基于多尺度特征和支持向量机的股市趋势预测

应用支持向量机方法对股票市场趋势性变动进行预测是金融市场行为研究领域里一个重要的研究课题。为了提高股市趋势预测的准确率,现有文献中基本将研究重点集中在改善支持向量机参数上,而没有对输入数据的特征进行深入研究。股票市场时序数据是不同时间尺度因素非线性作用的结果,因此具有本质的多尺度特性。据此构建了股票市场多尺度时序特征趋势预测方法,该方法首先基于小波多分辨分析对股市时序数据进行多尺度分解,然后提取了股票市场数据的记忆性和趋势性特征,最后应用支持向量机对股票市场趋势进行预测,预测结果表明该方法提高了股市趋势预测的准确率。

Forecasting Stock Market Trend Based on Multi-scale Character and SVM

In the field of financial market behavior research,forecasting change of stock market trend with SVM is an important problem in.To improve the forecast accuracy of stock market trend,most researches pay more attention to the improvement of the parameter of SVM and do not study the character of input data deeply.Stock market time series come from the reciprocity of multi-scale nonlinear factors.So it has essential multi-scale character.This paper proposes a stock market trend forecasting method which deals with multi-scale time series characters.The method makes multi-scale analysis of stock market time series and then pick-up memory and trend character of the multi-scale time series.Finally,it uses support vector machine(SVM) to forecast stock market trend.As shown in the experiment,the method can improve forecast accuracy of stock market trend.