在投資組合問題中,除了要考慮收益外,同時須考慮風險因素,所以投資組合問題是屬於一種多目標的問題。當可供選擇的股票標的數量愈多時,問題的複雜度就更高,其所需要的求解時間也相對的更多。因此,本研究希望藉由有效的啟發式演算法在合理的求解時間下,求得效益較佳的投資組合。近年來有學者分別提出以基因演算法、粒子群演算法來求解投資組合最佳化問題,而這類演算法都是在空間中反覆隨機搜尋,缺乏自我檢查的機制。本文提出的細菌覓食演算法,當解在某個搜尋方向無法再改善時,能夠隨機轉向繼續搜尋以求得最佳解。本文經由實驗證明,細菌覓食演算法確實能在合理的求解時間下,找到相當靠近題庫標準解的效率前緣曲線,而且在各種不同的風險條件下都能提供投資組合解,使投資者有更多的選擇機會。
Portfolio optimization (PO) is a mixed quadratic and integer programming problem, and an effective solution approach is essential for most investors in order to raise expected returns and reduce investment risks. To solve this problem, various heuristic algorithms, such as genetic algorithms and particle swarm optimization, have been proposed in the past. This paper aims to examine the potential of bacterial foraging optimization algorithms (BFO) for solving the portfolio optimization problem. Bacterial foraging optimization algorithm is a new swarm intelligence technique and has successfully applied to some real world problems. Through three operations, chemotaxis, reproduction, and elimination and dispersal, the proposed BFO algorithm can effectively solve a PO problem with cardinality and bounding constraints. The performance of BFO approach was evaluated by performing computational tests on five benchmark data sets, and the computational results were compared to those obtained with existing heuristic algorithms. Experimental results demonstrate that the proposed algorithm is very competitive in portfolio optimization.