基于ARIMA-BiLSTM模型的沪深300指数预测

Research on the Forecasting of CSI 300 Index Based on ARIMA-BiLSTM Model

股票价格指数是度量和反映股票市场总体价格水平及其变动趋势而编制的股价统计指标,是反映社会、政治、经济变化状况的“晴雨表”。准确预测股票价格指数波动,有助于防范股票市场风险和保障金融市场稳定发展。由于股票数据的复杂性,本文将时间序列模型与深度学习模型相结合,提出新的组合模型ARIMABiLSTM,对沪深300指数数据集进行性能评估。结果表明,ARIMA-BiLSTM模型可以规避单一模型带来的缺陷,较其他先进方法具有更高的预测精度。文章对构建的模型展开消融实验,研究模型参数设置的合理性,并基于不同数据集进一步验证了组合模型的预测效果。

Stock price index is not only an indicator of the overall price of stock market and its trend,but also a“barometer”of the social,political and economic changes in the country or region where the market is located.Accurate forecasting of stock price fluctuations helps to prevent risks and promote the sound development of financial markets.Due to the complexity of stock data,this paper combines time series models with deep learning models to propose a new combined model ARIMA-BiLSTM to evaluate the performance of the CSI 300 index data set.The ablation experiments are conducted on the proposed model to study its parameters settings,and further verify the prediction accuracy of the proposed model based on different datasets.The results suggest that the new model can circumvent the defects brought by a single model and has higher accuracy than other advanced methods.