Skip to main content

RL-HMM Trading System: Reinforcement Learning with Hidden Markov Models for Financial Trading

Project description

RL-HMM 交易系统库

PyPI version License: MIT

一个模块化的强化学习结合隐马尔可夫模型的交易系统库,支持多时间框架分析。

安装

从 PyPI 安装(推荐)

pip install rl-hmm

安装完整版本(含强化学习依赖)

pip install rl-hmm[gym]

从源码安装

git clone https://github.com/yourusername/rl-hmm.git
cd rl-hmm
pip install -e .

功能特性

  • 多时间框架支持:支持 5分钟、15分钟、1小时等多种时间框架
  • 特征工程:内置多种技术指标(RSI、MACD、ATR、布林带等)
  • HMM状态建模:隐马尔可夫模型捕捉市场状态
  • 强化学习:支持 PPO 算法的交易策略学习
  • 模块化设计:灵活配置特征、奖励机制、动作空间等

项目结构

rl_hmm/
├── __init__.py
├── config/              # 配置模块
│   ├── __init__.py
│   └── default_config.py
├── features/            # 特征工程模块
│   ├── __init__.py
│   └── base.py
├── hmm/                 # HMM模块
│   ├── __init__.py
│   ├── hmm_handler.py
│   └── multi_timeframe.py
├── environment/         # Gym环境模块
│   ├── __init__.py
│   └── trading_env.py
├── agents/              # RL Agent模块
│   ├── __init__.py
│   └── ppo_agent.py
├── data/                # 数据加载模块
│   ├── __init__.py
│   └── loader.py
├── evaluation/          # 评估模块
│   ├── __init__.py
│   └── metrics.py
└── system/              # 交易系统主类
    ├── __init__.py
    └── trading_system.py

快速开始

基础使用

from rl_hmm import TradingSystem, Config

config = Config()
system = TradingSystem(config)
system.run()

自定义配置

from rl_hmm import TradingSystem, Config

config = Config()
config.ticker = "AAPL"
config.total_timesteps = 100000
config.n_components = 4
config.win = 100
config.loss = -80

system = TradingSystem(config)
system.run()

模块化使用

from rl_hmm import Config, DataLoader, FeatureEngineer, HMMHandler
import numpy as np

config = Config()
loader = DataLoader(config)
df_15m, df_1h = loader.load_data()

hmm = HMMHandler(config.n_components)
features = df_15m['return'].values.reshape(-1, 1)
hmm.fit(features)
prediction = hmm.predict(features[-1:])

核心模块说明

Config

配置类,包含所有系统参数:

  • ticker: 交易代码
  • interval_15m: 15分钟K线间隔
  • interval_1h: 1小时K线间隔
  • train_window: 训练窗口大小
  • total_timesteps: RL训练总步数
  • n_components: HMM组件数量
  • n_env: 并行环境数量
  • win: 获胜奖励
  • loss: 失败惩罚

FeatureEngineer

特征工程类,提供技术指标提取功能:

  • add_all_features(): 添加所有特征
  • get_feature_names(): 获取特征名称列表

HMMHandler

单个HMM处理器:

  • fit(): 训练HMM模型
  • predict(): 预测状态
  • predict_proba(): 预测状态概率

MultiTimeframeHMM

多时间框架HMM处理器:

  • update(): 更新多时间框架HMM模型
  • get_regime_filter(): 获取市场状态过滤

TradingEnv

Gym交易环境,用于RL训练:

  • reset(): 重置环境
  • step(): 执行一步交易
  • _get_observation(): 获取观察状态

PPOAgent

PPO强化学习Agent:

  • train(): 训练Agent
  • predict(): 预测动作
  • save(): 保存模型
  • load(): 加载模型

DataLoader

数据加载器:

  • load_data(): 下载并加载数据

Evaluator

评估器:

  • evaluate(): 评估模型性能
  • plot_results(): 绘制结果

TradingSystem

交易系统主类,协调所有模块:

  • run(): 运行完整的交易系统

特点

  1. 模块化设计:各功能模块独立,便于扩展和维护
  2. 面向对象:清晰的类结构和职责划分
  3. 易于调用:提供简单易用的API
  4. 多时间框架:支持多时间框架分析
  5. 灵活配置:可自定义各种参数

依赖项

  • numpy
  • pandas
  • yfinance
  • hmmlearn
  • gym
  • stable-baselines3
  • pandas-ta
  • matplotlib

许可证

MIT License

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

rl_hmm-0.2.1.tar.gz (17.1 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

rl_hmm-0.2.1-py3-none-any.whl (18.4 kB view details)

Uploaded Python 3

File details

Details for the file rl_hmm-0.2.1.tar.gz.

File metadata

  • Download URL: rl_hmm-0.2.1.tar.gz
  • Upload date:
  • Size: 17.1 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.1.0 CPython/3.13.12

File hashes

Hashes for rl_hmm-0.2.1.tar.gz
Algorithm Hash digest
SHA256 0cb21428a8bf89ed155037afdcd72017762846861f5cde63cda6d2d5c24b7d10
MD5 8e48a60b3beb5acfaeb1fbba94327fae
BLAKE2b-256 2a367ed2c2c51d12b4df2f70c4c0840ae4582f4f5b5059ae04934d18ca9cf908

See more details on using hashes here.

File details

Details for the file rl_hmm-0.2.1-py3-none-any.whl.

File metadata

  • Download URL: rl_hmm-0.2.1-py3-none-any.whl
  • Upload date:
  • Size: 18.4 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.1.0 CPython/3.13.12

File hashes

Hashes for rl_hmm-0.2.1-py3-none-any.whl
Algorithm Hash digest
SHA256 e43900cc53016408bf06d03b83c274ae6faf0e5d5b04e6e9646f0791b8187c18
MD5 aa73a0403da98be6e16a0c2b66d3db85
BLAKE2b-256 14cd4a3ada7ada17186975ddd4d0320fb067cd6cacebcf6cedb16414eb4a0019

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page