以馬可夫鏈分析券商之目標價預測行為

Abstract

本文主要以馬可夫鏈探討券商所發佈的目標價預測是否存在一動態調整行為，並藉此分析券商預測的準確度之變動過程與長期下券商預測準確度的穩態機率分布。在台灣股市中有70%以上的投資人為散戶投資人，然而這類投資人往往因為缺乏足夠的專業投資能力與完備的資訊管道而迷失在茫茫股海之中。因此，具有專業形象與投資素養的投資機構所發佈的券商投資建議或目標價預測便逐漸受到散戶投資人的關注，進而作為投資決策之依據。本文以TEJ券商投資建議資料庫內券商對個股所發佈的目標價預測為研究對象，選取2007年3月至2009年12月為樣本期間，先將資料依照其屬性進行分類，再以本文自行定義之預測係數決定每筆券商預測資料的預測狀態，最後求出券商預測狀態之轉移矩陣與穩態機率分佈。結果發現，本研究根據目前所選取的十家券商對十支個股所做的目標價預測資料認為券商對個股所做的目標價預測不具有預測能力，因為這些券商的目標價預測在絕大多數的情況下是需要極長的等待時間才能實現或是根本無法實現，投資人在參考券商所發佈的目標價預測時應持謹慎保留的態度。

The purpose of this thesis is to investigate the existence of dynamic adjustment behaviors of the brokers and institutions when forecasting the target prices of securities on the Taiwan stock market, and to analyze how such behaviors affect the movement in the process and the steady-state probability distribution in the long run of the brokers and institutions’ forecasting accuracy by Markov chain analysis. The private investors account for over 70% of Taiwan stock market; however, these investors usually lost their direction in the stock market for the lack of professional investment knowledge and sufficient market information. Thus, more and more private investors are paying attention to the stock recommendations or target price forecasting issued by professional institutions to assist their investment decision. In this study, we introduce the data from TEJ database during the period between 2007 and 2009. A coefficient with a form of the theoretical realization period over the actual realization period is created in this work and is employed to determine the forecasting level of every prediction of the brokers or institutions. Finally, we evaluate the transition probability matrices and the steady-state probability distributions of the predictions of the brokers and institutions. The results based on our sample indicate most of the predictions of the brokers and institutions take too much time to realize or even don’t realize in our sample period, so that the private investors should keep a cautious attitude when putting the target price forecasting into investment decisions.