Bayesian Forecasting of Economic Time Series: A Case Study of MSG stock price

• Main Authors: Cheng-Long Hong, 洪正龍
• Other Authors: Chi Lo
• Format: Others
• Language: zh-TW
• Published: 2013
• Online Access: http://ndltd.ncl.edu.tw/handle/10560132284418984624

碩士 === 中華大學 === 應用統計學系碩士班 === 101 === The occurrence of important real-world events may cause a fundamental change in the movement of an economic time series. For example, Jeremy Lin’s fabulous debut in the NBA, even the Wall Street also felt. The Jeffreys-Savage(JS) Bayesian theory in this paper is used to test whether Linsanity event has a significant effect on the movement of the time series of the MSG stock price. If the Linsanity event is non-negligible, then we model the movement of the MSG stock price as an exchangeable sequence. In this dissertation, we only considered to forecast the ups and downs of the MSG stock price. In the simplest of 0-1 valued data, is a 2-valued Polya process and is a simple random walk with . We then use the JS theory to predict future observations by taking a weighted average of the optimal predictions for each model, with weights given by the posterior probability of the hypotheses. The fully Bayesian predictor are compared with the Polya predictor, predictor, and the predictors based on the “p-value”. From the result of computer simulation, we know that the fully Bayesian predictor is generally superior, especially in the small sample case. Finally, we conclude from the test result that Linsanity event has no significant effect on the movement of MSG stock price