LSTM-stock-price-prediction

Stock price predicetion (classification and regression) using LSTM.


Report Bug · Request Feature

Table of Contents

1. About The Project
2. Getting Started
   • Prerequisites
   • Installation
   • Usage
3. Hyper-parameter Search
4. Classification
5. Regression
6. Acknowledgments

About The Project

Stock price predicetion (classification and regression) using LSTM. Integrated with another homemade light-weight quant framework. Support Sliding windows, hyper-parameter search, backtesting, Reversing Trade and etc.

LSTM股票价格预测，调用了另一个自制框架。支持滑窗, 超参数搜索, 反向对冲, 回测等。

Features:

• Model
  • LSTM
  • Seq2seq
  • Resnet50-1D
• Prediction
  • Signal Classification (Buy, Sell, Hold) 信号分类
  • Regression (avg price in next window) 回归
• Backtesting Metrics 回调指标
  • Sharpe 夏普
  • Maximum Drawdown 最大回撤
  • Alpha (regression/annualized) (回归法/年化)
  • Beta (regression/annualized) (回归法/年化)
  • Interval rate of return 平均区间收益率
  • Annualized rate of return (baseline/stretegy) 年化收益率 (基准/策略)
  • backtesting rate of return 策略回测收益率
• others
  • Reversing Trade Support 反向对冲回调策略
  • Sliding Window 滑窗生成器
  • focal_loss
  • class_weighed_sampling 分类权重采样 (抑制类别不均衡)

Built With

• 
• 

Getting Started

Prerequisites

Clone repo.

git clone https://github.com/dr413677671/LSTM-stock-price-prediction.git

Installation

pip install <repo-directory>/requirements.txt

Usage

Prepare raw data in csv format.

Run relervant jupyter notebooks, and use pandas.dataframe to read raw_data.

.
├── README.md
├── docs 
├── Regression                   # Signal Regression 
├── hypertune                    # Hyper-parameter tuning
├── classification               # Window Classification
└── lib 
    └── quantflow                # Homemade quant framework 

Hyper-parameter Search

Classification

Regression

Contact

     

(back to top)

Acknowledgments

Based on these brilliant repos:

• Seq2seq
• LSTM
• Logo genetrared by Stable-Diffusion

(back to top)