基于积累线性模型的证券单笔交易模型

Equity's Transaction Model Based on Cumulative Link Model

积累线性模型是处理自变量为连续变量、因变量为离散变量时的常用方法。而单笔交易中证券价格的变动为离散变量,本文采用积累线性模型为其建模,得到了在一定的交易方向和交易量下,证券价格变动的条件概率分布。理论与实证结果均显示,积累线性模型更适合用于建立单笔交易模型;其中序次Logisic回归又比序次Prob it回归的拟合效果更好。使用积累线性模型建立的证券单笔交易模型具有良好的应用前景。

Cumulative link model is a useful tool when independent variable is continuous and dependent variable is discrete. Since the equity＇s price change is discrete when transacting, we use the cumulative link model to build transaction model. We get the conditional probability distribution of the change of equity＇s price. Both the theory and empirical result show that cumulative link model is appropriate. Moreover, the ordered logistic regress is proved better than ordered probit regress by the test of goodness of fit. The equity＇s transaction model based on cumulative link model can be applied well in practice.