论文部分内容阅读
引入符号时间序列分析方法从大尺度的角度分析收益变化的特征,提出了确定收益变化的主要模式并预测收益水平的方法。首先将收益序列转化为符号序列,由符号序列中不同的字代表不同的收益变化模式,根据符号序列直方图,可以确定收益变化的主要模式。然后,根据各收益变化模式的概率分布,在前几个时点收益水平确定的情况下,可以推知下一个或几个时点处于不同收益水平的概率,从而实现对收益水平的预测。对上证综指、深证成指以及上证工业股指数、上证商业股指数、上证地产股指数、上证公用事业股指数共六个股票指数的收益序列进行了实证分析,确定了各指数收益的主要变化模式,并基于主要变化模式进行了收益水平的预测,从而说明了该方法的有效性和可行性。
The method of symbolic time series analysis is introduced to analyze the characteristics of earnings changes from the perspective of large-scale, and the methods of determining the main modes of earnings changes and predicting the income level are proposed. First, the income sequence is transformed into a symbol sequence. Different words in the symbol sequence represent different patterns of revenue changes. According to the histogram of the symbol sequence, the main mode of the return can be determined. Then, according to the probability distribution of the pattern of the change in returns, the probability of the next or several time points at different income levels can be deduced under the condition that the return levels of the previous few times are determined, so as to realize the prediction of the income levels. The empirical analysis of the income series of SSE Composite Index, SSE Composite Index and SSE Industrial Index, SSE Commercial Index, SSE Real Estate Index and SSE PSI of six stock indices confirms the main income of each index The pattern of change, and the prediction of the level of return based on the main patterns of change, thus illustrating the effectiveness and feasibility of the method.