Comparison of the binary logistic regression model on the stock price index

Abstract

Abstract. The stock price index is the stock price stated in the index number. There are several methods for determining the price index, including the unweighted/simple aggregative and the chain price index methods. This paper proposes a mixed method between the two methods to determine the stock price index. The mixed method calculate price index based on the simple aggregative method and determining up or down for stock price based on the chain price index method. This paper also attempts to model the stock price index by a mixed method into a binary logistic regression model and compares a binary logistic regression model formed by the simple aggregative price index method. The result of research for Microsoft stock  data since 10 years shows that a binary logistic regression model from the stock price by a mixed method is better than a binary logistic regression model by  a simple aggregative method.