KHQuant (看海量化)

CSkhQuant (看海量化交易平台) is a quantitative trading system built with Python and PyQt5. The system is organized into several functional modules that handle everything from the user interface to core trading logic and data management.

Core Modules Overview

1. Main Interface: Provides the primary GUI for strategy execution, backtesting, account management, and real-time market monitoring (GUIkhQuant.py). It also includes a dedicated interface for batch downloading and cleaning stock data (GUI.py).
2. Data Management: Includes tools for browsing local data (GUIDataViewer.py), scheduling automated data updates (GUIScheduler.py), and specialized parsers/viewers for miniQMT data formats (miniQMT_data_parser.py, miniQMT_data_viewer.py).
3. Analysis & Visualization: Features interactive charting tools (GUIplotLoadData.py) and detailed backtesting result analysis windows (backtest_result_window.py).
4. Core Framework: The engine of the system, comprising the strategy execution engine (khFrame.py), trading tools (khQTTools.py), order/cost management (khTrade.py), and risk control (khRisk.py).
5. Technical Indicators: A high-performance library for technical analysis (MyTT.py).

Before deploying, be aware of the following limitations of the KHQuant system:

• Not for High-Frequency Trading (HFT):
  • Data: MiniQMT provides 3-second snapshots rather than tick-by-tick data, which is insufficient for microsecond-level precision.
  • Latency: The Python-based architecture and MiniQMT link cannot meet sub-millisecond execution requirements.
• Historical Data Constraints: MiniQMT (broker version) has limits on historical data depth. Tick data is typically limited to ~1 month, 1/5-minute K-lines to ~1 year, and daily data is more complete. For longer historical minute/tick data, the professional 'Research version' of QMT is required.
• Market Scope: Core optimization is focused on the A-share market (Stocks, ETFs, some Futures/Options) via MiniQMT. It is not natively optimized for complex cross-market arbitrage (e.g., Global markets, Forex, Crypto).
• Operating System: The system is primarily developed and tested on Windows due to the MiniQMT client requirements. Running on Linux or macOS via compatibility layers like Wine is not officially supported and may be unstable.

A strategy file in KHQuant is a standard Python script that implements specific callback functions defined by the framework. To create a new strategy, you should implement the following core functions:

• init(stock_list, context): Executed once at the start of the task for global initialization (defining variables, loading data).
• khHandlebar(context): The core logic function. It is called repeatedly based on the frequency set in the 'Run Driver' area. It is responsible for market analysis and generating trading signals.
• khPreMarket(context) (Optional): Called at a specified time before the market opens. Useful for daily stock selection or factor calculation.
• khPostMarket(context) (Optional): Called at a specified time after the market closes. Useful for performance attribution or saving strategy state.

from typing import Dict, List

def init(stock_list, context):
    """
    策略初始化函数，在任务开始时仅执行一次。
    用于定义全局变量、加载外部数据等。
    """
    pass

def khHandlebar(context: Dict) -> List[Dict]:
    """
    策略核心逻辑，会被框架根据设定的频率反复调用。
    负责行情判断和生成交易信号。
    """
    signals = []
    return signals

def khPreMarket(context: Dict) -> List[Dict]:
    """
    盘前处理函数（可选）。
    在每日开盘前的指定时间点调用。
    """
    signals = []
    return signals

def khPostMarket(context: Dict) -> List[Dict]:
    """
    盘后处理函数（可选）。
    在每日收盘后的指定时间点调用。
    """
    signals = []
    return signals

Core Functionality

• The current version of KHQuant is a strategy backtesting and research platform. Its core function is historical data validation. The official version does not include direct live trading execution functions.

Data and Reliability

• Data Accuracy: While the system performs basic integrity and format checks on data from MiniQMT, these are technical safeguards and do not guarantee data accuracy. Accuracy depends entirely on the broker's MiniQMT and its upstream sources.
• No Investment Advice: KHQuant and its outputs (backtest reports, performance metrics, example code) are for educational and research purposes only. They do not constitute investment advice or trading recommendations. Past performance does not guarantee future results.

Liability

• Users assume all responsibility for any losses (financial, data, or hardware) resulting from the use of this software, including errors in data, network issues, or user-modified code used for live trading.

The trigger mode defines the frequency and timing of the khHandlebar function calls (the strategy's "heartbeat").

Tick Trigger

Executes whenever market transaction data arrives (in MiniQMT, this is a snapshot every 3 seconds).

• Pros: Maximum flexibility for adding custom filtering logic.
• Cons: High data volume and slower backtesting speeds due to the large amount of data required.

K-Line Trigger

Executes based on fixed time intervals. In the KHQuant backtesting system, supported periods are 1-minute and 5-minute K-lines.

• Pros: Low resource consumption and fast backtesting. Ideal for trend-following and technical analysis strategies.
• Note: In live trading, even if subscribing to 1m/5m K-lines, the data feed typically pushes updates every 3 seconds. In backtesting, execution occurs strictly at the end of the K-line period.

Custom Time Trigger

Allows you to specify exact timestamps for execution. Ideal for scheduled trading (e.g., opening auction or market close strategies) or non-standard intervals (e.g., every 10 minutes).

• Important: Unlike Tick or K-line triggers, a Custom Time trigger does not automatically pass data to the khHandlebar function. You must manually call data acquisition interfaces (e.g., get_market_data) inside your function.
• Smart Adaptation:
  • If all timestamps are on the exact minute (e.g., 09:30:00), the system uses 1-minute K-line data.
  • If timestamps include non-minute seconds (e.g., 09:30:15), the system switches to underlying Tick data for precision.
• Precision Tip: For stability in MiniQMT, all custom timestamps should be set to a multiple of 3 seconds.

The Data Supplement (数据补充) feature is designed to update and maintain the internal historical database used by MiniQMT and the KHQuant backtesting engine.

How it works: Unlike Data Download, this mechanism calls the xtquant.download_history_data function to write data directly into the MiniQMT system's local data directory. During backtesting, strategies then use high-speed functions like get_market_data_ex to read this local data.

Data Storage Details:

• Location: Data is stored in the MiniQMT installation directory under userdata_mini\datadir.
• Format: Data is stored in an optimized internal binary format (.dat files).
• Directory Structure: Inside the datadir, you will find folders for different exchanges (e.g., SH for Shanghai, SZ for Shenzhen). Within these, subfolders represent the data frequency in seconds:
  • 0: Tick data
  • 60: 1-minute data
  • 300: 5-minute data
  • 86400: Daily (1d) data