在本文中，相較於過往基本的Black-Litterman model，在Zhou(2009)認為，除了投資人本身觀點以及市場外，應該要將樣本中的資訊納為考量，因此將Black-Litterman model中的估計結果，在貝氏分析的架構下，與樣本估計結合，並且我們透過該模型估計結果應用於切點投資組合上。此外，我們也提出了相對於以往不同對於市場均衡下權重估計的方法，使用歷史平均市值作為權重，能夠更準確地捕捉市場均衡。接著我們透過Baker and Wrugler(2006)中提出情緒對於未來股票報酬具有預測性，在市場樂觀時，後續報酬較差，反之，當市場悲觀，後續報酬會較好。因此，我們使用投資人情緒作為模型參數中Tau值變的判斷方法，在實證結果中可以發現，將情緒納入考量後的模型表現，在使用特定分類方法下建構出的投資組合確實會相對於固定Tau值的參數設定下，有更好的表現。;This paper examines the performance of the traditional Black-Litterman (BL) model and its extensions, with a particular focus on Zhou (2009) modification, under different asset classifications and market conditions. The Black-Litterman model, which integrates investor views with market equilibrium, has constituted a fundamental element in the field of portfolio management. Zhou’s enhancement incorporates a Bayesian approach to address the limitations of the original model, thereby providing a more comprehensive framework for asset allocation. As demonstrated by Baker and Wurgler (2006), investor sentiment provides insight into the market. By incorporating investor sentiment, the model is refined through the use of sentimentadjusted tau values, allowing for a dynamic adjustment of portfolio weights based on market sentiment. Moreover, we put forth the suggestion of incorporating a historical average market capitalisation, which serves to enhance the reflection of market states. This helps to reduce the impact of biases resulting from short-term market fluctuations. Our contributions include a demonstration of the practical benefits of sentiment-adjusted models and an illustration of the effectiveness of the historical average market capitalisation weights approach. The empirical analysis assesses the out-of-sample performance of a range of portfolios, classified according to their momentum, size, book-to-market ratio and industry. This comprehensive evaluation offers new insights into the applicability and benefits of these enhanced models, demonstrating their potential for improving asset allocation strategies in different market environments.
