mootdx

name	mootdx
category	data-source
description	MooTDX — TCP direct connection to TDX (通达信) for A-share K-line data. Free, no API key, no IP blocking. Supports daily/weekly/monthly/minute OHLCV + level-2 order book + tick-by-tick trades.

MooTDX (通达信 TCP 直连)

Overview

MooTDX connects directly to TDX (通达信) servers via TCP protocol, providing free A-share market data without any API key or registration. Unlike HTTP-based data sources, TCP connections are not subject to IP-based rate limiting.

• K-line: daily / weekly / monthly / 1m / 5m / 15m / 30m / 60m
• 分时图 (Minute line): per-minute price/volume/amount for a single trading day
• Level-2: 五档盘口 (5-level order book) + 逐笔成交 (tick-by-tick)
• No auth: no registration, no token, no IP blocking
• Speed: TCP direct connection, lower latency than HTTP polling

The project has a built-in MooTDX DataLoader at backtest/loaders/mootdx_loader.py. When backtesting, set source: "mootdx" or rely on source: "auto" (MooTDX is the #1 fallback for A-shares).

Quick Start

from mootdx.quotes import Quotes

client = Quotes.factory(market="std")

# Daily K-line
bars = client.bars(symbol="600519", frequency=9, start=0, offset=100)

# Minute K-line (1m=8, 5m=0, 15m=1, 30m=2, 60m=3)
bars = client.bars(symbol="000001", frequency=8, start=0, offset=50)

# Level-2 order book
orders = client.quotes(symbol=["600519", "000001"])

# Tick-by-tick
transactions = client.transaction(symbol="600519", start=0, offset=10)

分时图 (Minute Line) — Per-Minute Price Trace

The DataLoader exposes two dedicated methods for 分时图 data (not OHLCV K-line — this is the actual tick-level price trace per minute):

Single stock, single day

from backtest.loaders.mootdx_loader import DataLoader

loader = DataLoader()

# Fetch 分时图 for 贵州茅台 on a specific date
df = loader.fetch_minute_line("600519.SH", "2026-05-29")
# Returns DataFrame indexed by time, columns: price, volume, amount
# None on non-trading days or errors

Multiple stocks, date range

# Fetch 分时图 for multiple stocks over a week
result = loader.fetch_minute_lines(
    ["600519.SH", "000001.SZ"],
    "2026-05-25", "2026-05-29",
)
# Returns: {code: {date_str: DataFrame | None}}
# e.g. result["600519.SH"]["2026-05-29"] -> DataFrame(time, price, volume, amount)

Direct mootdx API (for custom use)

from mootdx.quotes import Quotes

client = Quotes.factory(market="std")

# Today's minute line (live)
df = client.minute(symbol="600519")

# Historical minute line (specific date)
df = client.minutes(symbol="600519", date="2026-05-29")

Limitations

• History: TDX server only keeps ~5–10 trading days of minute-line data
• Non-trading days: returns None (weekends, holidays, suspended)
• Format: This is NOT OHLCV — each row is one minute's price/volume/amount
• For intraday backtesting: use interval="1m" with the standard fetch() method (1-minute OHLCV bars)
• For 分时图 visualization: use fetch_minute_line() / fetch_minute_lines()

Available Intervals (K-line)

Interval	frequency	Description
1m	8	1-minute bars
5m	0	5-minute bars
15m	1	15-minute bars
30m	2	30-minute bars
60m	3	60-minute bars
1D	9	Daily bars (default)
1W	5	Weekly bars
1M	6	Monthly bars

Notes

• Code format: 600519.SH or just 600519 (the DataLoader normalizes automatically)
• No authentication or API key required
• TCP port (7709) must be accessible — no special firewall rules needed in most environments
• Not available on PyPI by default — the project installs it via pip install --no-deps mootdx to avoid the httpx version conflict