Reading A-share block trades: discount rate & institutional seats

1. What a block trade is (30 seconds)

A block trade is a large transfer executed through the exchange block-trading system by mutual agreement. Because it bypasses continuous auction: it does not appear on the intraday tape, and the trade is disclosed the next day (T+1).

That is the value: you can see who took how much, at what price, from whom — information the normal order book does not give.

• Sellers wanting to offload → use blocks (fast exit without crashing the price)
• Buyers wanting size → take blocks (faster than buying in the market)
• Discount = seller concedes for speed (common); premium = buyer is eager (rare, stronger signal)

2. One request to fetch it

`/v1/block-trade` is a paid endpoint. Only two parameters: `code` (optional) and `date` (optional; default = latest).

import requests

BASE = "https://api.ashareapi.com/v1"
H = {"Authorization": "Bearer ct-your-key"}

r = requests.get(f"{BASE}/block-trade",
                 params={"code": "sh600519"},     # Kweichow Moutai
                 headers=H, timeout=30)
body = r.json()

# Note: data may be an array; do NOT write body["structured"] (KeyError)
rows = body.get("data") if isinstance(body.get("data"), list) else body.get("structured")
if not body.get("ok"):
    print("no block trade that day (not an API failure)")   # see section 4
else:
    for x in rows:
        print(x["TurnoverPrice"], x["CloseDiscountRate"],
              x["BuySalesDepartment"], x["SellSalesDepartment"])

• `data` may be an array or Markdown — branch on type (or fall back to `body.get("structured")`)
• `ok:false` with an empty array = no block trade that day (a normal result, not a fault) — see section 4

3. The seven fields

Volume caveat: `TurnoverValue` is a single trade. Sum same-day trades for a daily total. The upstream reports "number of trading days with blocks" and "total trade count" as two different measures — state which one you use.

4. The discount rate is the most informative field

`CloseDiscountRate` is relative to the closing price of the day:

Key point: discounts are the norm — large stakes are hard to sell at market, so a discount alone is not a bearish signal. What carries more information is an unusually deep discount (e.g. -8%) combined with who bought (institutional seat vs an ordinary branch).

5. Live data (2026-09-29)

How to read it: the Hengrui trade — 252m at an 8.01% discount — shows a clearly motivated seller, with the buyer a brokerage branch. That combination (large + deep discount) is worth a closer look (no directional conclusion — it just shows seller urgency).

6. Scanning in bulk (find large / deep-discount trades)

CODES = ["sh600519", "sh600276", "sh601899", "sz300750",
         "sh601318", "sh600036", "sz000858", "sh600030"]

hits = []
for code in CODES:
    b = requests.get(f"{BASE}/block-trade", params={"code": code},
                     headers=H, timeout=30).json()
    if not b.get("ok"):
        continue                                   # no block trade that day
    rows = b.get("data") or b.get("structured") or []
    for x in rows:
        amt = float(x.get("TurnoverValue") or 0)
        dis = float(x.get("CloseDiscountRate") or 0)
        if amt >= 1e8 or dis <= -5:                # large or deep discount
            hits.append((code, amt / 1e8, dis, x.get("BuySalesDepartment", "")))

for code, amt, dis, buyer in sorted(hits, key=lambda z: -z[1]):
    print("%-10s %.2f bn  rate %+.2f%%  %s" % (code, amt, dis, buyer[:24]))

• Large: value ≥ 100m CNY (big enough to shift the shareholder structure)
• Deep discount: rate ≤ -5% (clear seller concession)
• Combined: "large + deep discount + institutional buyer" is a relatively rare and informative mix

7. The date parameter (historical days)

`date=YYYY-MM-DD` queries a specific day — non-trading days fall back to the most recent trading day (measured: `2026-09-26`, a Saturday, returned `2026-09-24`).

b = requests.get(f"{BASE}/block-trade",
                 params={"code": "sh600276", "date": "2026-09-26"},
                 headers=H, timeout=30).json()
print(b["ok"], b.get("data"))

• Measured: the first table (date / close and other meta) follows the date, falling back to the latest trading day
• Detail table behaviour: repeated queries returned the same trade details in our tests — either that stock had only one block, or the detail table only returns the latest. Verify with a sample whose days clearly differ before assuming multi-day history

8. Two common misreads (important)

• 1. `ok:false` does not mean the API is broken — it means no block trade for that stock on that day (a normal result). Most stocks have none on most days. To check the endpoint is fine, try a liquid large cap (e.g. `sh600519`)
• 2. A discount is not automatically bearish — large disposals need a discount to clear fast. "Has a discount" is normal; "unusually deep discount + institutional buyer" is what deserves attention