Abstract: Due to the artificial disturbance of the stock market, the effects of the stock market prediction algorithm based on sentiment are poor. In the paper, the Markov Chain Stock Market Trend Prediction Algorithm Based on Rational Indicator （RI_MCA） is proposed  for the problem of the buy and sale induced in stock market. First, the main stock market rational characteristics are extracted and quantized; then, by principal component analysis of these characteristics rational integration into rational indicators, the buy and sale is acquired to use of these rational indicators; finally, situation of the stock market to the buy and sale is led into Markov chain to realize the stock market trend forecast. The reliability of buy and sale is affected in the condition of the deviation of rational indicators and stock status. RICD_MCA algorithm （Markov Chain Stock Market Trend Prediction Algorithm Based on Rational Indicator Combining with Characteristics Deviate） is proposed by introducing characteristics deviate into RI_MCA algorithm which adjusts the RI_MCA algorithm results by the degree of the feature departure. RICD_MCA algorithm has higher prediction accuracy showed by the comparison and analysis of experimental results of Shanghai Composite Index.

Key words: rational indicator, Markov chain, characterized departing, principal component analysis

摘要： 由于股票市场存在人为扰动性，使得基于情绪的股市预测算法效果不佳。针对股市的诱多诱空问题，提出一种基于理性指标的马尔可夫链股市态势预测算法（RI_MCA）。提取股市的主要理性特征，并对这些理性特征进行量化；通过主成分分析将这些理性特征融合成理性指标，并利用理性指标获取股市的买卖点；将买卖点所对应的股市状态引入到马尔可夫链中，实现股市态势预测。在理性指标和股市状态相背离情况下会降低买卖点的可靠性，因而通过将特征背离引入到RI_MCA算法中提出了RICD_MCA算法，RICD_MCA算法根据特征背离程度对RI_MCA算法的结果进行调整优化。在上证指数上的实验比较与分析结果表明，RICD_MCA算法具有更高的预测精度。

关键词: 理性指标, 马尔可夫链, 特征背离, 主成分分析