Quantifying A-Share Market Mood

Why breadth beats the index

An index can rise 1% while 3,000 stocks fall — heavyweight names distort it. Advance/decline breadth shows what the majority of stocks actually did, which matches how retail investors experience the market.

Likewise, limit-up counts read money-making effect directly, and turnover reads participation.

Step 1: today’s breadth (free)

`/v1/changedist` is the recommended entry for breadth, and it is the current session figure.

import requests

r = requests.get("https://api.ashareapi.com/v1/changedist", timeout=10)
b = r.json()
rows = b.get("structured") or b["data"]

for row in rows:
    print(row)   # advances/declines, limit-ups/downs, turnover, distribution

⚠️ The caveat: current session vs T-1

• `/v1/changedist` = current session — use this for today’s mood
• `/v1/market-overview?type=updown` = T-1 (the response states its data date; it can lag by one trading day) — the two will not match
• Do not treat T-1 breadth as today’s — this is the most common mistake

Step 2: market profile and style rotation (free)

# Overall market profile
print(requests.get("https://api.ashareapi.com/v1/market-overview",
                   params={"type": "summary"}, timeout=10).json()["data"])

# Style rotation (large vs small cap, growth vs value)
print(requests.get("https://api.ashareapi.com/v1/market-overview",
                   params={"type": "rotation"}, timeout=10).json()["data"])

# Valuation percentile (CSI All Share PE/PB/PS vs history)
print(requests.get("https://api.ashareapi.com/v1/market-overview",
                   params={"type": "valuation"}, timeout=10).json()["data"])

• `type=summary` profile · `trade` close · `interval` multi-period · `technical` index technicals
• `type=margin` margin trading · `valuation` percentiles · `rotation` style rotation

Step 3: what money is chasing (free)

The hot list reflects retail attention. Read it with breadth: good breadth + concentrated hot list = a clear main line; poor breadth + scattered hot list = fading mood.

hot = requests.get("https://api.ashareapi.com/v1/hot",
                     params={"limit": 10}, timeout=10).json()
for row in hot.get("structured") or hot["data"]:
    print(row)   # attention ranking + change %

Turn three readings into a "market temperature"

Three rules to write down

• One day’s mood does not predict tomorrow — mood describes a state, it is not a buy/sell signal
• Always compare with history — is 80 limit-ups a lot? Only a 20-day average tells you
• Separate volume-up from volume-down — a rally on rising volume and one on falling volume are different events

Common mistakes and boundaries

• Mixing T-1 with the current session → use `/v1/changedist` for today, not `market-overview?type=updown`
• Only looking at limit-up counts → without the failure rate you can misread (high limit-ups + high failure = unstable mood)
• Units → `volume` is in lots, `amount` in CNY; do not mix them
• Anonymous limits are 5 req/min → this tutorial uses only free endpoints; batch jobs should `sleep()` or upgrade