應用多元迴歸產生情境預測行為股票最佳投資組合

Behavior stock portfolio optimization with complementary multi-index regression generated scenarios

鍾皓玶(Hao-Ping Chung)

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/碩士(2018年)

https://doi.org/10.6840/cycu201800726

摘要

現代投資有許多的方法可賺取額外收入例如購買黃金、房地產、基金、股票。而現在普遍大眾選擇以股票進行投資，購買股票不僅可以賺取公司分紅，透過投資買賣賺取額外收入，傳統經濟理論定義人們的投資行為都是理性的，但是逐漸被經濟學者定義為不理性行為。因此，在現代投資行為都是屬於不理性的，例如：過度期望股票的成長，又或者過度的樂觀但投資是有賺有賠。這些受到不理性行為之影響股票皆屬於行為股票(Behavior stock, B-stock)。
本研究將以不理性行為股票進行研究，其不理性行為依循過度反應之操作型定義(Operational Definition, OD)。操作型定義將說明行為股票發生日起因至結果影響的天數，顯示股票受到不理性行為後產生正報酬的機率。
本研究主要利用安全第一模型（Safety-first）求得最佳投資組合，以情境方式計算機率，針對過度反應行為股票起因所發生的天數，將發生天數與大盤使用多元迴歸方式產生情境預測明日報酬率。將情境運用安全第一模型（Safety-first）使用CPLEX軟體計算出行為股票最佳投資組合，最後將投資組合累積報酬率與大盤累積報酬率進行比較，並獲得較高超額報酬。

關鍵字

行為股票 ； 多元迴歸預測 ； 投資組合最佳化

並列摘要

For the past few years, many people was buying gold, real estate, funds, stocks, to earn extra income. Nowadays, the most popular choice is invest in stocks to earn some profit. Buying stocks can not only earn dividends, but also earn extra income. Investment behavior of individuals was long considered rational but from overlying evidences this is not the case. For example people expect too much of stock growth such that they are optimistic that they will invest on the stock will always be profitable but in any investment there is always a risk of losing. These types of behavior may possibly affect the stock price movement. Those stocks that are affected by irrational behavior are called Behavioral stock, B-stock.
This study uses these B-stocks from the related literature which are identified through the operational definition, OD of irrational behavior established by other researches. The OD defines the cause (trigger of behavior) and the resulting effect (say positive return) of each B-stock. Accordingly, following the literature, we know that after few days (time-to-effect) of spotting the cause the resulting effect (positive cumulative abnormal return) will occur with a corresponding probability.
The purpose of this study is to improve upon the work on B-stocks by confirming the accuracy and effectiveness of the multi-index predictions to predict future returns of identified B-stocks. Using similar data from literature, and considering the respective some days of each identified B-stock, we estimate the future return of these B-stocks and the market through multi-index regression models. We generate return scenarios for the return of B-stocks for the next period, and then used these scenarios to find the optimal portfolio considering the Safety-first portfolio selection model using the CPLEX software. The resulting portfolio is compared to the market which shows that the portfolio with B-stocks and the corresponding estimated return scenarios can outperform the market. Therefore, we can use multi-index regression to estimate future returns of behavior stocks to earn extra income.

並列關鍵字

behavior stock ； multi-index regression ； portfolio optimization

參考文獻

[1] Amirshahi, M. and V. Siahtiri (2010). Evaluating Behavioral Portfolio Theory on investors' purchase decisions at Tehran Stock Exchange. Financial Theory and Engineering (ICFTE), 2010 International Conference on, IEEE.

[2] Atta Mills, E. F. E., D. Yan, B. Yu and X. Wei (2016). "Research on regularized mean–variance portfolio selection strategy with modified Roy safety-first principle." SpringerPlus 5(1): 919.

[3] Baptista, A. M. (2012). "Portfolio selection with mental accounts and background risk." Journal of Banking & Finance 36(4): 968-980.

[4] Chang, K.-H., M. Young, M. I. Hildawa, I. Joshua R. Santos and C.-H. Pan (2015). Portfolio Selection Problem Considering Behavioral Stocks.Working paper, Chung Yuan Christian University.

[5] Chang, K.-H., M. Young and W. Kang Lin (2016). Portfolio Selection Problem Considering Behavioral Stocks under Holding Periods. Working paper, Chung Yuan Christian University.