基于LSTM的多指标股票预测
Multi-index Stock Forecast Based on LSTM
齐太威 1于文年2
作者信息
- 1. 武汉邮电科学研究院 武汉 430074;南京烽火天地通信科技有限公司 南京 210019
- 2. 南京烽火天地通信科技有限公司 南京 210019
- 折叠
摘要
该研究通过处理平安银行股票数据生成十个比较有代表性的技术分析指标,将技术指标的值经过预处理后分别作为基于机器学习的多元线性回归、BP神经网络和LSTM神经网络三种模型的输入,通过模型训练来预测股票的涨跌,然后比较三种模型在预测准确率以及回测中年化收益率的表现,证实LSTM神经网络模型对于非线性的股票走势预测效果最好.然后设计改进了一种基于LSTM模式分类的交易择时策略,获得了更高的年化收益率,并且这种策略可行性更高,最后说明利用LSTM模型进行量化交易是可行的.
Abstract
In this Research,ten representative technical indexes are generated by processing Ping An Bank stock data,and the values of technical indexes are preprocessed as the input of multiple linear regression based on machine learning,BP neural net-work and LSTM neural network respectively,training through the models to predict the rise and fall of the stock,and then compare the performance of the three models in the prediction accuracy and the calculation of annualized rate of return,it is confirmed that the LSTM neural network model has the best effect on the prediction of the nonlinear stock trend.Then a trading timing strategy based on LSTM pattern classification is designed to obtain a higher annual rate of return,and it's a more viable strategy,Finally,it is feasible to use the LSTM model to carry out quantitative trading.
关键词
多元线性回归/BP神经网络/长短期记忆网络/行为金融学/量化投资Key words
multiple linear regression/BP neural network/long short-term memory neural networks/behavioral finance/quantitative investment引用本文复制引用
出版年
2024
国家科技期刊平台
NSTL
万方数据