以MOEA/D為基礎之群組交易策略組合最佳化技術

MOEA/D-based Group Trading Strategy Portfolio Optimization Techniques

葉重佑(Chong-You Yee)

指導教授 ： 鄭建富

共同指導教授 ： 陳俊豪

淡江大學/工學院/資訊工程學系資訊網路與多媒體碩士班/碩士(2020年)

https://doi.org/10.6846/TKU.2020.00433

摘要

金融市場中，股票投資一直都是個熱門的項目，什麼時候該進行買入或賣出才能避開風險又獲得最大的收益一直都是一個很困難的問題。交易策略是常用來解決這個問題的方法且現有文獻已有許多方法被提出用來產生交易策略或交易策略組合，以追求最大的報酬。其中，群組交易策略組合最佳化方法更提供了投資者更多元的選擇機制，令使用者可以彈性更換組合內的策略。然而，一個群組交易策略組合無法滿足所有使用者的需求。故為了符合許多不同的面相，根據兩目標函數，本論文首先提出SPEA為基礎的群組交易組合最佳化方法來解決這個問題。第一個目標函數用來評估策略組合的風險與報酬，第二個目標函數用來評估群組內的交易策略是否相似。為求得更有效之柏拉圖解集合，我們進一步提出基於MOEA/D的群組交易組合最佳化方法。最後，透過真實資料集，實驗顯示所提的方法的是有效的且優於現有方法。

關鍵字

交易策略 ； 群組交易策略組合 ； 多目標最佳化演算法 ； SPEA ； MOEAD

並列摘要

In the financial market, stock investment is always a hot topic, and when to buy and sell is always a difficult problem for investor to avoid risk and to maximize the return. Trading strategies are a common method to be used to handle this problem, and there are many approaches have been proposed for obtaining trading strategies or trading strategy portfolio to maximize the profit. In addition, the group trading strategy portfolio (GTSP) optimization approach provides investors a friendly mechanism for investors to replace any trading strategy that they do not satisfy with in a given trading strategy portfolio. However, a GTSP cannot meet the needs of all users. In order to meet many different aspects, in this thesis, based on the two objective functions, we first propose a SPEA-based GTSP optimization approach. The first objective function is used to evaluate the risk and return. The second objective function is used to evaluate whether the trading strategies in the group and weights of groups are similar. Then, to reach a better Pareto front, we further propose the MOEA/D-based GTSP optimization approach. At last, experiments on a real dataset were conducted to show the proposed approaches are effective and better than the existing previous approach

並列關鍵字

Trading strategy ； group trading strategy portfolio ； multi-objective genetic algorithm ； SPEA ； MOEA/D

參考文獻

參考文獻

[1] C. H. Chen, C. Y. Lu, T. P. Hong and J. H. Su,"Using grouping genetic algorithm to mine diverse group stock portfolio," 2016 IEEE Congress on Evolutionary Computation (CEC),2016

[2] C. H. Chen and C. Y. Hsieh,"Actionable Stock Portfolio Mining by Using Genetic Algorithms,"JOURNAL OF INFORMATION SCIENCE AND ENGINEERING 32,pp. 1657-1678,2016

[3] C. H. Chen,C. C. Chenb and Y. Nojima,"An efficient and effective approach for mining a group stock portfolio using mapreduce," Intelligent Data Analysis, vol. 21, no. S1, pp. S217-S232, 2017

[4] C. H. Chen,J. Coupe and T. P. Hong ,"Optimizing Diverse Group Stock Portfolio without Setting a Number of Groups,"2017 IEEE 31st International Conference on Advanced Information Networking and Applications (AINA),2017