基于模糊逻辑的CNN股指趋势预测模型研究

Research on the Trend Forecasting of Stock Index Based on CNN with Fuzzy Candlesticks

目前,投资者大多通过证券投资分析软件进行金融产品价格趋势的预测,这类算法整合交易数据、技术指标数据等,并将这些数据以图形化的方式展示给投资者,投资者则通过"看图",结合相关指标和经验进行分析来发现合适的投资机会。经验主要受主观因素影响,不具备明细的界限,缺乏客观性,具有不确定性,使用模糊逻辑的方法可以有效解决以上问题。构建基于卷积神经网络与模糊逻辑的股指趋势预测模型,模糊表达K线图的上影线、下影线以及实体,以此作为CNN的输入,利用卷积神经网络在网格数据处理上的优势,进行有监督的学习,以达到预测的目的。

At present,most investors by securities investment analysis software for financial product price trend forecast,this software integrates the transaction data,technical indicators data,etc.,and to show the data in a graphical way to investors,investors through the picture,combined with related indicators and experience are analyzed to find the right investment opportunity.Experience is mainly affected by subjective factors,lack of detailed boundaries,lack of objectivity and uncertainty.The method of fuzzy logic can effectively solve the above problems.Constructs a stock index trend prediction model based on convolutional neural network and fuzzy logic,and expresses the upper and lower lines and entities of K line in fuzzy terms.With this as the input of CNN,the advantage of convolutional neural network in grid data processing is utilized to carry out supervised learning,so as to achieve the purpose of prediction.