A-Share Market Recap | 2026-09-28 Monday

Reconciliation baseline: reports/a-share/2026-09-28.md (this morning's pre-market list). Market data basis: limit-up / limit-down / broken-board pools are Eastmoney official closing pools (2026-09-28); index and individual stock quotes were read from the Tencent source after 15:30; two-market turnover is cited from the Gelonghui market close¹. Price limits: main board ±10% | ChiNext / STAR ±20% | Beijing Stock Exchange ±30%; every stock below is labeled with its board.

---

0. One-Line Recap

The expected "gap-up repair" never came — what came instead was a low-open, low-close, high-volume second leg down: SSE Composite -1.67% to close at 3823.62 (below the 9/24 low of 3888, intraday low 3806.67), ChiNext -4.53%, STAR 50 -4.06%; two-market turnover RMB 1.7 trillion (up RMB 49.4 billion vs 9/24), over 4,500 stocks fell; 33 limit-ups (9/24: 52), 56 limit-downs (9/24: 13); yesterday's 52 limit-up stocks averaged -1.95% today, median -3.74% — sentiment slid another notch from the upper edge of the ice point toward the lower edge. The pre-market list got the directional call wrong (the gap-up repair did not materialize), but the structural avoids were almost all right: media high-flyers fading, 山东黄金 (Shandong Gold) / 亨通光电 (Hengtong) / 光启技术 (Kuang-Chi) / 佰维存储 (Biwin) untouched, 康强电子 (Kangqiang) Pass — all validated; the only bright offensive recommendation was 均胜电子 (Joyson Electronics) (+6.19%). The real main line was none of the "holiday positives" but a risk-off low-level rotation: wind power, whole-vehicle autos, baijiu / pharma / power defensives, Fujian / Shenzhen local stocks. Tomorrow's tone: defensive in the fading stage; don't catch high levels, don't catch falling knives.

---

1. Market Overview

• Turnover: RMB 1.7 trillion across the Shanghai and Shenzhen markets, up RMB 49.4 billion vs 9/24 (RMB 1.65 trillion) — volume up 49.4 billion — this is not a quiet bleed lower on shrinking volume; it is panic selling with volume (source: Gelonghui market close¹).
• Limit-ups / limit-downs: 33 limit-ups (main board 30 / ChiNext · STAR 20cm 3), 56 limit-downs; 11 broken boards, broken-board rate 25% (11/44), seal rate ~75%.
• Advance rate: of yesterday's 52 limit-up stocks, only 8 continued to a board today (advance rate 15%), of which only ~3 succeeded at 1-into-2; of yesterday's limit-up stocks, 32 closed lower today and 8 hit limit-down.
• Sentiment temperature: ice point, second leg down. The 9/24 sentiment index of 18.97 was already extreme panic; today there are even fewer limit-ups, limit-downs are 4x those of 9/24, and the advance rate was halved — the panic was not repaired but deepened. The pre-market script "macro positives × sentiment ice point = gap-up repair" was falsified: the valuation pressure of the 30Y UST breaking 5.5% + pre-holiday profit-taking + tech-chain main funds continuing to exit — the triple pressure drowned out the head-of-state consensus positive.

---

2. Reconciling the Pre-Market List

2a. One-by-one check of the 13-stock recommendation list (quotes on the Tencent closing basis)

2b. Weather-vane check (not part of the recommendations)

2c. Hit rate and self-criticism

• Strict hit rate (absolute gainers among the 6 recommendation/watch names): 3/6 = 50% — 均胜电子 (Joyson Electronics) +6.19%, 中海油服 (COSL) +0.41%, 拓普集团 (Tuopu Group) +0.13%.
• Avoid-correct rate including "watch only / Pass": 9/13 (all 6 avoids correct: Biwin, Kuang-Chi, Hengtong, Sanhua, Shandong Gold, Kangqiang).
• But the core offensive calls (the 3 priority deep-dives) went 0/3, and the single biggest pre-market judgment — "today is very likely a gap-up repair" — was wrong: the auction opened lower directly, and the whole day was low-open low-close; even the chance to "fulfill on the gap-up" never appeared. The pre-market's own risk note (FTSE A50 -0.93% on Friday, "take another look before the auction") had in fact already sounded the warning, but it wasn't given enough weight.
• Self-criticism (how to fix the next pre-market):
  1. Overnight forward indicators must feed the conclusion, not just the caveats — when A50 falls 0.93% a day ahead, the confidence in "gap-up repair" should be downgraded directly;
  2. When slow variables like UST yields overpower the fast variables of headlines, the "macro-positive repair" scenario should be downweighted as a whole — on the day the 30Y broke 5.5%, high-PE growth (storage / optical modules) should not have been put in offensive recommendation slots, only in "wait for volume contraction and stabilization" watch slots;
  3. The "watch only / Pass" branch discipline proved effective today — keep it, or even strengthen it: everything that lost money today was tech growth in offensive slots, while the avoid-side calls were flawless;
  4. The robotics framework of "treat rumor-grade evidence as an option" was right (Joyson was the only bright spot today), but the branch should not have been given A- strength on that account; branch strength and position discipline must be written separately.

---

3. Today's Main-Line Validation

Conclusion: got the "fading" ones right (media / defense / precious metals / glass substrate / back rows), got the "repairing" ones wrong (macro positives, storage, optical modules) — today was not a repair day but a risk-off low-level defensive rotation day; the real main lines (wind power / autos / baijiu-pharma) were all outside pre-market coverage — this is the biggest blind spot of the pre-market list: it ranked purely by "holiday headline strength" and left no branch slot for the "risk-off scenario".

---

4. Limit-Up Echelons / High Boards

• Highest board: 新华传媒 (Xinhua Media, 600825, main board) 5 consecutive one-word boards (the only 5-board stock of the day, seal orders RMB 2.39 billion) — the last lone survivor amid the publishing/media fade; historical experience (Inner Mongolia Xinhua going straight to limit-down after 6 boards in 7 days) points to catch-down risk outweighing the odds of crossing through.
• 3 boards (3 stocks): 雪龙集团 (Xuelong Group, 603949, main board, auto parts / VC), 金辰股份 (Jinchen Machinery, 603396, main board, PV equipment), 福建水泥 (Fujian Cement, 600802, main board, Fujian / nuclear power) — three mutually unrelated lines, showing capital fighting its own battles with no unifying theme.
• 2 boards (3 stocks): 大业股份 (Jiangsu Daye, 603278, main board), 襄阳轴承 (Xiangyang Bearing, 000678, main board, reducers / robotics), 吉鑫科技 (Jixin Technology, 601218, main board, wind power).
• Mid-cap anchors: 江淮汽车 (JAC Motors, 600418, main board, 3 boards in 4 days, turnover RMB 4.4 billion) is the most recognizable mid-cap anchor of the day; 均胜电子 (Joyson Electronics, 600699, main board, +6.19%, turnover RMB 12.9 billion) is the trend anchor.
• Broken boards: of yesterday's 52 limit-ups, 8 continued to a board today (advance rate 15%), and the 1-into-2 advance rate was only ~3/39; the media high-flyer group (内蒙新华 (Inner Mongolia Xinhua), 泰慕士 (Jiangsu Taimushi), 天威视讯 (Topway Video)) all hit limit-down, 康强电子 (Kangqiang Electronics) failed at 2-into-3, and 雷科防务 (Leike Defense), 平潭发展 (Pingtan Development), 奥佳华 (OGAWA) all broke — the relay ecosystem is near the ice point.
• First-board structure: 26 of the 33 limit-ups are first boards, concentrated in autos, power, chemicals, pharma, and consumption — a textbook "low-level defensive + event first-board" structure with no high-level crowding.

---

5. Capital Flows

• Northbound: since the mechanism adjustment in August 2024, exchanges no longer disclose daily net buying (the Eastmoney interface shows 0); today's northbound net buy/sell cannot be confirmed; what can be confirmed is that Stock Connect underlying stocks fell far more than they rose (Shanghai Connect: 283 up / 1325 down; Shenzhen Connect: 254 up / 1603 down, Eastmoney basis).
• Southbound: net sell of about RMB 4.3 billion (Shanghai-Hong Kong Stock Connect -RMB 773 million, Shenzhen-Hong Kong Stock Connect -RMB 3.531 billion, Eastmoney basis) — domestic and foreign capital in a synchronized risk-off.
• Dragon-tiger list: as of this report's writing (~16:00), the exchanges had not yet published the 9/28 dragon-tiger data; active stocks and seat preferences cannot be confirmed for now — no fabrication.
• Main-force direction (inferred from the limit-up + limit-down pool structure): buying concentrated in power, autos, pharma, chemical products, film & TV and other low-level directions; the selling epicenters were components / PCB (江海股份 (Jianghai), 海星股份 (Haixing), 泰晶科技 (TKD) among others limit-down), communication equipment (世嘉科技 (Shijia Technology), 长江通信 (Yangtze Communication) limit-down), consumer electronics, precious metals (Shandong Gold limit-down), CPO / optical modules — consistent with the "electronics chain main-force continuous net outflow" trend since 9/24, and accelerating. (Note: Eastmoney's industry fund-flow interface malfunctioned today; the main-force net-inflow sector amount ranking could not be obtained, so the limit-up/limit-down pool structure is used for inference instead; detailed amounts are missing for now.)

---

6. Tomorrow's Outlook (2026-09-29, second-to-last trading day before the holiday)

① Main-Line Persistence Judgment

The fading stage is confirmed; overall persistence is poor. Three reasons: ① 56 limit-downs today and a 15% advance rate put the relay ecosystem near the ice point, with short-term capital risk appetite collapsed; ② tomorrow is the second-to-last selling day for cash-over-holiday funds, and the selling-pressure structure has not disappeared; ③ the root of the valuation derating (30Y UST at 5.5%, October rate-hike probability 69.7%) shows no sign of lifting. The only exception worth watching is the auto chain (JAC Motors as the mid-cap anchor + the Xuelong / Xiangyang echelon): it has both defensive attributes and event drivers, and is the only continuation candidate worth tracking tomorrow.

② Key Focus (branch + stocks + validation points)

③ Tomorrow's Avoids

• 新华传媒 (Xinhua Media, 600825, main board) 5-board lone survivor: the biggest catch-down risk in the whole market; sentiment weather-vane only, never relay;
• All 3-board-and-above names (unless Xuelong / Jinchen / Fujian Cement gap up and seal within seconds; a low open sealed tight is a second sentiment ice-point signal);
• Precious metals (the falling knife after Shandong Gold's limit-down), CPO / components / PCB back rows (main-force net outflow accelerating);
• Glass-substrate back rows (if Wufang breaks its board, pure-mapping names like Woge / Kaisheng fall first).

④ Input Notes for Tomorrow's Pre-Market List

1. Rewrite the tone: from "repair window" to "defensive in the fading stage + low-level event-driven"; delete all "gap-up repair" presets;
2. Expand the branch pool: must reserve slots for the risk-off scenario — defensive consumption (baijiu / pharma / power), the auto chain, wind power — filtered by "low level + a same-day catalyst"; do not blank the whole pool for being "low-grade themes" (today's pre-market missed them all);
3. Raise the bar for offensive recommendation slots: while the 30Y UST yield holds above 5.5%, high-PE growth (storage / optical modules / AI hardware) goes only into "watch slots", never "priority deep-dive", unless a hard volume-contraction-and-stabilization signal appears;
4. Slow variables to keep tracking: 30Y / 10Y UST yields, US-Iran dialogue progress (decides the direction of tanker two-way volatility), Micron's 9/30 earnings (the biggest pre-holiday variable for the storage chain), fund behavior at the 9/30 final pre-holiday position-reduction window;
5. Rumor verification for 均胜电子 (Joyson Electronics): if the Tesla plant audit / orders produce no announcement-grade progress, robotics drops back to B grade in the list; if there is progress, keep it as the only offensive watch item.

---

⚠️ Data-fetch failure log (credibility notes for this report):

• The Eastmoney push2 clist interface (industry sectors / concept sectors / whole-A quotes) returned persistent 502 / disconnects throughout the entire fetch window; stock_board_industry_name_em, stock_board_concept_name_em, stock_zh_index_spot_em, stock_sector_fund_flow_rank all failed (≥3 retries, intervals up to 60s, including re-tries after sleep 30/60). → the industry/concept gain rankings and the main-force net-inflow sector amount ranking are missing; the body infers from the limit-up/limit-down pool structure instead.
• Dragon-tiger: ak.stock_lhb_detail_em parsing failed (NoneType); the direct datacenter-web API returned an "empty data" error (code 9201), and a 16:00 retry still returned empty — the exchange had not yet published the 9/28 dragon-tiger list; the body truthfully marks it "cannot be confirmed for now".
• Northbound: the interface returned 0, a caliber issue (net-buy disclosure stopped since 2024-08), not a malfunction.
• Fallback sources: indices from Tencent qt.gtimg.cn (after 15:30); two-market turnover and advance/decline counts cited from the Gelonghui market close; limit-up / limit-down / broken-board / yesterday's-limit-up pools are Eastmoney official closing pools (read at 15:30, closing basis). akshare version 1.18.79.
• Risk: the sector-structure inference has not been validated against the industry gain ranking; the direction is reliable but precision is limited; for exact sector rankings, add a section after the dragon-tiger disclosure (~after 18:30).

⚠️ Risk notice: this recap is a post-close information collation and observation only and does not constitute investment advice. Data may differ in timeliness or caliber; please refer to exchange/company disclosures and do not use this directly as a trading basis. Pre-holiday liquidity and overseas events (UST yields, US-Iran dialogue, Micron earnings) can invalidate this report's judgments quickly.