghai Eight Measures point at two completely different markets:

In other words: what is mismatched is the three "demand stimulus" items (①②③); what matches is the two "activating existing stock" items (④⑤) — and only 城投控股 can catch those latter two. That is one reason this brief still keeps 城投控股 in a recommendation slot while downgrading 中华企业 (600675) to watch only — the two have similar Shanghai purity (98.56% vs 98.24%), but 中华企业 has no housing-voucher track record and no affordable-rental operating platform.

The average-price mismatch below applies only to the three demand-side tools ①②③:

For a RMB 20 million home, a down payment 5 percentage points lower means RMB 1 million less cash out, which is meaningful; but "up to RMB 80,000 of subsidy outside the Outer Ring" is entirely inapplicable to Inner Ring luxury, and paying a down payment with provident fund is all but irrelevant to buyers at that price point. As far as the three demand-side tools go, there is not a single name on this chain whose saleable value genuinely sits in the rigid-demand market outside the Outer Ring; what they can catch is the two stock-activation items (④⑤), and those generate no incremental sales revenue — they only change the sales channel and the rental supply. This is precisely where the line falls between "policy sentiment trading" and "earnings delivery".

④ Tiering (Strictly Applying "Concept ≠ Beneficiary")

Branch Two: Precious Metals — The Driver Is in Silver, the Criterion Is "Self-Mined vs Smelting"

Driver: after US public debt broke $40 trillion and the 30-year yield hit levels not seen since 2007, the Treasury announced on 8/19 that it would double the per-operation cap on liquidity-support buybacks in the 10–20-year and 20–30-year buckets from $2 billion to at least $4 billion (effective 9/9–11/4); at 23:07 on 8/20 Treasury Secretary Bessent said "the size could exceed the announced $4 billion".

8/20 US session: SLV +2.75%, GDX +2.59%, SIL +1.66%, while GLD gained only +0.34%. Silver led gold by about 1.57pp; "silver stronger than gold" holds.

Yesterday's recap verification point 7, "is COMEX gold above 4,542.50 at 09:00 the next day" → yes (4,587.50), passed, so the branch is not downgraded. Verification point 8, "does 赤峰黄金 Chifeng Gold (600988) outperform 山东黄金 Shandong Gold (600547)", was already judged wrong yesterday, so the "pick the cleanest structure" logic will not be used again in this round; this brief switches to the cost-behavior criterion of "own mines vs smelting".

Three pieces of counter-evidence that must be presented alongside:

1. The bond-market driver was already given back the same day: in the 8/20 US session the 10-year Treasury yield rose to 4.708% and the 30-year to 5.25%, both up versus the prior day. What is driving gold and silver is no longer "falling rates" but the credit concern implied by "the Treasury needing to step in and support the market".
2. Silver is rebounding inside a deep drawdown, not breaking out: SLV is −41.6% from its 1-year high on 2026-01-28 (GLD only −16.3%), and on a 3-month view silver still lags gold by about 10 percentage points. High beta cuts both ways.
3. The interim-report growth is built on a silver price that no longer exists: SLV's 2026H1 average was 71.10, whereas Q3 to date (7/1–8/20) has averaged only 54.77, 23.0% below H1; the current price of 61.66 is still 13.3% below the H1 average. (Cross-check: 盛达资源 (000603)'s interim report states "the domestic average silver price rose +147.06% YoY", which is mutually consistent with SLV's H1 average of +138.3% YoY within the range of FX and the domestic-overseas price spread.) Inference (a judgment, not a fact): even if silver rises another 10%, a pure self-mined silver producer's H2 realized average price will still be below H1 — the YoY looks great, the QoQ is a headwind.

Elasticity quantification (self-mined costs are rigid; smelting/purchased costs move with the price):

Branch Three: Broker Share-Swap Merger — A Computable Discount Plus a Defined Timetable

The plan (2026-05-19 draft): 中金公司 CICC (601995) swap price RMB 36.68, 东兴证券 (601198) RMB 16.05 (ratio 1:0.4376), 信达证券 (601059) RMB 19.11 (1:0.5210).

8/27 is a hard date, and the two names showed almost zero reaction on 8/20. Risks: if the review fails or the plan is revised, the discount widens in reverse; these are not limit-up-type names.

---

3. Master Ranking of Single-Name Positive Strength (Descending by Stock)

Board and price-limit rules: SSE/SZSE main board (60/00) ±10%, ChiNext (30) ±20%, STAR Market (68) ±20%, BSE (8/4) ±30%. Limit-up/limit-down is always judged against the applicable limit. "Open→close %" = the executable return relative to that day's opening price, shown alongside the close-to-close change — the real execution price for a pre-market list is the opening price, and looking only at the closing change systematically overstates results. ⚠️ Valuation-basis warning: the "PE" shown on quote screens is dynamic PE (latest single period annualized), not TTM. Example: 城投控股 (600649) shows 356.63, which is the product of "2026Q1 net profit attributable to parent of RMB 7.266 million × 4 ÷ market cap"; its true PE (TTM) is about 38x, and about 141x once the guided H1 loss is plugged in. Every PE cited in this table is tagged with its basis.

---

4. Single-Name Scoring Model (95-point scale + risk deductions)

Component scores for representative names:

Two paired comparisons illustrate the two core criteria of this brief: ① both carry the "silver" label, yet 盛达资源 (000603) scores 78 pts vs 湖南白银 (002716) at 25 pts, with almost the entire gap coming from "earnings elasticity" and "risk deduction" — i.e. self-mined ore vs smelting/processing. ② both are Shanghai Eight Measures names with 98%+ Shanghai purity, yet 城投控股 (600649) scores 70 pts vs 中华企业 (600675) at 52 pts, with the entire gap coming from contract liabilities (RMB 8.622 billion vs RMB 108 million) — i.e. whether there is a reservoir left to recognize.

---

5. Detailed Analysis of the Top 10 Names

1盛达资源00060378 pts — Priority deep-dive · SZSE main ±10%

• Related news: silver from 66.73 → 68.44 (+2.56%); 8/20 US session SLV +2.75% vs GLD +0.34%; SHFE silver up over 4% intraday in the night session.
• Positive logic: the purest play on silver-price elasticity in the entire A-share market. The company's interim report states verbatim: "revenue from mined silver metal accounts for 70.38% of non-ferrous mining-and-processing revenue and 62.14% of operating revenue, and the profit generated by sales of mined silver metal accounts for 71.51% of the company's non-ferrous mining-and-processing profit"; its production model is that "the main raw material for production at the company's operating mines is self-mined ore" — 100% self-mined, costs rigid, the highest-elasticity configuration there is.

  Note: you cannot judge this off the product line's "silver concentrate 34.55%" — its products are named "lead concentrate (silver-bearing)" and "zinc concentrate (silver-bearing)", so the silver metal is dispersed inside the lead and zinc concentrates; the true silver revenue share is 62.14%, nearly twice the product-line reading.

• Fundamental verification (the only name in a recommendation slot in this brief whose interim report is already disclosed and sharply profitable): 2026 interim (disclosed 8/1): revenue RMB 1.152 billion +27.08%, net profit attributable to parent RMB 390 million +456.46%, ex-non-recurring RMB 382 million +417.06%, blended gross margin 69.15%, mining-and-processing segment gross margin 78.37% (highest in the field), liability ratio 48.36%, ROE 12.14%. Retained silver resources are about 7,800 tonnes, and Jinshan Mining's Erentaolegai is "one of the largest standalone large silver mines by reserves in China".
• Two operating characteristics that must be counted in: ① seasonality — "operating mines are concentrated in central-eastern Inner Mongolia and, due to the climate, there is essentially no output in the first quarter", so H1×2 understates the full year; ② sales timing — for silver and gold ingots the company "determines the specific timing of sales in light of changes in the macro environment and future expectations", i.e. it deliberately holds inventory, creating an unrealized inventory-value two-way option (this also explains why silver rose +147% while revenue rose only +27%).
• Technical sentiment: closed 8/20 at 33.28, +6.36%, but it opened at 33.32, so open→close was −0.12% — 100% of the gain came from the auction gap, and executing at the open earned nothing; main force +605 million, turnover rate 14.28%.
• Final call: its fundamentals are the hardest in this brief and its overnight driver the freshest. But it already demonstrated once yesterday how a gap can eat the entire return — the key today is not whether to pick it but whether it gaps up: an auction gap of ≥ +5% is by itself a reason to pass. Also factor in the H2 QoQ headwind from §2.

2城投控股60064970 pts — Watch closely (pure policy-sentiment trade) · SSE main ±10%

• Related news: the "Shanghai Eight Measures" take effect today (original text on the Shanghai Municipal Housing Administration website¹⁸, Chinanews¹). It is the only company on this chain that can simultaneously catch policy item ④ (housing-voucher resettlement, focused on urban-village redevelopment and old-city renovation) and item ⑤ (acquiring second-hand homes and converting them to affordable rental housing) — and those two are precisely the stock-activation tools that generate no incremental sales revenue.
• Business characterization (this definitively rules out one common misconception): its registered "business scope" still lists raw-water supply, sewage treatment and so on (a historical legacy) — but the environmental assets were spun off back in 2015–2016 via a share-swap absorption merger with Yangchen B-share and separately listed as 上海环境 Shanghai Environment Group (601200). Its 2025 annual report describes it verbatim as "a comprehensive real estate group whose main businesses are property development, operation and investment", with real estate at 99.78% of revenue and Shanghai at 98.56%. Its controlling shareholder is Shanghai Chengtou (Group) Co., Ltd.
• Degree of policy-mechanism match (this is its only substantive edge over peers): the 2025 annual report states verbatim that "the Shuishang Huating project successfully practiced a localized relocation housing-voucher-linked sales model" — the only company in the group with an actual housing-voucher sales track record, mapping directly to item ④ of the Shanghai Eight Measures. Amount not disclosed. It also runs the affordable-rental platform "Chengtou Kuanting" with 16 projects in operation, 17,000 units and annual rent over RMB 600 million (about 4% of revenue), plus two urban-village projects outside the Outer Ring, Duxing and Zhuzhai, at the re-planning / pre-strategy stage. The word "acquisition" appears only once in the annual report, as industry background; the company discloses no business of its own in acquisitions.
• Fundamentals (these must be given in full, not just the policy narrative):
  • ⚠️ On 2026-07-11 it already filed H1 guidance: net profit attributable to parent of −50 million ~ −75 million and ex-non-recurring of −40 million ~ −60 million, its first-ever loss since listing (prior-year period net profit attributable to parent +RMB 153.27 million). Q2 alone is implied at −RMB 69.8 million, worse than the H1 cumulative figure. Interim report on 8/31.
  • 2026Q1 revenue RMB 853 million (−63.55%), net profit attributable to parent RMB 7.27 million, operating cash flow −RMB 1.077 billion, liability ratio 72.80%.
  • Of the RMB 289.3 million of 2025 net profit attributable to parent, RMB 260.8 million was investment income (mainly from its stake in 西部证券 002673) — about 90%. This means its income statement may be more sensitive to broker share prices than to the Shanghai property market; the H1 guidance itself notes "affected by securities-market volatility, the fair value of held equities, funds and other financial assets declined".
  • Finance costs eat the gross profit: RMB 871 million / RMB 2.519 billion of gross profit = 34.6% in 2025; RMB 205.6 million / RMB 321 million = 64.0% in 2026Q1. Net interest-bearing debt of about RMB 35.294 billion / equity attributable to parent of RMB 21.053 billion = net gearing of 168%.
  • Forward indicators are weakening across the board: 2026Q1 contracted floor area −69.13%, value −73.45%; Q2 −17.02% / +10.68%; H1 totals 14,467 sqm / RMB 1.816 billion, with floor area −46% and value −41% YoY; zero new construction starts in 2026Q2; contract liabilities fell from RMB 11.644 billion at end-2024 to RMB 8.622 billion in 2026Q1.
  • Saleable-value mismatch: H1 average contracted price about RMB 126,000/sqm (on sale: Luxiangyuan Phase II in Huangpu's old town, Jingyunli Phase II villas in Qingpu, etc.). The RMB 80,000 subsidy outside the Outer Ring is entirely inapplicable to Inner Ring luxury, and paying a down payment with provident fund is all but irrelevant to buyers at that price point.
• Valuation-basis correction: the "PE 356.63" shown on quote screens is a dynamic PE (2026Q1 net profit attributable to parent of RMB 7.266 million × 4 ÷ market cap), not TTM; the true PE (TTM) is about 38x, and about 141x once the guided H1 loss is plugged in. PB 0.49 does not have this problem (it uses the latest reporting period's net assets per share and does not lag).
• Technical sentiment: on 8/20 it opened at 3.76 (−0.53%), first touched the limit-up price of 4.16 only at 13:01 (i.e. the seal happened before the 14:59 policy release), and pulled back twice intraday (to 4.12 at 13:45, to 4.11–4.12 between 14:15–14:30) before re-sealing into the close; turnover rate 5.22%, main force +46 million; Dragon-Tiger buys of RMB 138.4 million / sells of RMB 58.0 million, list net buy +RMB 80 million (36.99% of the day's turnover), read as "1 institution buying". Market cap RMB 10.4 billion.
• Final call: it is the most recognizable property name today — 98.56% Shanghai purity, the only one with a housing-voucher track record, the only one closing at the limit-up price, and with institutions on the Dragon-Tiger list. But it has already guided H1 to a first-ever loss, contracted floor area has halved, Q2 had zero new starts, and its saleable value and the policy tools are not in the same market. So this brief assigns "watch closely" rather than "priority deep-dive", and explicitly characterizes it as a pure policy-sentiment trade: it can be traded on the sentiment line, but do not use "earnings delivery" as a reason to size up.

3兴业银锡00042669 pts — Watch closely · SZSE main ±10%

• Positive logic: 100% self-mined ore. FY2025 by product (mining is 99.64%, all of it "mined X"): mined silver is 39.17% of revenue, 39.93% of gross profit, at a 58.71% gross margin; mined tin is 29.70% of revenue, 34.98% of gross profit, at a 67.84% gross margin — silver and tin together are about seventy percent of gross profit, all self-mined, and its absolute silver elasticity is the largest in this group.
• Two demerits (which must be presented alongside the growth rate):
  1. Non-recurring items: Q1 net profit attributable to parent was RMB 1.338 billion while ex-non-recurring was only RMB 891 million (a gap of RMB 447 million, 33%); the H1 guidance splits the same way — net profit attributable to parent +168.95197.86% but ex-non-recurring only +110.87139.57%. The anomalous combination of a 64.63% net margin above a 69.07% gross margin can only come from large non-operating gains. Whenever "+257%" or "+197%" is cited, the ex-non-recurring basis must be given alongside.
  2. Capital allocation: on 8/19 it filed to acquire control of *ST 威领 and launch a partial tender offer, for a total outlay of up to RMB 1.8 billion; the target's net assets attributable to parent at end-2025 were −RMB 19.8891 million (negative), it is already under delisting-risk warning, and its main business is tungsten/tin/lead/zinc plus lepidolite/lithium carbonate — no silver. That money points at tin, tungsten and lithium, diluting rather than reinforcing the silver narrative.
• Technical sentiment: closed 8/20 at 38.56, +7.23%, opened at 38.28, open→close +0.73%; main force +574 million. Market cap RMB 68.469 billion. H1 disclosure on 8/29.
• Final call: silver elasticity is tied for first with Shengda (incremental gross profit of about 0.62% per unit of market cap), but non-recurring items are a third of profit and RMB 1.8 billion is going into a negative-net-asset target; those two together place it behind Shengda.

4东兴证券60119868 pts — Watch closely · SSE main ±10%

Filed at 19:20 on 8/20: the CICC absorption merger goes to the review committee on August 27. The 1:0.4376 swap ratio is locked, so at CICC's 8/20 close of RMB 34.06, each Dongxing share converts to RMB 14.905, a 10.08% discount to its close of 13.54. It closed flat on 8/20 (0.00%) with 0.49% turnover — the market gave this date zero reaction. The merged company's 2025 revenue was RMB 37.2 billion, net capital RMB 103.3 billion, and it had 441 branch outlets. Risks: if the review fails the discount widens in reverse; this is not a limit-up-type name and single-day elasticity is limited.

5山东黄金60054766 pts — Watch closely (upgraded in this brief) · SSE main ±10%

• One misreading that must be corrected: the consolidated gross margin is only 20.46%, and concluding from that that "earnings quality is poor" is wrong. Breaking FY2025 out by product: self-mined gold is 36.82% of revenue but 91.00% of gross profit, at a 50.57% gross margin; purchased gold is 41.42% of revenue at a gross margin of just 0.80%, small gold bars 2.28%, trading 1.38%. It is purchased gold / small bars / trading (together about 61% of revenue but only 3.4% of gross profit) that inflate revenue and dilute the consolidated gross margin. On a gross-profit basis, it is genuinely a self-mined gold miner.
• Its elasticity to the gold price comes only from that 36.8% of revenue in self-mined gold; purchased gold works the same way as 湖南白银 (002716)'s smelting. Silver exposure is zero.
• Technical sentiment: on 8/20 +8.59%, the strongest on the whole chain, opening at 34.23, open→close +3.42%, also the strongest in the field. 2026Q1 revenue RMB 32.516 billion +25.37%, net profit attributable to parent RMB 1.446 billion +40.87%, operating cash flow per share +1.32.
• Information gap: one of only two of the 7 precious-metals names without interim guidance, with H1 disclosure on 8/29 — the only one of the three pure gold names that still carries an interim information gap.
• Final call: strongest tape yesterday + self-mined gold at 91% of gross profit + the only remaining information gap. But with the driver on the silver side it is not on the core beneficiary chain, so it is listed as watch closely rather than in a recommendation slot.

6陆家嘴60066364 pts — Watch closely (downgraded in this brief) · SSE main ±10%

• The interim report is real and was actually disclosed on 2026-07-30: revenue RMB 8.098 billion +22.75%, net profit attributable to parent RMB 1.108 billion +35.91%, ex-non-recurring RMB 1.113 billion, operating cash flow +RMB 5.407 billion, gross margin 57.92%.
• But +35.9% should not be treated as a sustainable operating growth rate — three pieces of evidence point to a one-off asset disposal:
  1. the "real estate (including investment property) sales" segment's gross margin was 79.33% (2026H1) vs 30.09% (2025 annual report), a 49-percentage-point jump;
  2. the management discussion states explicitly that H1 office-property sales cash inflow of RMB 3.763 billion was "collections from the sale of the DBS Bank Tower"; investment property is depreciated under the cost model, so on disposal the carrying cost is far below market price, producing an extremely high accounting gross margin;
  3. minority interests of RMB 24.683 billion (49.2% of total equity), while 2026H1 minority profit was only RMB 13.7 million (versus RMB 362.3 million in the 2025 annual report) — indicating that H1 profit came almost entirely from wholly-owned entities, with near-zero contribution from JV project companies. True earning power excluding that transaction cannot be quantified (no breakdown disclosed); marked "to be verified".
• Its benefit from the Shanghai Eight Measures is the lowest in the group: H1 residential contracted sales of 34,300 sqm (−16%) / RMB 5.778 billion (+21%), an average price of about RMB 168,000/sqm (Qiantan and Lujiazui ultra-luxury), more than twice Shanghai's market-priced average of RMB 81,000/sqm. Sales floor area is already −16%, and the +21% in value comes entirely from a mix shift to higher unit prices — it sells to precisely the buyers not constrained by down-payment policy.
• The investment portfolio is weakening: H1 office leasing cash inflow of RMB 1.265 billion, −5% YoY; average occupancy of Shanghai's mature Grade-A offices is only 74% at RMB 6.15/sqm/day; Tianjin occupancy is 60%.
• Highest leverage in the group: net interest-bearing debt of about RMB 56.605 billion / equity attributable to parent of RMB 25.509 billion = net gearing of 222%.
• Technical sentiment: on 8/20 only +1.27%, turnover value RMB 204 million, turnover rate 0.54%, main force −7 million.
• Final call: its financials can be read without the Shanghai Eight Measures, and the Shanghai Eight Measures will not improve them much. The first draft gave it a recommendation slot on "the interim report's +35.91% is verified"; that is now corrected to watch closely.

7信达证券60105963 pts — Watch closely · SSE main ±10%

Swap ratio 1:0.5210, converted value RMB 17.745, a discount of 8.53% to its 8/20 close of 16.35 — thinner than Dongxing's. 8/20 +0.37%, turnover 0.15%. Same logic as Dongxing, lower priority.

8赤峰黄金60098861 pts — Watch closely (downgraded in this brief) · SSE main ±10%

• The original criterion has been overturned: yesterday's recap and this brief's first draft recommended it as "the only one on the chain still rising after the gap". Computing open→close for 8/20 name by name, 4 were positive — 山东黄金 (600547) at +3.42%, 赤峰黄金 (600988) +2.44%, 紫金矿业 (601899) +0.74%, 兴业银锡 (000426) +0.73%; Chifeng is neither the only one nor the strongest. That criterion is void.
• Fundamentals: FY2025 mined gold was 89.72% of revenue and 100.07% of gross profit — a figure above 100% that by itself shows the non-gold businesses are a negative contribution: electrolytic copper gross margin −30.90%, rare earths −17.82%, electronics dismantling −12.91%; on 8/8 it announced the suspension of its Mengkang rare-earth project in Laos. Silver exposure is effectively zero.
• 2026Q1 net profit attributable to parent RMB 988 million +104.43%, gross margin 59.27%, liability ratio 29.36%. H1 was already guided on 7/15 at RMB 1.70–1.78 billion (+54~61%), with ex-non-recurring essentially equal to net profit attributable to parent — clean quality — and disclosure on 8/31. Its 2026 gold production and sales target is 14.7 tonnes.
• Final call: high quality among the pure gold names, but already guided + zero silver exposure + the original recommendation criterion falsified, so it is downgraded to watch closely.

9紫金矿业60189959 pts — Watch closely · SSE main ±10%

• Qualitative correction: the earlier "copper-led" description is inaccurate on a gross-profit basis. FY2025 segment gross profit: mined gold 43.2% > mined copper 36.5% > mined silver 2.11%. Gold has already slightly overtaken copper; but at only 2.11%, using it to express the "silver" theme is a mismatch.
• On 8/20, net main-force inflow of +RMB 2.081 billion was first in the whole market, +4.55%, open→close +0.74%, turnover value RMB 14.179 billion. 2026Q1 net profit attributable to parent RMB 20.079 billion +97.50%; PE (TTM) 14.76 and PB 4.87 are both the lowest in this group. H1 was guided on 7/10 (net profit attributable to parent RMB 39.1 billion, ex-non-recurring RMB 37.9 billion) and will be disclosed on 8/22.
• It also carries RMB 125.8 billion of revenue at a 1.03% gross margin from smelted/traded gold, plus a negative-gross-margin smelted zinc business. A RMB 910.7 billion market cap means the marginal elasticity of any single metal is diluted by a huge denominator. Suits the role of the branch's low-elasticity anchor.
• (Its inclusion here executes item 6 of §6④ of yesterday's recap: once a branch is finalized, check back whether the top 3 names by net main-force inflow in that branch all appear in the table.)

10中华企业60067552 pts — Watch only (downgraded in this brief) · SSE main ±10%

• Highest Shanghai purity in the group: in the 2025 annual report real estate is 89.58% of revenue, Shanghai is 98.24%, gross margin 30.68%. Cash of RMB 13.383 billion ≈ its market cap of RMB 13.91 billion, liability ratio 65.60%, PB 0.97.
• But the reservoir is empty, and that is decisive: contract liabilities went from RMB 8.869 billion at end-2024 → RMB 2.867 billion at end-2025 → only RMB 108 million in 2026Q1 (−96% in one quarter), while revenue recognized over the same period was RMB 3.290 billion and the gross margin fell from 33.98% to 17.14%. This was a low-margin, clear-the-shelves recognition; RMB 108 million of contract liabilities means there is almost nothing left to recognize over the next 12 months.
• 7/15 already guided H1 to a first-ever loss of −212 million ~ −352 million (prior-year period +RMB 521.0 million), with Q2 implied at −119 million ~ −259 million. Interim report 8/27.
• Final call: however precisely the policy lands, it must first become new contracts, then presales, then deliveries. Its 2026 financials are essentially already set. The first draft gave it a recommendation slot; that is now corrected to watch only.

---

6. Pass List

One qualitative conclusion spanning the pharma chain: after checking all 8 companies' public pipelines one by one, not one of these 8 A-share names has a "personalized neoantigen therapeutic oncology mRNA vaccine" pipeline. The mRNA pipelines of Walvax, BCHT and CSPC are all registered as preventive biologics; mRNA appears 0 times in 康希诺 (688185)'s July 2026 investor-relations records; 智飞生物 (300122) and 金迪克 (688670) have no mRNA platform at all. What yesterday's 37 pharma limit-ups mapped to was a technology-route label, not a business connection.

---

7. Intra-Branch Ranking

Branch One: Shanghai Real Estate, "Shanghai Eight Measures" (criteria = contract liabilities / recognizable reservoir + whether saleable value and the policy tools are in the same market)

• Hardest single name: none. Not one company on this chain has 2026 financials capable of carrying the Shanghai Eight Measures — the best of them, 上实发展 (600748), is merely a "narrower loss", and the improvement driver may come from Quanzhou rather than Shanghai.
• Highest trading recognizability: 城投控股 (600649). First that can be falsified: 上实发展 (8/22 interim report).
• In one sentence: the Shanghai Eight Measures target "outside the Outer Ring + down payments + second-hand turnover", and not one name on this chain has its saleable value genuinely sitting in the rigid-demand market outside the Outer Ring. The mismatch is not about "whether it can be delivered", but that "it is simply not aimed at the same market".

Branch Two: Precious Metals (criterion = self-mined ore vs smelting)

• Greatest elasticity: 盛达资源 (000603) tied with 兴业银锡 (000426) (incremental gross profit of about 0.62% per unit of market cap, roughly 2x 湖南白银 (002716) and 20x 紫金矿业 (601899)).
• One reminder covering the whole group: the ranking above measures sensitivity to a further rise in the silver price. If silver stays where it is, the two names with the greatest elasticity are also the two facing the greatest QoQ headwind (see counter-evidence 3 in §2).

---

8. Opening Verification Signals for Today

Following item 3 of §6④ of yesterday's recap: every criterion of the form "A outperforms B" must add an absolute-direction gate ("and A's change > 0").

Auction signals (09:15–09:25)

1. [Is the Shanghai Eight Measures a launch or a cash-out] What is 城投控股 (600649)'s auction shape? This is the single most important fork today — yesterday's seal happened before the policy text was released, meaning informed money is already in.
   • One-word board / gap up ≥ +7% → the policy is fully priced plus there was front-running yesterday, so the day is most likely a cash-out, and chasing carries more risk than opportunity.
   • Gap up +2%~+5% → the ideal shape; today is day 1 of pricing the text.
   • Flat or gap down → the front-runners are starting to distribute, and the whole branch is downgraded on the spot.
2. [Branch breadth] Among Shanghai local property names (中华企业 600675, 陆家嘴 600663, 上实发展 600748, 浦东金桥 600639, 外高桥 600648), are ≥ 3 gapping up more than +2%? ≥3 = a sector effect; ≤1 = a single-stock move in 城投控股 (600649) only.
3. [Will precious metals repeat yesterday] The size of the opening gaps in 盛达资源 (000603), 兴业银锡 (000426) and 山东黄金 (600547).
   • Any one of them gapping up ≥ +5% counts as entering chase territory — yesterday Shengda gapped up +6.49% and its full-day open→close was −0.12%; that shape has already occurred once this week.

Sector signals (09:30–10:30)

4. [Can the new main line be established] Before 10:30: are there ≥ 3 limit-ups within Real Estate Development (BK0451)? ≥3 → the main-line rotation is confirmed; 0–1 → treat as single-stock moves.
5. [Is yesterday's main line ebbing] At 10:00: the direction of Biologics (BK1044) and Vaccines (BK1598); at the close: are fewer than 15 (< 40%) of yesterday's 37 pharma limit-ups up on the day? Yes → ebb confirmed.
6. [Can the core take the baton] At the close: is 石药创新 (300765)'s seal/turnover ratio > 0.3? (only 0.073 yesterday)

Single-name signals

7. [Are institutions consistent] At the close: is 城投控股 (600649)'s net main-force flow > +RMB 100 million, and does it appear on the Dragon-Tiger list again? (main force was only +46 million yesterday)
8. [Is the self-mined vs smelting criterion valid] At the close: is 盛达资源 (000603)'s change > 湖南白银 (002716)'s, with Shengda positive?
   • If 湖南白银 (002716) is instead the stronger → the market is pricing off "headline growth rate" rather than "cost structure", and the elasticity ranking in §2 of this brief needs to be re-examined within this round. This is the direct test of this brief's most central criterion.
9. [Is silver still stronger than gold] At the close: do 盛达资源 (000603) and 兴业银锡 (000426) both close up, and both outperform 山东黄金 (600547)?
   • Absolute gate: both must post positive gains; merely "outperforming" does not count.
10. [Can the commodity leg hold] Beijing 14:30 reading: is COMEX silver @SI.1 > 68.105 (the 8/20 settlement)? A break below leaves A-share silver names without a fresh driver.
11. [Has the arbitrage started] At the close: is 东兴证券 (601198) up more than +2%? (0.00% yesterday)
12. [The first thing that can be falsified] 上实发展 (600748)'s interim report on 8/22: check the regional composition of its RMB 1.644 billion of contract liabilities, and the revenue-recognition timing of the RMB 1.979 billion Quanzhou bulk transaction. If the improvement comes mainly from Quanzhou (Fujian), the chain "上实发展 (600748) benefits from the Shanghai Eight Measures" snaps on the spot. This is the item that can be verified tomorrow at the lowest cost and with the highest information content.

Risk signals

13. [Is the pseudo expectation gap being misread] At the close: does 拓荆科技 (688072) gap up > +3% and then close down? If so, that confirms "the +1324% headline was misread and then corrected".
14. [Survival rate of yesterday's limit-ups] At the close: are ≥ 32 (40%) of 8/20's 79 limit-up names up today? (the corresponding 8/19→8/20 reading was 14/36 = 38.9%) Below 30% and all continuation strategies fail.
15. [High board] At the close: does 金健米业 (600127) break its seal? (4 boards yesterday but the seal broke 9 times, seal ratio 0.023)
16. [Compliance] Intraday: does 智飞生物 (300122) disclose its interim report or a postponement announcement? (the scheduled date of 8/18 has passed) — for a RMB 34.6 billion market cap, disclosure compliance outranks any thematic narrative.
17. [Coverage self-check (the standing criterion required by item 4 of §6④ of yesterday's recap)] At the close: do the three industry sectors with the most limit-ups today all appear in the branch list in §2 of this brief? If any one does not, this brief still has a main line it "did not see".

---

9. Final Recommendations

① The 5 Names Most Worth Watching Today

② The 3 Strongest Branches Today

③ Directions Not Worth Chasing Today, and Why

1. The entire pharma chain (vaccines / biologics / innovative drugs) — four independent pieces of evidence: ① the A-share market only begins pricing this on day 1 after the one-off US repricing was already complete (Moderna −23.55% on the day, XBI −3.64%, IBB −2.83%); ② last night the A-share vaccine names' own interim results were losses (百克生物 (688276)'s Q2 loss worsened versus Q1, 金迪克 (688670) had half-year revenue of only RMB 0.9109 million with a widening loss, 智飞生物 (300122) lost RMB 388 million in Q1 and its interim report is overdue); ③ the VAT rate on vaccine products has already been raised from 3% to 13%, an industry-wide gross-margin suppressor that is unpriced; ④ on a name-by-name check, not one of these 8 companies has a therapeutic personalized oncology mRNA vaccine pipeline. Of yesterday's 37 pharma limit-ups the highest was only 2 boards — no ladder, no core.
2. The "interim-report surges" disclosed last night — the shoe has dropped; this is not fresh good news. 大族数控 (7/9 guidance RMB 0.90–1.00 billion, actual RMB 957 million), 天赐材料 (7/9 guidance RMB 2.7–3.0 billion, actual RMB 2.861 billion), 中国巨石 (7/15 guidance +65%~85%, actual +73.87%), 长电科技 (7/14–15 guidance RMB 770–950 million, actual RMB 845 million), 四川黄金 (7/11 guidance RMB 390–450 million, actual RMB 433 million), 盛达资源 (6/30 guidance RMB 350–400 million, actual RMB 390 million) — all six landed inside their own guided ranges, zero expectation gap.
3. 拓荆科技 (688072)'s "+1324%" — of the RMB 1.343 billion of net profit attributable to parent, RMB 975 million or 72.6% is non-recurring, and ex-non-recurring is only RMB 367 million; the RMB 571 million of Q1 was disclosed back in April. The number most likely to be misread off the headline today.
4. 湖南白银 (002716)'s "+170.34%" — 97.06% of revenue comes from smelting at a 4.74% gross margin; the company itself says that when raw materials are unpriced, a rising silver price "will produce a cost inversion"; operating cash flow is −RMB 1.377 billion; cash of RMB 546 million cannot cover RMB 1.203 billion of short-term debt. It is precisely the only one of the 7 that had not pre-announced and is therefore a "first disclosure" this morning — the brightest headline over the most fragile structure, and that mismatch is the shape most worth guarding against today.
5. 志邦家居 (603801) and the "trade-in" association — the "trade-in" in the Shanghai Eight Measures is a home-purchase subsidy, not a furniture subsidy, and is a different policy from the 2024–25 national subsidies for home furnishing; 7/14 already guided H1 to a first-ever loss of −126 million ~ −189 million; its estimated Shanghai revenue exposure is just 0.1%–0.8% of revenue, 1–2 orders of magnitude away from the loss; the 8/20 "7% intraday deviation" Dragon-Tiger list was net selling of −RMB 2.62 million.
6. 光明地产 (600708) — 2025 blended selling price of RMB 8,741/sqm, only 19% of Shanghai's citywide average, so the bulk of saleable value is most likely not in Shanghai; net gearing 324%; H1 guided to a widening loss; PB 1.15 is the only developer in the group above 1.
7. 近岸蛋白 (688137) — on 8/15 the company itself announced an H1 loss "wider than the same period last year" and proactively flagged PS (TTM) of 32.38 above peers; mRNA revenue in 2025 was only RMB 14.34 million, 9.79% of revenue; yesterday's close of 67.49 is still below 68.43 on 8/17.
8. Names that only sealed in the afternoon yesterday, or that broke their seals repeatedly — 安科生物 Anke Biotechnology (300009,13:00 seal, seal ratio 0.031), 康泰生物 Kangtai Biological (300601, seal ratio 0.054), 通鼎互联 (002491, seal broken 13 times), 金健米业 (600127, seal broken 9 times, seal ratio 0.023).
9. Non-Shanghai property names — 京投发展 (600683, Beijing), 首开股份 (600376, Beijing, already limit-down, H1 guided loss of RMB 1.7–2.2 billion). The "Shanghai Eight Measures" is a policy for one city, Shanghai.

④ Final Judgment in One Sentence

Today is a "policy rotation day", but you must separate "strongest catalyst" from "strongest evidence" — they are two different stocks: the strongest catalyst is the Shanghai Eight Measures, yet all three main names on that chain guided their H1 to losses back in July, and what they sell is central-district luxury at RMB 126,000 to RMB 168,000/sqm — the policy targets rigid demand outside the Outer Ring, which is simply not the same market, so it can only be traded on the sentiment line; the one that genuinely combines "a fresh overnight driver + interim results already delivered + verifiable business purity" is silver, not property, and not yesterday's pharma line.

---

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

Data Collection and Run Log (not sent to clients)

1. Data Collection Failures

• push2.eastmoney.com returned HTTP 502 (third consecutive trading day unavailable). Quotes/sector/fund-flow calls were switched to push2delay.eastmoney.com (200 throughout). That endpoint returns data.diff as a dict rather than a list (keys "0", "1", ...), so slicing it as a list raises TypeError: slice(None, 3, None).
• Cailianshe telegraph is entirely unavailable: www.cls.cn/nodeapi/telegraphList → 404; api3.cls.cn/... → {"errno":50101}; www.cls.cn/telegraph is Next.js client-rendered with no body text in the static HTML. News sources were switched to Tonghuashun news.10jqka.com.cn/tapp/news/push/stock/ (14 pages, 700 items, reaching back to 8/19 20:31) + Sina zhibo.sina.com.cn/api/zhibo/feed (800 items), cross-checked across both.
• Three Eastmoney financial-data reportNames all failed on the main path (RPT_PUBLIC_OP_NEWPREDICT, RPT_LICO_FN_CPD, RPT_PUBLIC_OP_PREDICT; every filter syntax returned success:false). Main-path financial data was switched to F10 emweb.securities.eastmoney.com/PC_HSF10/NewFinanceAnalysis/ZYZBAjaxNew and BusinessAnalysis/PageAjax. That endpoint intermittently returns gzip without a Content-Encoding header (688276 raised UnicodeDecodeError on the very first request), requiring manual decompression keyed off the magic number \x1f\x8b.

  ⚠️ This endpoint failure directly caused the most serious error in this brief (see section 5). The sub-agent successfully retrieved the earnings pre-announcements using the same RPT_PUBLIC_OP_NEWPREDICT and akshare's stock_yjyg_em — which shows the endpoint is intermittently available / syntax-sensitive, and that the main path's failure was not a global outage. The right move at the time was to retry via akshare, not to give up.

• newsapi.eastmoney.com/kuaixun/... returned HTTP 567; esmchina.com WebFetch timed out at 60s — the "chip price-hike wave / Maxscend price-adjustment letter" item was removed from the report entirely because its publication date could not be confirmed.

2. Search Engine Returned a Three-Year-Old Article (caught)

A WebSearch for "Cailianshe August 21 morning news roundup" returned m.cls.cn/detail/1438014; a WebFetch confirmed its publication time was 2023-08-21 08:04 Monday (the content being the 2023 CSRC fee cuts, Country Garden extension and Evergrande), while the search summary treated it as 2026 content and summarized it as such. That item was discarded in full. Same shape as the remembered lesson "recurring column headlines hide the publication date".

3. Full-Universe Coverage Checks

4. Freshness Verification (symbol by symbol)

• CNBC: last_time for XBI/IBB/SMH/IGV/XLE/XLU/XLV/GDX/KWEB/FXI/SLV/GLD/SIL/WMT/MRNA/MRK/BNTX was all 2026-08-20T16:00~16:15 -0400, confirming an 8/20-close basis. @GC.1/@SI.1 were read at 2026-08-20T19:0919:16 -0400 = Beijing 8/21 07:0907:16, which is tagged in the body.
• ⚠️ @CL.1 (crude oil) has last_time = 2026-08-20T01:00:00 -0400, about 18 hours stale — the known "dead data" pattern. This brief uses no crude oil quotes at all.
• A-share quote snapshot timestamp 2026-08-20 16:11:50 (push2delay field f86), i.e. the 8/20 close.
• The fixed overnight-sector checklist required by yesterday's recap was read through item by item — it was precisely the XBI/IBB readings that revealed the main line was decaying.

5. ⚠️ The Most Serious Error in This Brief: Failing to Check Earnings Guidance for the Property Names

The first draft described 城投控股 and 中华企业 as having "2026 interim reports not yet disclosed" and used that as "uncertainty", placing both in recommendation slots 1 and 4. The fact is: both companies' H1 earnings guidance had already been published on 7/11 and 7/15 respectively, and both were first-ever losses. The same applies to 志邦家居 (7/14), 光明地产 (7/15, widening loss) and 上实发展 (7/15, narrower loss).

The most glaring part is this: within the same report, I strictly applied the criterion "go back and check July earnings guidance" to the semiconductor and electrolyte names (and on that basis downgraded 大族数控, 天赐材料, 中国巨石 and 长电科技 all to "the shoe has dropped"), yet did not do the same thing at all for the property names. The criterion was right, the application was selective — and the selectivity landed exactly on the line I most wanted to recommend.

The proximate cause is that I abandoned the check after the Eastmoney guidance endpoint failed as logged in §1, but that is only an excuse: the correct action on an endpoint failure is to retry another source (the sub-agent got it on the first try with akshare's stock_yjyg_em), not to downgrade "could not retrieve" into "not yet disclosed". "I could not find it" was silently written as "it has not happened" — exactly the same failure mode as the remembered lesson "'no news found' ≠ 'no news'".

6. Other Substantive Corrections by Sub-Agents to the Main-Path First Draft

Three factual errors in the precious-metals section (all overturned by primary sources):

1. "赤峰黄金 is the only one still rising after the gap" — does not hold. Computing 8/20 open→close name by name: 山东黄金 +3.42% (strongest) > 赤峰 +2.44% > 紫金 +0.74% > 兴业银锡 +0.73% > 盛达 −0.12% > 湖南白银 −1.35% > 四川黄金 −1.84%, 4 of them positive. That criterion came from the wording of yesterday's recap and was inherited by the first draft without re-verification — a textbook case of "inheriting the previous piece's conclusion without recomputing it".
2. "山东黄金's gross margin is only 18.30%, so quality is poor" — a misreading. The low consolidated gross margin is diluted by purchased gold (41.42% of revenue / 0.80% gross margin) + small gold bars + trading (together about 61% of revenue but only 3.4% of gross profit); on a gross-profit basis, self-mined gold contributes 91.00% at a 50.57% gross margin. The first draft downgraded it to "watch only" on that basis; this version upgrades it into a recommendation slot. Lesson: when comparing miners' gross margins across companies, strip out the trading/purchased segments first.
3. "紫金 is copper-led" — inaccurate on a gross-profit basis (mined gold 43.2% > mined copper 36.5%). The direction of the conclusion is unchanged (silver is only 2.11%).

Key facts missed entirely by the first draft and supplied by agents: ① 兴业银锡's 8/19 plan to acquire *ST 威领 for up to RMB 1.8 billion (negative target net assets, no silver); ② 近岸蛋白's 8/15 self-announced H1 loss warning that proactively flagged PS of 32.38 above peers; ③ 陆家嘴's +35.9% includes the one-off DBS Bank Tower disposal (the 79.33% vs 30.09% gross margin and minority profit collapsing from RMB 360 million to RMB 13.7 million being corroborating evidence); ④ 90% of 城投控股's 2025 net profit attributable to parent came from investment income (its stake in Western Securities); ⑤ on 8/20 Chengtou sealed at 13:01 and Zbom at 09:33, both earlier than the 14:59 policy release; ⑥ the "PE" on quote screens is a dynamic PE, not TTM.

The pharma section had its wording strength revised in the opposite direction: the first draft's "the catalyst has been repudiated in its place of origin" overstated the counter-evidence — MRNA is still +111.7% versus the pre-catalyst close of 62.96 on 8/18, and the accurate statement is "a one-off repricing was completed within a single US session and the next day enters decay". Writing "drawdown" as "falsification" makes readers overreact to the whole chain.

7. Methodological Takeaways From This Brief

1. "Go back and check the first disclosure date of earnings guidance" must be applied without exception to every branch, especially the one you like most (see section 5).
2. "H1 cumulative growth" must be broken down to the single quarter and read alongside ex-non-recurring: this overturned 拓荆科技 (non-recurring 72.6%), 湖南白银 (Q2 −92% QoQ) and 天赐材料 (Q2 −27% QoQ).
3. The first criterion for property names is contract liabilities, not current-period revenue: 中华企业's revenue of RMB 3.290 billion looks acceptable, but contract liabilities collapsing from RMB 2.867 billion to RMB 108 million means that was a clear-the-shelves recognition, with nothing left to cook over the next 12 months.
4. The first criterion for precious-metals names is "self-mined vs smelting", ahead of growth rate: 湖南白银's own statement that "when raw materials are unpriced, a rising silver price will produce a cost inversion" is decisive primary evidence.
5. Checking back on revenue composition / segment gross profit changes conclusions: 城投控股's registered business scope still lists water services (spun off to 上海环境 601200 back in 2016) while real estate is actually 99.78% of revenue; 山东黄金's consolidated gross margin is the reverse case, where the trading segment must be stripped out before real earning power becomes visible. Business-licence basis ≠ revenue basis; consolidated basis ≠ segment basis.
6. For policy catalysts, always ask one more question: "is it even aimed at the same market?" The Shanghai Eight Measures' tools point at rigid demand outside the Outer Ring, while the chain's saleable value is central-district luxury at RMB 126,000–168,000/sqm. Fit cannot be judged from the word "Shanghai".

8. What This Brief Did Not Do / Could Not Obtain / Remains Uncertain

• The risk-control quality check (risk-auditor) was not run before finalization — red-flag screening in this brief was carried by cross-corrections from three fundamentals-analyst runs. Given that those three already overturned the first draft's two top recommendation slots and three factual errors, an independent quality check should still be added, particularly for the newly promoted 盛达资源 and 山东黄金 (neither of which has yet been challenged by anyone).
• 城投控股 closed 8/20 at 4.16 = the day's high = the limit-up price, yet it does not appear in Eastmoney's official 79-name limit-up pool, while in the Dragon-Tiger data it exists with CHANGE_RATE=10.05. The two official bases contradict each other, and the reason has not been determined. The body follows the official pool (not counting it in the limit-up tally) and states only the independently verifiable fact that it "closed at the limit-up price".
• The Dragon-Tiger seats were not broken down to institutional detail. BILLBOARD_NET_AMT is the whole-list net amount, not the institutional-seat net amount, so 城投控股's "+RMB 80 million" does not mean institutions bought RMB 80 million — the body consistently uses the phrase "list net buy".
• The revenue and gross-profit contribution of the DBS Bank Tower within 陆家嘴's RMB 3.744 billion of H1 "property sales" could not be obtained — what H1 net profit attributable to parent is after excluding it, and whether the YoY is positive or negative, directly determines whether the "+35.9%" description can be used at all. This requires the interim report's revenue-breakdown notes.
• The specific number of units / amount of 城投控股's housing-voucher sales is undisclosed (the annual report only describes "Shuishang Huating successfully practiced" it qualitatively); its consolidated interest-bearing debt of RMB 41.818 billion is a sum of four line items, while the annual report separately states "RMB 25.503 billion on a non-consolidated basis", and the attribution of the RMB 16.3 billion difference must await the interim report notes (8/31).
• 兴业银锡's silver-elasticity ranking (~0.62% of market cap) rests on the assumption of "flat FY2025 output + normalization against SLV average prices", making it the only non-measured entry in the elasticity table; H1 segment data must wait for 8/29.
• The absolute silver output (in tonnes) of 盛达资源 and 兴业银锡 could not be obtained — Shengda only discloses output +8.82% YoY, and since the company times its sales, output ≠ sales volume.
• Neither 志邦家居 nor 上实发展 discloses its Shanghai revenue share (segments only go to "domestic 84.49%" and "East China 90.04%"); the magnitude estimate for Zbom in the body is explicitly flagged as a top-down assumption.
• The specific composition of 拓荆科技's RMB 975 million of non-recurring items could not be obtained (it may relate to the semiconductor-company acquisition it suspended trading to plan in July).
• The official primary text of the "Shanghai Eight Measures" has been retrieved: Shanghai Municipal Housing Administration website¹⁸. Checking against it corrected two distortions introduced by second-hand summaries: ① the media-reported "first homes uniformly 15%, second homes inside the Outer Ring 25%" is not new in this round — the notice's only operative change is the single tier "second homes outside the Outer Ring → no less than 15%" (Shanghai already issued "Shanghai Seven Measures" on 2026-02-25, and most tiers are pre-existing levels) — so the first draft's "three-pronged push" written on that basis overstated the marginal force; ② official clauses ④ and ⑤ contain qualifiers absent from the second-hand summaries: housing-voucher resettlement "with priority focus on urban-village redevelopment and old-city renovation projects", and acquisition "in each central district ... converted to rental use ... increasing the supply of affordable rental housing in central districts". Those two qualifiers in fact make 城投控股 (Luxiangyuan Phase II sits in Huangpu's old town, i.e. old-city renovation, and it has the Chengtou Kuanting affordable-rental platform) a better policy match than the first draft judged, and §2③ was revised accordingly from "the whole chain is mismatched" to "the three demand-side items are mismatched, the two stock-activation items match". Lesson: reading only second-hand summaries simultaneously overstates the force and understates the structure — deviations in both directions occurred on the same policy.
• The SGE Au99.99 closing price was not obtained. Northbound flows have not been disclosed since 2024-08-19; southbound flows net sold HKD 10.412 billion on 8/20.