本文基于隐马尔可夫模型(HMM),选取上证指数近10年的历史数据(开盘价、最高价、最低价和收盘价)进行实证分析,得出HMM模型在股票预测方面具有一定的可行性。同时,通过对传统HMM模型的输入和预测方法进行改进,对股票价格变化作出了更加...本文基于隐马尔可夫模型(HMM),选取上证指数近10年的历史数据(开盘价、最高价、最低价和收盘价)进行实证分析,得出HMM模型在股票预测方面具有一定的可行性。同时,通过对传统HMM模型的输入和预测方法进行改进,对股票价格变化作出了更加准确的预测。主要步骤为:1) 数据处理。对股票价格序列进行检验并做处理,以股价波动率作为HMM模型的输入。2) 根据池化信息准则(AIC)和贝叶斯信息准则(BIC)固定最佳隐状态数目,并通过训练模型确定参数。3) 预测。相较于传统HMM模型根据股票价格序列直接得到预测数据,改进后的HMM模型则通过股价波动率计算后得出的预测得到了进一步提升。Based on the Hidden Markov Model (HMM), this paper selects the historical data of the Shanghai Composite Index in the past 10 years (opening price, high price, low price and closing price) for empirical analysis, and concludes that the HMM model has certain feasibility in stock prediction. At the same time, through the improvement of the input and prediction methods of the traditional HMM model, more accurate predictions are made for stock price changes. The main steps are: 1) Data processing. The stock price series is tested and processed, and the stock price volatility is used as the input to the HMM model. 2) The number of optimal hidden states is fixed according to the Pooling Information Criterion (AIC) and Bayesian Information Criterion (BIC), and the parameters are determined by training the model. 3) Forecasting. Compared with the traditional HMM model, which directly obtains the forecast data based on the stock price series, the improved HMM model further improves the prediction obtained by calculating the stock price volatility.展开更多
文摘本文基于隐马尔可夫模型(HMM),选取上证指数近10年的历史数据(开盘价、最高价、最低价和收盘价)进行实证分析,得出HMM模型在股票预测方面具有一定的可行性。同时,通过对传统HMM模型的输入和预测方法进行改进,对股票价格变化作出了更加准确的预测。主要步骤为:1) 数据处理。对股票价格序列进行检验并做处理,以股价波动率作为HMM模型的输入。2) 根据池化信息准则(AIC)和贝叶斯信息准则(BIC)固定最佳隐状态数目,并通过训练模型确定参数。3) 预测。相较于传统HMM模型根据股票价格序列直接得到预测数据,改进后的HMM模型则通过股价波动率计算后得出的预测得到了进一步提升。Based on the Hidden Markov Model (HMM), this paper selects the historical data of the Shanghai Composite Index in the past 10 years (opening price, high price, low price and closing price) for empirical analysis, and concludes that the HMM model has certain feasibility in stock prediction. At the same time, through the improvement of the input and prediction methods of the traditional HMM model, more accurate predictions are made for stock price changes. The main steps are: 1) Data processing. The stock price series is tested and processed, and the stock price volatility is used as the input to the HMM model. 2) The number of optimal hidden states is fixed according to the Pooling Information Criterion (AIC) and Bayesian Information Criterion (BIC), and the parameters are determined by training the model. 3) Forecasting. Compared with the traditional HMM model, which directly obtains the forecast data based on the stock price series, the improved HMM model further improves the prediction obtained by calculating the stock price volatility.
