论文部分内容阅读
为提高金融时间序列的预测精度,本文提出了基于MODWT、MCP变量选择方法和RELM_Adaboost的混合预测模型。该模型由三步构成:第一步,收集特征变量,包括MODWT分解得到的特征变量以及常用的技术指标;第二步,利用MCP惩罚方法从上述特征变量中选取重要的作为输入变量;第三步,利用Mnet惩罚正则化ELM,将RELM视作弱预测器,然后用Adaboost算法生成强预测器进行预测。实证结果显示:第一,经过MCP方法的筛选,最终的输入变量中不仅包含常用技术指标,还有小波分解所得的变量。第二,混合预测模型RELM_Adaboost有良好的泛化误差表现。本文提出的模型在量化交易时代具有良好的应用前景。
In order to improve the prediction accuracy of financial time series, this paper proposes a hybrid forecasting model based on MODWT, MCP variant selection and RELM_Adaboost. The model is composed of three steps: the first step, collecting feature variables, including the feature variables obtained by MODWT decomposition and commonly used technical indicators; the second step, using the MCP penalty method to select important variables from the above characteristic variables as input variables; Step, the use of Mnet penalty regularization ELM, RELM as a weak predictor, and then use Adaboost algorithm to generate strong predictors for prediction. The empirical results show that: First, after the screening of the MCP method, the final input variables include not only the commonly used technical indicators, but also the variables obtained by the wavelet decomposition. Second, the mixed prediction model RELM_Adaboost has good generalization error performance. The model proposed in this paper has a good application prospect in the era of quantitative trading.