vn.py (VeighNa)

A trading application is assembled here rather than configured: a script constructs an EventEngine and a MainEngine, adds gateways such as CtpGateway, adds apps such as CtaStrategyApp or CtaBacktesterApp, and runs. The core package ships alpha, chart, event, rpc and trader subpackages, while broker gateways, datafeed adapters and the strategy and backtesting apps live in roughly 55 separate vnpy_* repositories under the same account, each installed individually — pip install vnpy on its own is not yet a runnable trading setup.

Version 4.0 added vnpy.alpha for multi-factor ML strategy research. The module carries dataset/, model/, strategy/ and lab.py; the shipped models are Lasso, LightGBM and MLP, and the shipped feature sets are Alpha 158 and Alpha 101 — the README attributes Alpha 158 to Microsoft’s Qlib project. Supported Python is 3.10+ (64-bit, 3.13 recommended) on Windows 11 and Server 2022 or later, or Ubuntu 22.04 LTS and later.

Two things shape who this fits. The README and official documentation are primarily in Chinese, with no English equivalent in the repository root, and gateway and datafeed coverage is heavily China-centric — CTP futures, SimNow simulation accounts, RQData, TuShare, Wind and iFinD. The README also opens by promoting VeighNa Fusion, a commercial product activated through partner futures brokers; the framework indexed here is the MIT-licensed open core (copyright 2015-present, Xiaoyou Chen), which advertises no returns and sells no signals.