Tuesday was ugly for Chinese A-shares. Here are the hard numbers, then the interesting part: what 800+ retail investors actually said while it happened.
The numbers
- Shanghai Composite: 3658.03, -1.48%
- Shenzhen Component: -2.96%
- ChiNext: -4.37% (lost the 2600 level)
- STAR 50: -2.78%
- Limit-down stocks: 49 vs limit-up: only 33
- Margin trading balance: -5.53 billion yuan in a single day (deleveraging)
- The 4 biggest equity ETFs (ChiNext ETF, STAR50 ETF, CSI300 ETF, KWEB): -5.51 billion yuan net outflow
The sentiment paradox
I run a daily sentiment scraper over Chinese retail forums (Sina Guba, East Money Guba). Yesterday's 24-hour window pulled 800+ comments across 5 platforms. The split:
- Bull-weighted platform (Sina Guba): 55.0/100 - retail investors still buying the dip
- Bear-weighted platform (East Money): 42.8/100 - panic, but easing after hours
- Composite index: 50.5 - right on the bull/bear line, down from 53.1 the day before
One comment captured the mood better than any chart:
"If you don't drop, how can I get in? Give me a chance to board."
Another one, posted after hours:
"Brutal today. Dozens of stocks at limit-down."
The takeaway: retail sentiment cracked but did not break. Margin deleveraging says institutions moved first. When the two converge - either both capitulating or both re-risking - that's the signal I watch.
How I collect this (the AI agent part)
The whole pipeline - scraping forums, scoring sentiment, generating the daily report - runs on AI agents I built myself. No vendor, no black box.
I turned that system into a course: AI Agents from Scratch - 12 Lessons, Zero Code, built for non-programmers. You build real agents: scrapers, sentiment scorers, market monitors.
Check it out here: AI Agents from Scratch
I post a daily China market sentiment read on Substack - data-backed, not vibes.
