It's Monday lunchtime in Shanghai, and the morning session just delivered the kind of tape that separates people who watch the index from people who read the market.
Here's what actually happened in the first half of today's session (2026-08-31, live quotes):
| Index / Stock | Price | Change |
|---|---|---|
| Shanghai Composite | 3,944.46 | -0.20% |
| Shenzhen Component | 13,812.99 | -1.00% |
| ChiNext | 3,380.15 | -1.29% |
| BYD | Β₯88.47 | -4.17% |
| CATL | Β₯358.17 | -2.80% |
| LONGi Green Energy | Β₯11.96 | -3.78% |
| Seres | Β₯49.30 | -2.55% |
| Kweichow Moutai | Β₯1,296.08 | -0.10% |
| Wuliangye | Β₯71.20 | -0.43% |
The index printed "a mild dip." But underneath, a violent rotation was happening: new-energy (EVs + solar) got sold hard, while defensive white-liquor barely blinked. That gap β BYD down 4% while Moutai is flat β is the signal. The index is just the noise.
Why the headline number lies to you
The Shanghai Composite is dominated by financials and old-economy heavyweights. When money rotates out of growth and into defensives, the index can look "stable" while a growth-heavy portfolio gets shredded. Read the index for mood; read the divergence for direction.
Today's tape is a textbook risk-off rotation: risk appetite is shrinking at the margin, and capital is hiding in the names that pay dividends and survive recessions.
The three-pillar way to read this
My A-share scanner doesn't try to predict the market. It triangulates three independent signals and waits for them to agree:
- Technical β price/volume structure (MACD, KDJ, RSI, Bollinger, volume-price). Says what is happening.
- Fundamental β ROE decomposition, PEG, valuation percentile. Says why it might be justified.
- Intelligence β ιΎθζ¦ (dragon-tiger list of top buy/sell seats), block-trade and capital-flow data, sector rotation, policy catalysts. Says who is doing it and where the money is going.
On days like today, pillar #3 earns its keep. The dragon-tiger list and block-trade data will tell you this afternoon whether the EV selling was institutional rotation or a few hot-money desks panic-exiting. Those two scenarios have completely different follow-throughs.
Three rules you can use right now (no paid data required)
Rule 1 β When the index and the sectors disagree, trust the sectors. Compute the gap between ChiNext and the Shanghai Composite. Today it's ~1.1 percentage points of underperformance. If that spread widens three sessions in a row, it's a rotation, not a blip.
Rule 2 β Don't buy the dip in the sector being sold until volume confirms a floor. A falling knife with rising volume is still a falling knife. Wait for a session of shrinking volume with a higher low β that's the first footprint of distribution ending.
Rule 3 β In risk-off rotations, the rotation target is usually the previous laggard. Money doesn't leave the market; it moves. Today's defensiveness (liquor, banks, dividends) is the same money that was chasing EVs in July. Watch where volume accumulates over the next 3 sessions β that's the next leg.
How the scanner automates this
The scanner pulls free quote data (Tencent/Sina endpoints β no API key, no cost), computes 24 indicators per stock, scores each of the three pillars 0-10, and only flags a stock when all three agree. The Aug 31 morning scan automatically flagged the rotation by scoring new-energy names down on technicals while defensive names held their pillar scores.
The full pipeline (Python, zero paid dependencies) is documented in my awesome-content-tools repo β a collection of self-updating tools including the A-share scanner, a GitHub Trending aggregator, and the market-data scripts used for this article. If you're on the Chinese skill marketplace θΎθ― (xiaping.coze.com), you can also find my published data-analysis and stock tooling skills there.
No paid API, no black box, no crystal ball β just three independent signals forced to agree before you act. That's the whole edge.
