qbot

Qbot is an AI-oriented automated quantitative investment research platform aimed at practically applying AI technologies such as machine learning, reinforcement learning, and deep learning to quantitative research. The project is fully locally deployed, free and open-source, covering a complete closed loop from data acquisition, strategy development, factor mining, backtesting, simulation trading, to live automated trading. Architecturally, it adopts a modular layered design: data layer, strategy layer, and trading engine are abstractly decoupled, facilitating integration with multiple data sources and trading interfaces, as well as user-defined strategies and factors. The backtesting part is based on backtrader and easyquant; the live trading part connects to vnpy, pytrader, and pyfunds; visualization and online operations leverage quantstats; and AI strategies integrate qlib and deep learning models. The strategy library is quite rich. Classic strategies cover Bollinger Band mean reversion, dual moving averages, KDJ+MACD, ARBR, Aroon, RSI divergence, StochRSI, SSA sparrow optimization, small market cap, low-valuation stock selection, RSRS timing, pairs trading, grid trading, turtle strategy, dynamic balancing, index enhancement, Alpha hedging, multi-factor stock selection, etc., covering stocks, funds, futures, and cryptocurrencies. On the intelligent strategy side, it includes GBDT (XGBoost, LightGBM, CatBoost), Boost (DoubleEnsemble, TabNet), RNN (LSTM, GRU, ALSTM, ADARNN, ADD, KRNN, Sandwich), CNN (MLP, ImVoxelNet), Transformer series (Transformer, TCTS, TRA, TCN, IGMTF, HIST, Localformer), reinforcement learning (TFT, GATs, SFM), and LLM (ChatGPT, FinGPT), with the Model Zoo housing about 300+ models and 40+ paper methods. In terms of factors, it supports alpha-101 and alpha-191, and implements automatic factor generation based on deap. Built-in indicators include EMA, MACD, KDJ, RSRS, RSI, StochRSI, BIAS, BOLL, OBV, SAR, VOL, PSY, ARBR, CR, BBI, EMV, TRIX, DMA, DMI, CCI, ROC, ENE, SKDJ, LWR, P/E ratio, P/B ratio, etc. Some intelligent parameters and predictive indicators are paid features. Live trading interfaces cover futures (CTP, CTPMini, Femas, Aikronko, Yida), options (CTPOpt, Jinzheng maOpt, QWIN secondary development), stocks (Zhongtai XTP/XTPXAlgo, Huaxin Singularity, Huarui ATP, Kuangrui OES, Hithink, East Money, Huatai, Guotai Junan, Zhonghui Yida, Hengsheng UFT, Juejin, Dingdian Feichuang), as well as broker clients like Haitong, Huatai, Guojin, and Hithink. Message push supports email, Feishu, pop-up windows, WeChat, and other methods, used for trade buy/sell notifications, daily returns, stock recommendations, etc. The project provides a GUI frontend/client (some features support web), with the backend handling data processing and trading scheduling, implementing an event-driven trading process. Runtime environment requires Python 3.8 or 3.9; after installing dev/requirements.txt via pip, run main.py to start. The project is continuously updated; it is recommended to Star rather than fork. Official documentation is at https://ufund-me.github.io/Qbot, with derivative projects like DeepWiki and qbot-mini (iQuant).