基于模式识别的股票价格预测研究

Research on Stock Price Prediction Based on Pattern Recognition

基于模式识别的时间序列数据分析方法,因其能够揭示复杂、非线性的股价运行模式,在经济学、统计学和机器学习领域受到了广泛的关注。为了避免人判断的主观性,因此采用一种与双底形态的目标模板进行对比的方法,把一维的股价时间序列变换成二维的图形与目标模板进行对比,对比后相似度高的时间片段能作为未来市场价格上涨的信号。实验结果表明,此方法能够产生正收益,且对股票的未来涨跌具有一定的预测能力。

The method of time series data analysis based on pattern recognition has attracted wide attention in the fields of economics,statistics,and machine learning due to its ability to reveal complex and nonlinear stock price patterns.A heuristic based template called the“Double Bottoms”,which transfers a one-dimensional stock price time series to a two-dimensional template,and identifies time segments with high similarity to the template as signals which can predict future market price increments.Experimental results show that this method can generate positive returns and has a certain predictive ability for future stock price movements.