The Effectiveness of Strategy Following the Net Long/Short of Institutional Investors in Black Litterman Model-Empirical Evidence from Listed Companies in Taiwan

• Main Authors: Ling-Jie Kao, 高令潔
• Other Authors: Day-yang Liu
• Format: Others
• Language: zh-TW
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/j7239e

碩士 === 國立臺灣科技大學 === 財務金融研究所 === 105 === The main purpose of this study is to test the effectiveness of utilizing Black-Litterman Model on practice investment. As distinguished from previous research, we go deep into the analysis on the part of investor views. The manner to establish our investor views is to do linear regression to analyze the relationship of return between the groups which are overbought and oversold. Furthermore, we try to adjust confidence of investor views by p-value of the regression. This study is aimed to provide all investors, particularly mutual funds, with a model usage guide. The sample period runs from January 1, 2004 to April 30, 2017. The empirical results are as follows: 1. The 3-year or longer effectiveness of strategy following the net long/short of institutional investors in Black-Litterman Model is better than Markowitz Mean-Variance Model and Equal Weighted Model, and also beat the Taiwan Stock Exchange Weighted Index and Yuanta Taiwan 50 ETF. 2. As number of assets in the asset pool increase, Black-Litterman Model shows better stability than Equal Weighted Model and Markowitz Mean-Variance Model. Different from the two models, its return(volatility) would not obviously decrease(increase) when the size of asset pool expanded. The result suggests that Black-Litterman Model is more suitable for the asset allocation in larger asset pool. 3. Black-Litterman Model is able to create a portfolio which is more diversified than what Markowitz Model created. Meanwhile, it is also able to actively invest in the merit assets. The features reflect on the transaction costs, Black-Litterman Model pursues higher effectiveness with investor views and lower costs. 4. The empirical result shows that performance of portfolio under p-value adjustment is obviously better than which theτset to 0.025 and adjusted by the number of samples, but still can not beat the portfolio which theτset to 1.