| Summary: | 碩士 === 國立高雄大學 === 統計學研究所 === 98 === In the arbitrage pricing theory, the most important spirit is to establish the relationship between return and risk in order to asset pricing. A major problem in studying asset pricing is that common factors affecting asset returns are unobservable. In relevant literatures, there were many methods to estimate risk factors, such as semiautoregression approach, asymptotic principle component analysis, and so on. In this article, we try to use the macroeconomic factors transformation methods to quantify risk factors.
In order to describe the relationship between risk and return more appropriate, many scholars use different models to reflect the interaction mechanism in market. We propose a model to combine a dynamic programming model and arbitrage pricing theory .We expected it can make optimal decisions in the next step on different investment period. We find that multi-armed bandit model is better than other traditional methods.
|
|---|
