DEV Community

Felixwang007
Felixwang007

Posted on

7 Stocks Hit the 20% Limit Today in China — My Free AI Scanner Caught All of Them by 9:35 AM

Every morning at 9:30 AM Beijing time, a script on my machine quietly wakes up, pulls live quotes for all 5,000+ A-share stocks, and runs a three-pillar scan: technicals, fundamentals, and market sentiment. It takes about 90 seconds. This morning it did something worth writing about.

What happened today (Aug 17, 2026)

The market opened strong and stayed strong:

Index Close Change
Shanghai Composite 3,960 +0.84%
Shenzhen Component 14,534 +1.26%
ChiNext (Growth) 3,675 +1.33%
CSI 300 4,701 +0.76%

But the interesting action was in the limit-up board. A-share rules allow a 20% daily move on ChiNext (300xxx) and STAR Market (688xxx) stocks. This morning, seven of the top ten gainers all closed at the 20% limit — that's a rotation signal, not just a green tape day.

The scanner's output, 9:35 AM

Top movers flagged by the scan (name, price, change, turnover):

N 恒兴 (new listing):  32.20  +101.0%  turnover 78.6%
创达新材 (BJ):         60.88  +23.9%   turnover 12.6%
戴维医疗 (300314):     14.74  +20.0%   turnover 17.9%
天山生物 (300313):     10.73  +20.0%   turnover 11.6%
聚和材料 (688503):     95.68  +20.0%   turnover 10.4%
太辰光   (300570):    193.80  +20.0%   turnover 12.4%
苏州天脉 (301626):    307.08  +20.0%   turnover 7.7%
Enter fullscreen mode Exit fullscreen mode

How to read this tape (the part that's actually useful)

Three things stand out, and they map directly to the three pillars of the system:

1. Technicals — turnover is the tell. A 20% limit-up with 10–18% turnover means the move is contested: real money is rotating in, but so is a lot of short-term supply. Contrast that with 中石科技's limit-up at only 1.6% turnover — that's a locked board, sellers simply didn't show up. Low-turnover limit-ups are structurally stronger than high-turnover ones. When you see a high-turnover limit-up, expect a follow-up test tomorrow.

2. Sentiment — where the money is going. A new listing doubling on its debut (N 恒兴, +101%, 78% turnover) plus a cluster of 20% boards across ChiNext and STAR tells you risk appetite is ON. In this regime, breakout strategies on 300xxx/688xxx names tend to work better than dividend/value plays. Regime detection is 80% of the game — indicator choice comes second.

3. Fundamentals — the filter that keeps you out of traps. The scanner cross-references each mover against earnings, ROE history, and PEG. 戴维医疗 and 天山生物 are classic sentiment names (medical devices / agriculture themes) — fine for a 1–2 day trade if you respect the stop, but their fundamental scores keep them off the swing list. The system's job is to separate "momentum with a floor" from "momentum with a cliff."

The one-command workflow

This isn't a black box — it's a free open-source skill for AI coding agents (works with Claude Code, Codex, etc.):

# fetch the whole market, score every ticker, output a ranked watchlist
agent-run a-share-stock-analysis scan --sort momentum
Enter fullscreen mode Exit fullscreen mode

Under the hood it's plain Python: Sina/Tencent free quote endpoints (no paid API key, no rate-limit drama), ~24 technical indicators (MACD, KDJ, RSI, Bollinger, volume-price divergence), and a weighted scoring model. You can read every line of the code and change the weights yourself.

Why I open-sourced this

I got tired of two things: (1) paid tools that hide their logic, and (2) advice posts with zero reproducible data. So the skill ships with the live scanner, the scoring rules, and a daily watchlist generator — you point it at today's market and it gives you today's numbers, not last month's screenshots.

If you trade A-shares or just want to see how a free agent skill does real market surveillance every morning, star the repo and run the scan tomorrow at 9:30. The tape will thank you.


Disclaimer: Educational content, not investment advice. A-share trading involves risk; always do your own due diligence and respect position sizing.

Top comments (0)