A-Share Market Recap | 2026-07-30 Thursday

⚠️ Risk disclaimer (up front): This recap is post-close information review and observation only. It does not constitute investment advice, and contains no ratings, target prices, position sizing, or buy/sell recommendations. "Watch / avoid" in this report is only a ranking of research attention and does not represent any judgment about future returns.

Data-basis disclosure, stated first

1. Time window: 2026-07-30 09:30 open → 15:00 close; report cut-off 2026-07-30 15:42.
2. Today's main market-data interfaces are partially restored, but gaps remain. Available: Eastmoney limit-up pool / limit-down pool / failed-limit-up pool / previous-day limit-up pool (stock_zt_pool_em, stock_zt_pool_dtgc_em, stock_zt_pool_zbgc_em, stock_zt_pool_previous_em), Sina full market quotes (stock_zh_a_spot, 5533 names), Sina index quotes (stock_zh_index_spot_sina), Tonghuashun industry sector summary and industry fund flows (stock_board_industry_summary_ths, stock_fund_flow_industry), Eastmoney historical Dragon-Tiger List detail (stock_lhb_stock_detail_em). Unavailable: stock_zh_a_spot_em, stock_board_industry_name_em, stock_board_concept_name_em (5 consecutive RemoteDisconnected) — therefore all sector data in this report is on the Tonghuashun basis, which is not the same classification set as the Eastmoney basis cited in yesterday's report; the two sets' sector moves and net fund amounts cannot be compared directly across days. This is the data-basis discontinuity readers most need to note this issue.
3. ⚠️ No reliable data available for today's Dragon-Tiger List (the reason is clear; it is not an interface failure): the cut-off time is 15:42, and the exchange normally publishes the Dragon-Tiger List after 17:30–18:00; in testing, stock_lhb_stock_statistic_em still returned 2026-07-29 as the most recent listing date. Therefore Section 5 provides no Dragon-Tiger List or seat preferences for today; it uniformly states "no reliable data available" and makes no inferences whatsoever.
4. ✅ But this issue clears a two-day-old debt: both yesterday's recap and this morning's pre-market brief recorded "the nature of the 7/29 Dragon-Tiger List seats has not been backfilled" as the same outstanding item for a second consecutive day. This report has used stock_lhb_stock_detail_em to obtain the complete top-five buy and sell seats for 7/29 for 风华高科 (Fenghua Advanced Technology, 000636), 天融信 (Topsec, 002212) and 一鸣食品 (Yiming Foods, 605179) (see 5.3). Among them, 风华高科's seat structure directly rewrites this morning's pre-market characterization of it, and is the single most important new piece of evidence this issue.
5. Northbound flows cannot be obtained today: the exchange discontinued single-day net-buy disclosure from August 2024; today stock_hsgt_fund_flow_summary_em returned 0.00 for "net turnover purchase" on both northbound rows, with trading status code 3 — this means the interface has no valid value, not that "net buying was zero", so this report gives no northbound net inflow figure. The only usable substitute basis is the advance/decline count of constituent stocks (see 5.1).
6. Two data-basis conflicts are disclosed side by side without picking a winner: ① the Tonghuashun sector summary table shows banks' "net inflow" as RMB 11.975 billion, while the Tonghuashun industry fund-flow table shows banks' "net amount" as RMB 10.852 billion, a gap of RMB 1.123 billion whose cause could not be verified; this report uniformly adopts the industry fund-flow table basis and discloses the discrepancy here; ② the market-wide limit-up count is 52 on the Eastmoney limit-up pool, while stocks with "gain ≥9.8%" in Sina's full market quotes number 66 (the difference comes mainly from ChiNext/STAR 20cm names and ST stocks that did not seal the limit), the two are not the same basis and must not be mixed.
7. Dimensions this report does not cover at all: today's Dragon-Tiger List and hot-money seats, today's northbound net amount, stock-level main-fund net inflow rankings (the Eastmoney fund-flow interface is unavailable, and only the top of the Tonghuashun real-time single-stock list was obtained), and closing-auction turnover structure.

---

0. One-Sentence Recap

Today's sentiment temperature: divergence to the point of tearing — at the index level it was a "mild correction," at the structural level it was an extreme two-tier market of "AI hardware collapse + big consumer and financials rallying independently." The SSE Composite fell −0.62% and the SSE 50 bucked the trend at +0.39%, while the STAR 50 fell −5.38% and the ChiNext Index −3.97%, the CSI 500 −2.86% and CSI 1000 −2.74%; market-wide limit-downs of 74 exceeded limit-ups of 52, with 3635 decliners versus only 1767 advancers, and a median single-stock change of −1.15%. The strongest main lines were baijiu (+4.48%, first market-wide) and banks (+2.43%, net fund inflow of RMB 10.852 billion, first market-wide), followed by auto OEMs (+2.85%) and education (+2.78%).

The reconciliation of this morning's pre-market list is uncomfortable and must be said plainly: the name-by-name hit rate of 31/38 = 81.6% looks fine, but the "top 3 branches" hit 0/3, and the two lines that actually led today (baijiu, banks) were explicitly written into the pre-market's "directions not recommended to chase today." The three main lines the pre-market recommended — optical chips, passive components/MLCC, and copper/nonferrous — all failed. And they failed in exactly the way the pre-market's own Section 8 "failure signals" described: 永鼎股份 (Yongding Holding, 600105) touched the limit as a lone name while 长光华芯 (Everbright Photonics, 688048) fell −15.17%; 风华高科 (000636) broke its limit 6 times while the components sector fell −6.58%; 西部矿业 (Western Mining, 601168) at −0.70% did nothing. In other words, the pre-market's "falsification conditions" were designed correctly and its "branch ranking" was wrong — and the pre-market's own Section 9.4 wrote "go by the observable falsification conditions in Section 8, not by ranking on news strength," a sentence fully vindicated today.

Tone for tomorrow: the sell-off in AI hardware shows no stabilization signal yet (semiconductors 176 down, 7 up), and the Central Politburo meeting reported at 14:38 on July 30 left only about 22 minutes of trading time in the session — tomorrow (7/31, Friday, the last trading day of the month) is the first full trading day on which that policy tone will be fully priced.

---

1. Market Overview

1.1 Indices and Turnover

Combined turnover of the two markets was RMB 2,342.809 billion (Shanghai 1,106.477 + Shenzhen 1,236.332, Sina index basis).

⚠️ Verdicts on the two technical levels flagged pre-market (explicitly set as risk signals in pre-market Section 8.4):

• SSE Composite 3800: intraday low of 3767.50 — lost, close of 3804.69 barely recovered. By the pre-market's own standard, "if the SSE Composite loses 3800 → the 7/29 deep-V repair is negated" — it triggered intraday and was narrowly held at the close, an in-between state that is "neither confirmed nor falsified."
• ChiNext Index 3333: intraday low of 3158.14 and close of 3244.62 — decisively lost.

1.2 Limit-Ups/Downs and Sentiment Temperature

Sentiment temperature verdict: not "freezing," not "warming," but "structural euphoria and structural freeze coexisting." Basis:

• Evidence for freezing: 74 limit-downs > 52 limit-ups; decliners were 2.06x advancers; yesterday's limit-up stocks had a median of −1.42% today and only a 12.5% advance rate; failed-limit-up rate of 26.8%.
• Evidence for euphoria: the baijiu sector had 19 advancers, 0 decliners; the banking sector had 42 advancers, 0 decliners (Tonghuashun sector summary basis); 贵州茅台 (Kweichow Moutai, 600519) +3.09% hit a near-two-month high, and 工商银行 (ICBC, 601398) and 建设银行 (China Construction Bank, 601939) reportedly hit all-time highs per media.
• The two are not contradictory: this was a one-directional, violent migration of capital out of tech and into consumer and financials, not a broad decline. The reason the index drop (SSE Composite −0.62%) was far smaller than the single-stock median (−1.15%) is precisely that heavyweight blue chips were holding the index up — the SSE 50 was the only broad-based index to close green.

---

2. Pre-Market List Reconciliation

Reconciliation basis (rules stated first, to avoid adjusting standards after the fact):

• A positive label (priority deep-dive / watch closely) is judged "delivered" = closed up (>0%) on the day.
• A negative label (watch only / Pass) is judged "correct" = the day's gain was <+5% (i.e., no significant move was missed). Note this rule only claims "should not have been bought," not "will fall" — this is the lesson acknowledged for 天融信 (002212) in yesterday's recap, carried over today.
• It also attaches outperformance/underperformance versus the day's median (−1.15%) as a second yardstick, because on a day when the median is negative, "a small decline" and "a gain" are not the same in nature.

2.1 Positive Recommendations (priority deep-dive / watch closely) — 8 names

Positive recommendation hit rate: 6/8 = 75.0% (7 of the 8 beat the median; only Western Mining fell into "beat the median but fell in absolute terms").

2.2 Watch Only — 16 names

Watch-only hit rate: 15/16 = 93.8%. Among them, 江海股份 (002484) at −9.61% was the cleanest correct call on the whole list — the pre-market's reason was "its core business is aluminum electrolytic capacitors ≠ MLCC, a category mismatch, so the mapping must be discounted," and today it came in nearly 16 percentage points below the genuine MLCC name 风华高科 (+6.29%). That piece of supply-chain common sense, "category mismatch," was worth 16 points today. 环旭电子 (601231) −8.82% and 京东方Ａ (000725) −6.13% likewise delivered on the judgment that "they belong to the AI hardware chain, and the overnight SOX plunge is a headwind."

兆易创新 (603986) +1.94% deserves its own note: the pre-market listed it as "today's biggest point of disagreement — not recommended to chase, but also not to be declared a certain decliner" — that wording was accurate today. Against a backdrop of semiconductors −7.08% (176 down, 7 up), it closed up 1.94% with turnover of RMB 35.484 billion, validating neither the bear case ("buyback support is zero") nor the bull case ("ex-non-recurring P/E of 26.3x is cheap"). The pre-market was right to refuse a one-directional conclusion.

2.3 Pass List — 14 names

Pass hit rate: 10/14 = 71.4%.

⚠️ All 4 wrongly judged Passes hit the limit, and they share one trait: all are "pure event / pure sentiment" names. 天和磁材 (603072) at +0.06% was a clean correct call — the same "amount is noise-level" logic was right on Tianhe Magnetic Materials and wrong on 广电电气 (601616). This shows that the inference "small amount → should not rise" holds on the earnings dimension but not on the sentiment dimension, see 2.5.

2.4 Hit-Rate Summary

Versus yesterday: yesterday's recap read "avoid 10/10 all correct, recommendations only 2/11 (18.2%)." Today's positive recommendations improved from 18.2% to 75.0%, a genuine improvement after this morning's pre-market added a "positive scan," and it should be credited. But at the branch level it was 0/3, and today's two strongest main lines were written into "not recommended to chase" — which shows the improvement happened at the stock-selection level, not at the direction-selection level.

2.5 Post-Mortem: What Went Wrong Today and How to Fix It Tomorrow Morning

① What was called right was the "falls," what was called wrong was the "rises" — the same structure as yesterday, just in a different place.

The pre-market was extremely accurate on "what would fall": the advice to avoid the entire AI hardware chain delivered in full today — 通富微电 (Tongfu Microelectronics, 002156) −10.00% limit-down, 紫光股份 (Unisplendour, 000938) −9.99% limit-down (the two names called out in pre-market Section 9.3, both limit-down), 江海股份 (002484) −9.61%, 环旭电子 (601231) −8.82%. But the best-performing directions today were, without exception, on the pre-market's negative list:

② The methodological error most deserving of acknowledgment today: the pre-market used "single-day fund flow" to pronounce a branch "already falsified."

Pre-market Section 4b had just set a rule — "the 'expectation gap' must not be measured by a single day's price change; it must use medium-term price change and valuation levels" — on the grounds that "a single day's price is a poor proxy variable for the information state." But the same report then wrote in Section 9.3: "banks had main-fund net outflow of RMB 1.874 billion → the defensive logic has been falsified" — that used 7/29's single-day fund flow to negate a branch. This is the very error it had just prohibited, with "single-day price" merely swapped for "single-day fund flow."

Today's counter-evidence is extremely clean: banks had main-fund net outflow of RMB 1.874 billion on 7/29, and on 7/30 net fund inflow of RMB 10.852 billion, first market-wide, with 42 sector advancers and 0 decliners. One day's fund-flow direction has no predictive power for the next day.

→ First revision for tomorrow morning's pre-market: expand the scope of 4b.1 from "expectation gap" to "branch-survival judgments." Any conclusion that "a branch has been falsified / removed from the list" may not rest solely on a single day's price or a single day's fund flow; it must also have either "the news has already been digested" or "3 or more consecutive days of price-volume divergence." The aviation sector, cleared from the list today by the "zero reaction for 2 consecutive days" rule (airports & aviation +1.02%, 春秋航空 601021 +2.62%, beating the median by 3.77 percentage points), is likewise a victim of that too-fast removal rule.

③ Pass logic must be split into two columns: the earnings dimension and the sentiment dimension.

天和磁材 (603072) and 广电电气 (601616) were given exactly the same Pass reason ("noise-level amount, does not change operations"); the former came in at +0.06%, the latter at +10.14% limit-up. The difference is not in the announcement but in the chip structure and the share-price level: 广电电气 closed at RMB 3.80 with a 6.35% turnover rate, a textbook low-priced, high-turnover sentiment name; 天和磁材 closed at RMB 31.71 with turnover of RMB 58 million. "This announcement is worthless" and "this stock will not rise tomorrow" are two independent propositions, and the pre-market merged them into one.

→ Second revision for tomorrow morning's pre-market: add a "Pass type" column to the Pass list, with values "earnings-dimension Pass (does not affect revenue)" and "dual earnings + sentiment-dimension Pass (and the chip structure does not support sentiment speculation)." Only the latter claims "will not rise"; the former only claims "should not be bought on earnings logic."

④ One thing that must be acknowledged: the pre-market's "verification signal" design (Section 8) was far higher quality than its "recommendation ranking" (Section 9).

Pre-market Section 8.2 wrote a "failure signal" for each branch, and all of them triggered precisely today:

• Optical chips: "Yongding hits the limit as a lone name with no followers → a pure single-stock event, not a branch" → Yongding +7.33%, Everbright Photonics −15.17%, holds.
• MLCC: "MLCC strengthens on its own while AI hardware keeps plunging → leans toward supporting the within-sector defensive catch-up reading" → 风华高科 +6.29%, components sector −6.58%, semiconductors −7.08%, holds.
• Dairy/consumer: "tech stabilizes and rebounds while consumer pulls back in tandem → proves it is merely the mirror image of tech bleeding out" → tech did not stabilize and consumer surged, this one did not trigger; on the contrary, the independence reading gained support.
• Gaming: "opens high, closes low and the net ratio turns negative → a one-day wonder" → gaming −0.85%, net outflow of RMB 857 million, holds.
• Copper/nonferrous: "opens high, closes low → earnings already fully expected" → Western Mining opened at 37.55 and closed at 36.65, holds.

The line in pre-market Section 9.4 — "go by the observable falsification conditions in Section 8, not by ranking on news strength" — was the single most correct sentence in the whole pre-market report today. → Third revision for tomorrow morning's pre-market: move Section 8 (verification signals) ahead of Section 3, so falsification conditions appear before rankings.

---

3. Verification of Today's Main Lines

3.1 The Three Pre-Market Main Lines: 0/3

3.2 Today's Real Main Lines (all missed or treated in reverse pre-market)

Three things that must be stated, so this does not become hindsight rationalization:

① Baijiu was not pulled up by the Politburo meeting; the timing does not line up. The Central Politburo meeting press release was reported at 14:38 (NBD basis), only about 22 minutes before the close. Yet a Sina Finance report published at 12:27 had already recorded that "Kweichow Moutai was up more than 3% intraday, broke through its semi-annual moving average, and hit a near-two-month high, while Shede Spirits hit the limit at the open." That is, baijiu's rally was already complete in the morning session; the Politburo meeting was a late-session confirmation, not the cause. Per media attribution, the morning's drivers were industry-level medium-term factors: institutional holdings in the baijiu sector fell to a historic low of 0.97% in Q2, Q2 earnings were cleared, consumption data beat expectations, and Moutai raised prices for the second time this year (all of the above are media and sell-side sourced; this report obtained no primary data source, and the precision remains to be verified).

② Banks' independence was the strongest today, and also the least "defensive." All 42 bank names closed green with net fund inflow of RMB 10.852 billion, first market-wide — "defensiveness" usually shows up as price rising while money does not come in (today's education sector was exactly that: +2.78% with a net amount of −RMB 44 million), whereas banks saw prices rise and money flow in heavily, which is allocation behavior, not risk-aversion behavior. The pre-market used "7/29's index green, money leaving" to characterize it as "defensiveness has failed," and today proved the reverse.

③ The storage chain "must be viewed in three segments" — this was the pre-market's only correct structural judgment and must be credited. Pre-market Sections 8.5 and 9.4 twice stressed that "there is no full-chain storage collapse; the design / packaging / compute-hardware segments must be viewed separately." Today's result:

The spread from this segmentation judgment reached 20 percentage points (德明利 +10.00% vs 通富微电 −10.00%), making it the most valuable single judgment in today's pre-market. But one inaccuracy must be acknowledged at the same time: the pre-market also put 澜起科技 (Montage Technology, 688008) in the interface segment that "closed up or flat on 7/29, with valuation de-rating essentially complete"; today it fell −5.17%, so that classification does not hold.

---

4. Limit-Up Ladder / High Boards

4.1 Ladder Structure (52 limit-ups)

⚠️ The fatal flaw of the ladder: a gap. Between the 8-board and the 4-board names, the 5, 6 and 7 heights are all empty, and only 1 name bridges the 4-board and 3-board levels. This is not a healthy ladder but three isolated high boards plus a pile of first boards. Of yesterday's 80 limit-up stocks, only 10 advanced a board today (12.5%), with a median of −1.42% — continuation trades lost money.

4.2 Quality of the High Boards

• 爱丽家居 (603221), 8 boards: turnover of only RMB 68 million, turnover rate 1.37%, sealing orders RMB 307 million, locked one-line at 09:25:03 and never opened all day. Exactly the same pattern as 7/29, still a "volumeless one-line isolated high board" — the pre-market's reason for passing on it ("an isolated high board rather than the top of a ladder") was right on the structural judgment and merely wrong on the price direction. The risk in this kind of name is not whether it rises, but that once the limit opens you cannot sell.
• 明新旭腾 (605068), 4 boards: turnover RMB 123 million, turnover rate 3.67%, broke the limit once before resealing, sealing orders RMB 44 million. Sealing orders have visibly thinned (relative to the other 2-board names).
• 传智教育 (003032), 4 boards: turnover RMB 197 million, turnover rate 8.80%, sealed at 09:30:09 and never opened all day, sealing orders RMB 108 million. It is the only one of the three high boards whose sealing quality is improving.
• 一鸣食品 (605179), 3 boards: turnover of RMB 49 million, turnover rate of only 0.69%, sealed at 09:30:48 and never opened, sealing orders RMB 159 million. Pre-market Section 8.1 set "can Yiming Foods make a third board" as the criterion for the dairy branch's survival, and the third board did happen today — but a third board on RMB 49 million of turnover shows that nobody is selling, not that many people are buying.

4.3 Core Leaders and Main-Line Identifiability

Among the 42 first boards, the only ones with "core leader" characteristics (large turnover + high identifiability + main board) are:

• 德明利 (001309): turnover of RMB 11.366 billion (the largest among limit-up stocks), sealing orders RMB 716 million, turnover rate 18.57%, close RMB 390.04. Hitting the limit against a backdrop of semiconductors −7.08% and 176 decliners, it was the strongest single-stock anomaly today.
• 江淮汽车 (600418): turnover RMB 1.808 billion, sealing orders RMB 195 million, market cap RMB 54.979 billion — the core leader of the auto OEM +2.85% main line.
• 舍得酒业 (600702): turnover RMB 522 million, sealing orders RMB 117 million; 金徽酒 (603919): turnover RMB 378 million, sealing orders RMB 64 million. Together the two form the limit-up end of the baijiu main line.

Industry distribution (Eastmoney limit-up pool basis): auto parts 6, power grid equipment 4, agrochemicals 2, beverages & dairy 2, education 2, general equipment 2, software development 2, real estate development 2, baijiu 2, chemical pharmaceuticals 2, conglomerates 2.

⚠️ One easily misread data point: the limit-up pool contains 6 auto parts names, but the limit-down pool also contains 6 auto parts names (晋拓股份 Jintuo 603211 −10.02%, 天普股份 Tianpu 605255 −10.00%, etc.). Auto parts is "extremely divided," not a main line; what is genuinely clean is auto OEMs (+2.85%, 21 up 1 down). Do not read the limit-up count in auto parts as main-line strength.

4.4 Broken Boards and Limit-Down Structure

The industry distribution of the 74 limit-downs is highly concentrated in tech hardware: semiconductors 10, communications equipment 7, components 7, auto parts 6, specialized equipment 5, general equipment 4, power grid equipment 4, optics & optoelectronics 3, specialized engineering 3, automation equipment 3.

6 names that hit the limit yesterday hit the limit-down today, the most brutal breaks among them: 兴欣新材 (Xingxin New Materials, 001358) went straight to limit-down at −10.01% after 3 boards, 信邦智能 (Xinbang Intelligent, 301112) −11.49%, 圣晖集成 (Shenghui Integration, 603163) −10.00%, 博通集成 (Beken, 603068) −9.99%, 中岩大地 (Zhongyan Dadi, 003001) −9.98%, 贤丰控股 (Xianfeng Holdings, 002141) −10.00%.

The two largest limit-down names by size: 杰瑞股份 (Jereh, 002353) −10.00% (turnover RMB 3.286 billion, market cap RMB 128.7 billion), 景旺电子 (Kinwong Electronic, 603228) −10.00% (turnover RMB 3.617 billion, market cap RMB 66.9 billion). Large-cap stocks hitting limit-down is the hardest evidence on the freezing side of today's sentiment.

---

5. Fund Flows

5.1 Northbound / Southbound

⚠️ No reliable data available for today's northbound net amount. The exchange discontinued single-day net-buy disclosure from August 2024; today stock_hsgt_fund_flow_summary_em returned 0.00 for "net turnover purchase" on both northbound rows, with trading status code 3 — this means the interface has no valid value and must not be read as "net buying was zero."

The only usable substitute basis (constituent-stock advance/decline counts from the same interface):

Decliners among Shenzhen Connect names were 2.12x advancers, markedly worse than Shanghai Connect's 1.31x — consistent with the structure of "Shenzhen carries a higher tech weight, Shanghai a higher consumer and financials weight."

Southbound (obtainable): Southbound Stock Connect (Shanghai) net outflow of RMB 1.100 billion, Southbound Stock Connect (Shenzhen) net outflow of RMB 2.353 billion, a combined net outflow of about RMB 3.453 billion, with the Hang Seng Index at −0.04%.

5.2 Industry Fund Flows (Tonghuashun "industry fund flow" basis, unit: RMB 100 million)

(Data-basis disclosure: the Tonghuashun sector summary table gives banks a "net inflow" of RMB 11.975 billion, RMB 1.123 billion different from the RMB 10.852 billion in the table above; the cause of the discrepancy could not be verified.)

Answering the question pre-market Section 8.3 mandatorily left open — "is the money still in those four directions from 7/29":

⚠️ Cross-day comparison warning on data basis: the 7/29 data uses the Eastmoney industry classification, today's uses the Tonghuashun classification, and the constituents of the two classification sets differ (for example, Eastmoney's "passive components" and Tonghuashun's "components" are not the same set). The table above may only be used to judge direction, not to compare amounts.

In one sentence: of the five inflow directions on 7/29, three (passive components, gaming, nonferrous) reversed outright today, one (food & beverage) converged internally into baijiu, and one (securities) migrated within financials to banks and insurance.

5.3 Dragon-Tiger List

⚠️ No reliable data available for today's (7/30) Dragon-Tiger List — the cut-off was 15:42, the exchange normally publishes after 17:30–18:00, and in testing the most recent listing date was still 2026-07-29. This report makes no inferences whatsoever about today's seats.

✅ But the 7/29 Dragon-Tiger List seats, outstanding for two consecutive days, have now been backfilled (source: Eastmoney stock_lhb_stock_detail_em). This is the item that this morning's pre-market self-reported in point 7 as "unfinished for a second consecutive day."

① 风华高科 (000636) 7/29 — the most important new piece of evidence this issue; it directly rewrites the pre-market's characterization of this name

What this data shows: when 风华高科 hit the limit on 7/29, northbound net sold RMB 347 million and institutional seats net sold about RMB 201 million in total, and the buyers were uniformly hot-money branches. Yet this morning's pre-market ranked it Top5 #2 on the core grounds of "sealing orders of RMB 415 million, the thickest market-wide + industry main-fund net inflow of RMB 2.594 billion, ranked first," and on that basis judged it "the core leader of the branch, the one capital most endorses." The Dragon-Tiger List shows the direction of that inference is backwards: the thickest sealing orders were put up by hot money, while institutions and foreign capital were distributing into the limit-up.

→ Today it broke its limit 6 times, with an amplitude of 12.38% and a turnover rate of 23.11% — not an accident, but the inevitable result of that chip structure.

→ Fourth revision for tomorrow morning's pre-market: for any stock where "sealing-order thickness / main-fund net inflow" is used as a recommendation reason, the direction of institutional and northbound seats on the prior trading day's Dragon-Tiger List must also be checked. "Main-fund net inflow" is a statistical estimate bucketed by single-order size; "institution-dedicated seats" are disclosed by name — when the two conflict, the Dragon-Tiger List governs.

② 天融信 (002212) 7/29 (the backfill item designated pre-market)

Conclusion: pure hot-money play, institutions net sellers, northbound not participating. The listing reason was "cumulative deviation of price change over three consecutive trading days reaching 20%." Today's −4.70% validates the pre-market judgment that "the main funds are distributing into the rebound." With this, two of the three seat-nature items designated in yesterday's recap — Topsec and Fenghua Advanced Technology — have now been backfilled.

③ 一鸣食品 (605179) 7/29 (representative of the food & beverage limit-up cluster)

The top five buyers totaled only about RMB 58 million (Guotai Haitong Kunming Renmin Middle Road +RMB 15 million, Guoxin Securities Beijing Zhongguancun +RMB 14 million, Huatai Securities Head Office +RMB 12 million, Shanghai Stock Connect +RMB 10 million, Huaxin Securities Shanghai Xuhui Ruiping Road +RMB 6 million); the top five sellers totaled about −RMB 19 million (Shanghai Stock Connect −RMB 7 million, the largest).

Conclusion: the scale of participating capital is tiny — a textbook case of "locked float on no volume" rather than "capital scrambling for chips." This is entirely consistent with today's pattern of a third board on turnover of only RMB 49 million and a turnover rate of 0.69% — nobody is selling, but not many people are buying either. A third board of this kind is not evidence of "branch strength."

5.4 Single-Stock Fund Flows (the obtainable portion)

Among the top of the Tonghuashun real-time single-stock fund-flow list, those that hit the limit yet had main-fund net outflow include: 广电电气 (601616) net outflow of RMB 64 million (limit-up), 均瑶健康 (605388) net outflow of RMB 43 million (2 boards), 返利科技 (600228) net outflow of RMB 39 million (2 boards), 南宁百货 (Nanning Department Store, 600712) net outflow of RMB 52 million (limit-up). That is, a fair share of today's limit-ups were "low-volume ramps + net capital outflow" rather than real money flowing in — which corroborates the 12.5% advance rate in Section 4.1.

(The Eastmoney single-stock fund-flow ranking interface was unavailable today, so a market-wide list of top main-fund net-inflow stocks cannot be given; this is a gap in this section.)

---

6. Outlook for Tomorrow (2026-07-31, Friday · last trading day of the month)

This section gives only conditional judgments and verification points; no directional conclusions, no position sizing, no buy/sell advice.

6.1 Main-Line Continuation Judgments

6.2 Key Watch Items for Tomorrow (branches + main-board stocks + verification points)

The following are research observation directions and observable verification conditions only; no buy/sell advice is included.

6.3 What to Avoid Tomorrow

6.4 Input Prompts for Tomorrow Morning's Pre-Market List

Six items that must be carried into tomorrow morning's pre-market:

1. [Methodology · highest priority] Expand the scope of 4b.1 from "expectation gap" to "branch-survival judgments." Using a single day's price or a single day's fund flow to declare a branch "falsified / removed from the list" is prohibited. Today's banks (net outflow of RMB 1.874 billion on 7/29 → net inflow of RMB 10.852 billion, first market-wide, on 7/30) and aviation (removed from the list for "zero reaction for 2 consecutive days" → airports & aviation +1.02% today, beating the median) are two direct counterexamples.
2. [Methodology] Add a "Pass type" column to the Pass list, distinguishing "earnings-dimension Pass" from "dual earnings + sentiment-dimension Pass," with only the latter claiming "will not rise." Today's 4 wrongly judged Passes (广电电气 601616, 神雾节能 000820, 爱丽家居 603221, 明新旭腾 605068) were all pure sentiment names and all hit the limit.
3. [Methodology] Move the "verification signals" section ahead of the "master stock list." Today all 5 failure signals in pre-market Section 8 hit, while Section 9's branch ranking was 0/3.
4. [New mandatory check item] For any stock recommended on the grounds of "sealing-order thickness / main-fund net inflow," the direction of institutional and northbound seats on the prior trading day's Dragon-Tiger List must be checked. 风华高科 (000636) is today's lesson: sealing orders of RMB 415 million, the thickest market-wide, while the Dragon-Tiger List showed northbound net selling of RMB 347 million and institutional net selling of about RMB 201 million.
5. [Three things that must be checked tomorrow] ① The full text of the Central Politburo meeting (reported 7/30 14:38; tomorrow is the first full trading day) — the focus is the industry mapping of three lines: the "six networks," "deepening comprehensive reform of capital-market investment and financing," and "stabilizing the property market, implementing the package debt-resolution plan" — but the "concept ≠ beneficiary" grading required by the iron rules must be applied, and concept stocks must not be named directly; ② Apple and Amazon earnings (7/31) and their impact on the AI capex narrative; ③ the Samsung Electro-Mechanics MLCC price increase formally taking effect on 8/1 — this is the passive components branch's only reverse variable, but today's components sector at 4 up 58 down already proves the market does not buy it, and extremely strong evidence would be required for it to return to the list.
6. [Debt carried forward, must be backfilled tomorrow morning] Today's Dragon-Tiger List (7/30) could not be obtained because the cut-off preceded the exchange's publication time, and tomorrow morning's pre-market must backfill it, focusing on: the seat nature of 德明利 (001309), 江淮汽车 (600418), 舍得酒业 (600702) and 爱丽家居 (603221), and where the seats went in 风华高科 (000636)'s 6 broken limits today. This is a new debt this report registers proactively, and it must not roll over for two days or more again.

---

⚠️ Risk disclaimer: This recap is post-close information review and observation only and does not constitute investment advice. Data may differ in timeliness or basis (this issue's sector data is on the Tonghuashun basis and cannot be compared directly across days with the prior day's Eastmoney basis; neither today's Dragon-Tiger List nor the northbound net amount was obtained). Please rely on exchange/company disclosures, and do not use this directly as a basis for trading. A-share volatility risk is extremely high, and automatically generated content may contain information lags or supply-chain mapping errors.