使用自適應量子啟發式禁忌搜尋演算法解決具權重之投資組合最佳化問題並使用趨勢值及考量投資者心情波動之新穎指標

A Novel Weighted Portfolio Optimization Model Based on Trend Ratio and Emotion Index Using ANGQTS

徐逸睿(Yi-Rui Hsu)

指導教授 ： 周耀新

國立暨南國際大學/科技學院/資訊工程學系/碩士(2021年)

https://doi.org/10.6837/ncnu202100360

摘要

不論是為了對抗通貨膨脹，或是為了退休生活準備，好的理財規劃是都是必要的。由於股票資訊公開透明且入手門檻低，許多人藉由投資股票來理財，俗話說，雞蛋不要放在同一個籃子，投資者通常會選擇多檔股票組成投資組合來分散投資風險。具權重之投資組合最佳化問題就是從眾多股票中選出多檔股票，並且找出資金分配的最佳比例。本研究選擇美國股票市場作為投資標的，並提出基於趨勢值、心情指數、自適應量子啟發式禁忌搜尋演算法、滑動視窗，且具權重之投資組合最佳化模型。趨勢值是在訓練期評估投資組合表現的新穎指標，相較於知名的指標──夏普值，兩者的核心概念皆是每單位風險下的報酬，但是夏普值的風險定義會導致穩定上漲的投資組合被視為高風險，而穩定上漲的投資組合對於趨勢值來說是低風險的。心情指數是在測試期評估投資組合表現的指標，其概念為承擔每單位心情波動下所獲得的實際收益。要在有限時間內從複雜的解空間中找出最佳解是非常困難的，元啟發式演算法能夠在有限的時間內找出近似最佳解，因此常被應用於解決各種複雜的最佳化問題，自適應量子啟發式演算法便是其中之一，該演算法利用量子反閘、自適應、以歷史已知最佳解導引等機制改善了演算法的廣度及深度搜尋能力。透過滑動視窗選擇適當的訓練期及測試期，能夠避免過度擬合、擬合不足的問題。實驗結果指出，本研究提出的模型在測試期表現優於道瓊工業指數。

關鍵字

人工智慧 ； 計算智能 ； 演化計算 ； 元啟發式演算法 ； 量子啟發式禁忌搜尋演算法 ； 投資組合最佳化 ； 選股問題 ； 股票資金分配問題 ； 趨勢值 ； 心情波動指數 ； 滑動視窗

並列摘要

A financial plan is crucial due to inflation, retirement, insurance, etc., and many people choose stock trading as one part of their overall investment portfolio. When investing in the stock market, investors usually select various stocks to form a portfolio with the goal of forming the best portfolio among numerous stocks and to properly allocate funds. This study targets the U.S. stock market as the largest stock market worldwide and proposes a weighted portfolio optimization model based on trend ratio, emotion index, Global-Best Guided Quantum-Inspired Tabu Search with Self-Adaptive Strategy and Quantum-NOT Gate (ANGQTS), and sliding window. The proposed model focuses on the fund allocation coding scheme, by considering fund allocation is more comprehensive than equally-weighted fund allocation. The trend ratio is a novel indicator to quantify portfolio performance during the training period. Compared to the most famous indicator, the Sharpe ratio, the return per unit of risk is the core concept of them both. However, different from the Sharpe ratio, the trend ratio evaluates a stable uptrend portfolio as low risk, which meets investors’ expectation. The emotion index is also a novel indicator and instead quantifies portfolio performance in the testing period. Different from the trend ratio, the emotion index concept is the profit per unit of fluctuation based on investors’ emotion. Evolutionary algorithms are typically used to solve various optimization problems since they can provide a near-optimal solution in a limited time. ANGQTS is an evolutionary algorithm and utilizes self-adaptive, Quantum-NOT gate, and global-best guided mechanisms to improve its search abilities, including exploration and exploitation. The sliding window can help investors select suitable periods for both training and testing and also avoid both overfitting and underfitting problems. The proposed weighted portfolio optimization model outperforms the Dow Jones Industrial Average index.

並列關鍵字

Artificial Intelligence ； Computation Intelligence ； Evolutionary Computation ； Metaheuristic ； ANGQTS ； Portfolio Optimization ； Stock Selection ； Fund Allocation ； Trend Ratio ； Emotion Index ； Sliding Window

參考文獻

[1] W. F. Sharpe, “The Sharpe Ratio,” Journal of Portfolio Management, vol. 21, no. 1, pp. 49–58, 1994.

[2] W. F. Sharpe, Investors and Markets: Portfolio Choices, Asset Prices, and Investment Advice. Princeton, NJ, USA: Princeton university press, 2011.

[3] S.-Y. Kuo and Y.-H. Chou, “Entanglement-Enhanced Quantum-Inspired Tabu Search Algorithm for Function Optimization,” IEEE Access, vol. 5, pp. 13236–13252, 2017.

[4] H.-P. Chiang, Y.-H. Chou, C.-H. Chiu, S.-Y. Kuo, and Y.-M. Huang, “A Quantum-Inspired Tabu Search Algorithm for Solving Combinatorial Optimization Problems,” Soft Computing, vol. 18, no. 9, pp. 1771–1781, 2014.

[5] H. Markowitz, “Portfolio Selection,” Journal of Finance, vol. 7, no. 1, pp. 77–91, 1952.