论文部分内容阅读
本文对我国股票市场技术交易规则预测能力进行了实证检验,发现移动平均规则所产生的买入区间收益率更大而波动率却更小,卖出区间的收益率为负而波动率却更大。运用自举(Bootstrap)方法检验发现,四种常用的收益率线性模型均不能解释买卖出区间收益率与波动率所表现出的非对称现象,尤其无法解释卖出区间收益率为负的现象。为此,本文通过人工神经网络方法,将条件异方差结构引入到现有的收益率非线性模型,发现该模型能更好地解释买卖出区间收益率与波动率模式,表明收益率动态过程中存在非线性特征。
This paper empirically tests the ability of forecasting the trading rules of the stock market in China. It finds that the moving average rule yields a higher yield and a lower volatility in the buy-in interval. The yield of the sell range is negative while the volatility is greater . Using the bootstrap test, we find that the four commonly used linear regression models can not explain the asymmetry between returns and volatilities in the buy-sell interval. In particular, it is impossible to explain the negative yield of the sell interval. Therefore, this paper introduces conditional heteroscedasticity into the existing non-linear model of returns through artificial neural network method and finds that the model can better explain the range-to-return and volatility models, indicating that the dynamic process of the yield There are non-linear features.