行為股票投資組合優化

Behavioral Stocks Portfolio Optimization

楊千億(Michael Young)

指導教授 ： 張國華

中原大學/電機資訊學院/工業與系統工程研究所/博士(2019年)

https://doi.org/10.6840/cycu201900353

摘要

自從行為投資組合理論引入以來，行為金融領域一直在快速的發展。金融世界現在比以往任何時候都充滿了更多的"人性投資者"，而不是專注於所謂的理性投資方式。行為投資組合理論建議人性投資者根據個人看法來估計回報情境。SP/A理論表達了這些投資者根據他們的害怕及期望的等級來衡量風險情境。心理會計學解釋了一般人在精神框架中對金字塔內不同層級的投資，其中每一個層級對應特定的目標和風險等級，這一直是行為投資組合理論中的基礎。
人性投資者還推動了行為投資組合管理，利用投資者的非理性行為來建立優質的投資組合。行為股票是受投資者集體行為影響很大的股票，具有因果關係，在識別相似原因之後有相當高的機率會在特定的時間間格重複發生。
本論文在一系列的研究中適當的介紹了這些行為股票，並透過相應的混合整數行為投資組合優化模型，利用其所擁有獨特資訊來選擇更好的投資組合。BPOM1是一項經過修改以安全第一為主的投資選擇計畫，最大限度地提高行為股票的持有期間回報。BPOM2也是一種經過修改以安全第一為主的投資組合選擇計畫，通過賣空來利用的一種行為股票。BPOM3是BPOM1和BPOM2的混合計畫，它遵循投資組合重新平衡策略，同時考慮實際交易條件和成本。BPOM4是一種改進的安全第一投資組合選擇計畫，考慮了隔日的情境機率以及行為股票模式的機率。BPOM5是一種通用的安全第一投資組合選擇計畫，通過回歸分析將生成的回報情境納入其中，以此來估算和利用行為股票的日效益回報。
通過過往的資料測試，每種模型得到的投資組合提供了統計證據，證明其優於傳統基準，如共同基金、交易所交易基金和市場。此外，比較BPOM投資組合，觀察到當利用行為股票的持有期間回報時，BPOM3提供了優越的投資組合。同樣，當利用行為股票的日效應回報時，BPOM5會產生優越的投資組合。此外，隨著BPOM投資組合的表現超過基準時，就可以得出結論，行為股票可以透過持有期間收益和日效應回報，利用隔日的情境機率以及行為股票模式的機率來幫助估計收益，
和透過修改後的BPOM1到BPOM4混合整數程序來選擇模型。
綜上所述，本文提出了一種利用行為股票來改善投資組合選擇基本框架的高階方法，以產生優秀的投資組合。

關鍵字

行為股票 ； 組合選擇 ； 行為投資組合理論 ； 混和整數規劃 ； 情境權重

並列摘要

The field of behavioral finance has been growing rapidly since the introduction of behavioral portfolio theory. Instead of focusing on the so called rational way of investing way back modern portfolio theory days, the finance world is now more than ever filled with more "human investors". Behavioral portfolio theory suggests human investors estimate return scenarios based on individual perceptions. Security, potential, and aspiration theory conveys that these same investors weigh risk scenarios based on their fear and hope levels. Mental accounting explains the prevalent human nature of mental framing different investment into a pyramid where each layer corresponds to a specific goal and risk level. This has been the norm portfolio selection framework in behavioral portfolio theory. Human investors also drive behavioral portfolio management to promote the exploitation of the irrational behavior of investors to build superior portfolios. Opportunely, it is just the right time for the introduction of behavioral stocks. Behavioral stocks are stocks which are significantly affected by a collective behavior of investors. Behavioral stocks possess cause-and-effect patterns that repeat at a specific time interval after the identification of similar causes at significantly higher probabilities. Appropriately this dissertation introduces behavioral stocks and presents different ways of exploiting the unique information they possess through corresponding mixed-integer behavioral portfolio optimization models (BPOM). BPOM1 is a modified safety-first portfolio selection program that maximizes the holding period returns of behavioral stocks. BPOM2 is also a modified safety-first portfolio selection program that exploits behavioral stocks through short-selling. BPOM3 is a hybrid program of BPOM1 and BPOM2 which follows a portfolio re-balancing strategy while considering real trading conditions and costs. BPOM4 is an improved safety-first portfolio selection program with an embedded two-dimensional weights assignment system on the day effect stock return scenarios and the likelihood of behavioral stock patterns. BPOM5 is a generic safety-first portfolio selection program that incorporates generated return scenarios through regression analysis to estimate and exploit the day effect returns of B-stocks. Through back-tests, the resulting portfolios from each model provided statistical evidence of dominance over traditional benchmarks like mutual fund, exchange traded fund, and Market. Moreover, comparing the BPOM portfolios, it was observed that when exploiting the holding period returns of behavioral stocks, BPOM3 provides the superior portfolio. Similarly, when exploiting the effect day returns of behavioral stocks, BPOM5 generates the superior portfolio. In addition, with the performance of BPOM portfolios over the benchmarks, it can be concluded that B-stocks can help in the estimation of returns through the holding period returns and day effect return, assignment of weights through the two-dimensional weights, and selection models through the modified mixed-integer programs in BPOM1 to BPOM4. Overall this dissertation presents a noble method for improving the basic framework of portfolio selection through the exploitation of behavioral stocks in order to generate superior portfolios.

並列關鍵字

behavioral stocks ； portfolio selection ； behavioral portfolio theory ； mixed-integer programming ； scenario weights

參考文獻

1. Alexander, G. J., Baptista, A. M., & Yan, S. (2017). Portfolio selection with mental accounts and estimation risk. Journal of Empirical Finance, 41, 161-186.

2. Amin, A., Shoukat, S., & Khan, Z. (2009). Gambler’s Fallacy and Behavioral Finance in the Financial Markets (A Case Study of Lahore Stock Exchange). Abasyn University Journal of Social Sciences, 3(2), 67-73.

3. Amirshahi, M., & Siahtiri, V. (2010, June). Evaluating Behavioral Portfolio Theory on investors' purchase decisions at Tehran Stock Exchange. In Financial Theory and Engineering (ICFTE), 2010 International Conference on (pp. 205-209). IEEE.

4. Baptista, A. M. (2012). Portfolio Selection With Mental Accounts and Background Risk., Journal of Banking & Finance, 36(4), 968-980.

5. Barber, B. M. & Odean, T. (2000). Trading is Hazardous to Your Wealth: The Common Stock Investment Performance of Individual Investors., The Journal of Finance, 55(2), 773-806.