Skip to content

SKILL.md

Latest commit

 

History

History
75 lines (57 loc) · 2.86 KB

SKILL.md

File metadata and controls

75 lines (57 loc) · 2.86 KB
name miniqmt-skills
description Comprehensive XtQuant (MiniQMT) quant trading skill for Python, covering xtdata行情接口、xttrader交易接口、常量与数据结构、故障排查和实战模板. Use when implementing or debugging A股量化策略 involving historical download, realtime quote subscription, account/order/trade/position queries, order placement/cancellation, callback handling, or MiniQMT path/session/port issues.

MiniQMT / XtQuant Skill

Use this skill to deliver executable, risk-aware XtQuant solutions quickly.

Workflow

  1. Verify runtime prerequisites.
  2. Load only the reference file needed for the current subtask.
  3. Build the smallest runnable script first, then extend.
  4. Add callback-safe logic (non-blocking, recoverable).
  5. Validate with a simulated account before live trading.

Runtime Prerequisites

  • Start MiniQMT client before running Python code.
  • Use 64-bit Python 3.6 to 3.12.
  • Point path to the correct user directory:
    • Broker client: ...\\userdata_mini
    • Research client: ...\\userdata
  • Use a unique session_id per strategy process.
  • Keep reconnect interval > 3 seconds for the same session_id.

Reference Map

  • references/getting_started.md: environment, startup sequence, minimal bootstrap.
  • references/api.md: full API index (quick lookup).
  • references/xtdata_api.md: market data APIs and return shapes.
  • references/xttrader_api.md: trading APIs, callbacks, account/query flow.
  • references/data_structures.md: tick/kline fields, financial fields, trade objects, constants.
  • references/examples.md: end-to-end strategy templates.
  • references/troubleshooting.md: common errors and operational guardrails.
  • references/index.md: navigation and search shortcuts.

Implementation Guardrails

  • Download required history (download_*) before reading local historical series.
  • Use get_market_data_ex as the default read API.
  • Keep callback handlers fast; move heavy logic to queues/workers.
  • Stock order volume should follow lot rules (typically multiples of 100).
  • order_remark in MiniQMT has length limits; avoid long Chinese text.
  • Do not hardcode account IDs, tokens, or live credentials in source files.

Minimal Bootstraps

xtdata quick check

from xtquant import xtdata

xtdata.download_history_data("000001.SZ", period="1d", incrementally=True)
data = xtdata.get_market_data_ex(["close"], ["000001.SZ"], period="1d", count=5)
print(data["close"])

xttrader quick check

import time
from xtquant.xttrader import XtQuantTrader
from xtquant.xttype import StockAccount

path = r"D:\\迅投极速交易终端 睿智融科版\\userdata_mini"
session_id = int(time.time())
trader = XtQuantTrader(path, session_id)
account = StockAccount("YOUR_ACCOUNT", "STOCK")

trader.start()
print("connect:", trader.connect())
print("subscribe:", trader.subscribe(account))