Zhongji Innolight (300308) rose 10.28% that day with a turnover of 235.7 yi, the largest turnover among the names on the list, yet the company made no announcement at all. Which piece of already-public information was this 235.7 yi buying?
Zhongji Innolight (300308) rose 10.28% that day with a turnover of 235.7 yi, the largest turnover among the names on the list, yet the company made no announcement at all. Which piece of already-public information was this 235.7 yi buying?
中际旭创(300308)当天涨10.28%、成交235.7亿，是名单里成交额最大的一只，公司却没有任何公告，这235.7亿买的是哪条已经公开的信息？
The desk answers已核实Walnut, from public filings
The company truly had not a single announcement that day; in Cninfo's announcement list, the latest one for Zhongji Innolight still sits at October 23. So the 235.7 yi bought not that day's news: it bought a voluntary announcement published at noon on October 23, a date already written clearly on October 15, a batch of similar disclosures handed in by peers in the same week, and the rest was made up by the sector rebound.
Not just this one stock rose that day
October 28 was a day of overall rebound for tech stocks. The Shanghai Composite rose only 0.18%, the ChiNext Index rose 3.40%, the Shenzhen Component rose 2.02%, and STAR 50 rose 2.45%; across the whole market 4156 stocks rose and 73 hit the limit up, with turnover of 1.64 wan yi, 1693 yi more than the previous day. The SW Communications sector rose 5.01%.
Overnight external markets also cooperated: the Nasdaq rose 1.18%, the S&P rose 0.60%, and the Dow rose 0.03%. External markets can explain sentiment, but cannot explain a single stock's 10%.
| Company | 10/8→10/27 (%) | 10/28 (%) | 10/28 turnover (yi CNY) |
|---|---|---|---|
| Zhongji Innolight | (26.2) | 10.28 | 235.7 |
| Eoptolink | (37.2) | 8.08 | 150.9 |
| TFC Optical Communication | (19.8) | 6.79 | 73.7 |
| Accelink Technologies | (21.1) | 3.03 | 32.6 |
| T&S Communications | (26.7) | 5.96 | 21.4 |
| Yuanjie Semiconductor | (35.4) | 0.77 | 58.8 |
| CIG Cambridge | (28.3) | 4.91 | 28.4 |
The sector probably gave about 5 points, and the remaining 5 points have to be found in the company itself. And it is not "the deeper the fall, the higher the bounce": Eoptolink fell 37.2% in this stretch, a notch deeper than Innolight's 26.2%, but on the 28th it rose only 8.08%.
First: that voluntary announcement at noon on October 23
The timing is crucial. That announcement was released at 12:30 on October 23, during trading hours. That morning Innolight was still falling, hitting a low of 554.34, then it was pulled up all the way in the afternoon to close at 613.58, up 3.82% for the day, with turnover of 279.2 yi.
Announcement No. 2026-100, two sentences are the substance:
On overseas markets, orders from major customers have been placed through 2027, and some key customers have given clear guidance on 2028 demand
At present, downstream demand is growing rapidly, manufacturers with large-scale delivery capability are limited, the market price system is healthy, and product quotes to customers are relatively rational
There is another item that is easy to skip: the 565.3 wan shares repurchased in September (0.48% of total share capital, costing 49.97 yi), whose use is to be changed from employee stock ownership to full cancellation, to be deliberated by the board on October 29. Cancellation is an action that directly lifts earnings per share.
On the same day, Eoptolink also issued the same type of announcement (No. 2026-061), saying that 1.6T has entered a stage of continuous volume ramp-up, orders on hand are sufficient, and it holds positive expectations for this year and next. Two leading companies simultaneously making voluntary disclosures on the second day of the plunge was a collective response after the whole industry was hit—it explains the sector, but cannot explain the several points by which Innolight rose more than others.
Second: a date made public as early as October 15
Innolight is A+H, with Hong Kong code 3308. Hong Kong rules require that when the board meets to review results, the date must be announced first. So the H-share announcement on October 15 nailed it down: the board will meet on October 29 to review the 2026 first three quarters results and their publication, and at the same time review the cancellation of repurchased shares.
A-shares have no such requirement. Thus a situation emerged: Cambridge Technology, also A+H, also announced its third-quarter report date in advance, which is October 30; Innolight's is October 29—it is the earliest to reveal results in this batch.
Late October is exactly the window for optical module results. Accelink already released its third-quarter report on October 26: single-quarter revenue 50.74 yi, net profit attributable to parent 7.02 yi, up 54% and 102% year-on-year.
Innolight's third-quarter report is not out yet, and the date is known to be October 29. Those buying this 235.7 yi are buying a known event whose results come out the next day.
The same announcement was worth only 3.82% three days earlier
This is the real answer to this question.
For the October 23 announcement, the market gave 3.82% that afternoon, using up 279.2 yi. The next two days gave it all back: October 26 fell 2.96%, October 27 fell 2.94%, and turnover dropped to 169.8 yi and 140.1 yi.
On October 28, the same information, 235.7 yi—43.5 yi less than October 23—pushed the price up 10.28%.
The information did not change; what changed was the selling. Margin financing balance had already fallen for 13 consecutive declines, with net financing outflow of over 1000 yi in October; during the same period, equity ETFs saw net inflows of over 3500 yi in the past month (International Finance News, October 28, citing Ziruixing Investment researcher Fu Hao). Those using leverage were running, those buying indexes were entering, and Innolight is a name this batch of money cannot avoid.
Hong Kong shares rose on the same day too: 3308 closed at HK$785, up 8.28%, with turnover of 28.5 yi Hong Kong dollars. Both markets moved together, showing this is not a matter of local A-share funds.
Which part has not yet been priced
The third-quarter report itself.
| Period | Single-quarter revenue (yi CNY) | Single-quarter net profit attributable to parent (yi CNY) |
|---|---|---|
| 2025Q3 | 102.16 | 31.37 |
| 2026Q1 | 194.96 | 57.35 |
| 2026Q2 | 222.81 | 79.17 |
| 2026Q3 | Not disclosed | Not disclosed |
2026Q2 single-quarter gross margin 46.42%, which is back-calculated: first-half revenue 417.78 yi × 46.25%, minus first-quarter 194.96 yi × 46.06%, the gross profit difference 103.44 yi ÷ 222.81 yi.
The sentence in the October 23 announcement that "orders have been placed into 2027, and some customers have given 2028 guidance" has so far not been verifiable by any single financial report. The first verifiable time point is October 29, the second is contract liabilities and inventory in the annual report. So strictly speaking, what was bought on October 28 was expectation, not facts already on the statements.
How a smart person would rebut me
He would say: this is pure beta. On the day 4156 stocks rose, Zhongji Innolight's circulating market cap was 7062 yi, it is a must-hold for broad-based ETFs and public funds, its share price fell from the June 22 high of 1418.56 to 577.89 on October 27, more than halved, so its rebound elasticity was naturally the largest; the October 23 announcement just happened to appear three days before the rebound, with no causal relationship.
This rebuttal carries weight, because the public funds' third-quarter reports show it is still the most tightly held stock, and such stocks are naturally swept up first by index funds.
The difference is in the details: if you really rank by "fell deep + large weight", Eoptolink fell deeper yet bounced less; if you really rank by "has a due event", Cambridge Technology's reveal was one day later, and on the 28th it rose only 4.91%.
Under what circumstances this does not hold
- If on October 29 the third-quarter report's single-quarter net profit attributable to parent is below 79.17 yi (no quarter-on-quarter increase), or the single-quarter gross margin falls below 46.42%, then this front-running was a wrong buy, and this 10.28% has to be given back.
- If within three trading days after the third-quarter report comes out, turnover falls back below 150 yi, and the share price retreats to the level before the third-quarter report disclosure, it shows that the 28th was just one pulse of sector beta, and that piece of company information was never truly priced.
- Conversely, if single-quarter net profit attributable to parent is clearly above 79.17 yi, and traces of "2027 orders already placed" can be seen in the third-quarter report's contract liabilities and inventory, then what was bought on the 28th was not just beta.
Sources: Zhongji Innolight "Announcement on the Company's Recent Production and Operation Situation", 2026-10-23, announcement no. 2026-100; Zhongji Innolight "H-share Announcement (Date of Board Meeting)", 2026-10-15, Hong Kong stock code 3308; Eoptolink "Announcement on the Company's Recent Production and Operation Situation", 2026-10-23, announcement no. 2026-061; Accelink Technologies "2026 Third Quarter Report", 2026-10-26; Cambridge Technology "H-share Announcement—Date of Board Meeting", 2026-10-20, Hong Kong stock code 6166; Zhongji Innolight 2026 Semi-Annual Report (disclosed August 2026, first-half revenue 417.78 yi, net profit attributable to parent 136.51 yi); Tencent Finance quotes 300308, 03308.HK, 2026-10-28 close; "21st Century Business Herald" "Latest Fund Position-Adjustment Path Exposed: Zhongji Innolight Retains No. 1 Heavy Holding", 2026-10-28; "International Finance News" "Tech Stocks Strengthen! A-shares Welcome a Technical Rebound, Three Major Signals Determine Sustainability", 2026-10-28.
公司当天确实一条公告都没有，巨潮的公告列表里，中际旭创最新的一份还停在 10 月 23 日。所以这 235.7 亿买的不是当天的消息：它买的是一份 10 月 23 日中午发的自愿性公告、一个 10 月 15 日就写明了的日期、一批同行在同一周交出的同类披露，剩下的由板块反弹补上。
当天涨的不只是这一只
10 月 28 日是科技股整体反弹的一天。沪指只涨 0.18%，创业板指涨 3.40%，深证成指涨 2.02%，科创50 涨 2.45%；全市场 4156 只上涨、73 只涨停，成交额 1.64 万亿，比前一天多 1693 亿。申万通信板块涨 5.01%。
隔夜外围也配合：纳指涨 1.18%、标普涨 0.60%、道指涨 0.03%。外围能解释情绪，解释不了单只股票 10%。
| 公司 | 10/8→10/27（%） | 10/28（%） | 10/28 成交额（亿元） |
|---|---|---|---|
| 中际旭创 | (26.2) | 10.28 | 235.7 |
| 新易盛 | (37.2) | 8.08 | 150.9 |
| 天孚通信 | (19.8) | 6.79 | 73.7 |
| 光迅科技 | (21.1) | 3.03 | 32.6 |
| 太辰光 | (26.7) | 5.96 | 21.4 |
| 源杰科技 | (35.4) | 0.77 | 58.8 |
| 剑桥科技 | (28.3) | 4.91 | 28.4 |
板块大概给了 5 个点，剩下 5 个点得从公司自己身上找。而且不是「跌得越深弹得越高」：新易盛这段跌了 37.2%，比旭创的 26.2% 深一截，28 日只涨 8.08%。
第一条：10 月 23 日中午那份自愿性公告
时间点很关键。那份公告是 10 月 23 日 12:30 发的，盘中。当天早上旭创还在跌，最低 554.34，下午一路拉起来收 613.58，全天涨 3.82%，成交 279.2 亿。
公告编号 2026-100，两句话是干货：
海外市场方面，主要客户订单已下至 2027 年，部分重点客户已给出 2028 年需求的明确指引
目前下游需求快速增长，具备大规模交付能力的厂商有限，市场价格体系健康，给客户的产品报价相对理性
还有一条容易被跳过的：9 月回购的 565.3 万股（占总股本 0.48%，花掉 49.97 亿），用途要从员工持股改成全部注销，10 月 29 日董事会审议。注销是直接抬每股收益的动作。
同一天，新易盛也发了同一类型的公告（编号 2026-061），说 1.6T 已进入持续放量阶段、在手订单充足、对今明两年持积极预期。两家龙头在暴跌第二天同时自愿披露，这是全行业被砸之后的集体回应——它解释板块，解释不了旭创比别人多涨的那几个点。
第二条：一个 10 月 15 日就公开的日期
旭创是 A+H，港股代号 3308。香港规则要求董事会开会审议业绩，日期必须先公告。所以 10 月 15 日那份 H 股公告就把话说死了：10 月 29 日开董事会，审议 2026 年前三季度业绩及其刊发，同时审议回购股份注销。
A 股没有这条要求。于是出现一个局面：同为 A+H 的剑桥科技，三季报日期也提前公告了，是 10 月 30 日；旭创是 10 月 29 日——它是这一批里最早开奖的那个。
十月底正是光模块交卷的窗口。光迅科技 10 月 26 日已经出了三季报：单季收入 50.74 亿、归母 7.02 亿，同比增 54% 和 102%。
旭创的三季报还没出，日期已知在 10 月 29 日。买这 235.7 亿的人，买的是一个隔天就开奖的已知事件。
同一条公告，三天前只值 3.82%
这才是这道题真正的答案。
10 月 23 日那份公告，市场当天下午给了 3.82%，用掉 279.2 亿。接下来两天全吐回去：10 月 26 日跌 2.96%，10 月 27 日跌 2.94%，成交额掉到 169.8 亿、140.1 亿。
10 月 28 日，同样的信息，235.7 亿——比 10 月 23 日还少 43.5 亿——把价格推高 10.28%。
信息没变，变的是卖盘。两融余额已经 13 连降，10 月融资净流出过千亿；同期股票型 ETF 近一个月净流入超 3500 亿（《国际金融报》10 月 28 日引述资瑞兴投资研究员傅豪）。加杠杆的在跑，买指数的在进，旭创是这批钱绕不开的名字。
港股同一天也涨：3308 收 785 港元，涨 8.28%，成交 28.5 亿港元。两地一起动，说明这不是 A 股本地资金的事。
还没被定价的是哪一块
三季报本身。
| 期间 | 单季收入（亿元） | 单季归母（亿元） |
|---|---|---|
| 2025Q3 | 102.16 | 31.37 |
| 2026Q1 | 194.96 | 57.35 |
| 2026Q2 | 222.81 | 79.17 |
| 2026Q3 | 未披露 | 未披露 |
2026Q2 单季毛利率 46.42%，是倒算出来的：上半年收入 417.78 亿 × 46.25%，减去一季度 194.96 亿 × 46.06%，毛利额差额 103.44 亿 ÷ 222.81 亿。
10 月 23 日公告里「订单下到 2027 年、部分客户给了 2028 指引」这句话，至今没有任何一张财报能验证它。第一个可验证的时点是 10 月 29 日，第二个是年报里的合同负债和存货。所以严格讲，10 月 28 日买的是预期，不是已经落表的事实。
一个聪明人会怎么反驳我
他会说：这就是纯 beta。4156 只上涨的日子里，旭创流通市值 7062 亿、是宽基 ETF 和公募的必配，股价从 6 月 22 日高点 1418.56 跌到 10 月 27 日 577.89，腰斩还多，反弹弹性自然最大；10 月 23 日那份公告只是恰好出现在反弹前三天，没有因果关系。
这个反驳有分量，因为公募三季报显示它仍是被抱得最紧的那只，这类股票本来就会被指数资金优先扫到。
区别在细节：真按「跌得深+权重大」排，新易盛跌幅更深却弹得更少；真按「有到期事件」排，剑桥科技晚一天开奖，28 日只涨 4.91%。
什么情况下这条不成立
- 10 月 29 日三季报单季归母低于 79.17 亿（环比不增），或单季毛利率跌破 46.42%，那这次抢跑就是买错了，这根 10.28% 要还回去。
- 三季报出来后三个交易日，成交额跌回 150 亿以下，且股价退回三季报披露前的水平，说明 28 日只是板块 beta 的一次脉冲，公司那条信息始终没被真正定价。
- 反过来，若单季归母明显高过 79.17 亿，且三季报的合同负债、存货上能看到「2027 年订单已下」的痕迹，那 28 日买到的就不只是 beta。
来源：中际旭创《关于公司近期生产经营情况的公告》，2026-10-23，公告编号 2026-100；中际旭创《H 股公告（董事会会议日期）》，2026-10-15，港股代号 3308；新易盛《关于公司近期生产经营情况的公告》，2026-10-23，公告编号 2026-061；光迅科技《2026 年三季度报告》，2026-10-26；剑桥科技《H 股公告—董事会会议日期》，2026-10-20，港股代号 6166；中际旭创 2026 年半年度报告（2026 年 8 月披露，上半年收入 417.78 亿、归母 136.51 亿）；腾讯财经行情 300308、03308.HK，2026-10-28 收盘；《21 世纪经济报道》《基金最新调仓路径曝光：中际旭创蝉联第一重仓》，2026-10-28；《国际金融报》《科技股走强！A 股迎来技术性反弹，三大信号决定持续性》，2026-10-28。