基于基本面数据和注意力机制的股票趋势预测

Stock Trend Prediction Based on Fundamental Data and Attention Mechanism

在金融市场中,资产的价格和回报的预测一直是金融从业者和学者们最具挑战和激动人心的问题之一。机器学习对股票进行中长期趋势的预测时,多采用股票的基本面数据,但由于基本面数据其特征数较大,不同的特征对预测出的股票的回报其影响是不一致的。基于这些因素,提出带有注意力机制的前馈神经网络模型对其预测。实验结果表明,该机制与随机森林(RF)和前馈神经网络(FNN)的效果对比有显著提升。

In the financial market,the prediction of asset prices and returns has always been one of the most challenging and exciting issues for finan-cial practitioners and scholars.When machine learning predicts the medium and long-term trend of stocks,the fundamental data of stocksis mostly used.However,due to the large number of characteristics of fundamental data,different characteristics have inconsistent effectson the predicted stock return.Based on these factors,a feedforward neural network model with attention mechanism is proposed to predictit.Experimental results show that the effect of this mechanism is significantly improved compared with Random Forest(RF)and Feedfor-ward Neural Network(FNN)