運用啟發式演算法於股價指數套利投資組合模式之建立

Abstract

台灣的金融市場在朝著國際化及自由化的目標下，逐漸開放一些新興金融 商品的交易。其中台灣股價指數期貨的推出，使得整個金融市場的發展更 具完整性。期貨市場中，投機、避險及套利三大功能提供了投資人多元化 的交易方式，相對的，在複雜多變的金融環境中，投資風險的管理也益顯 其重要性。故本研究針對套利交易中的現貨投資組合的建構方式部份，嘗 試提出一個新的投資組合建構模型，利用「啟發式演算法」在搜尋方面的 特性，在可接受的時間範圍內，建立出一個可以有效模擬現貨指數的投資 組合。實証研究結果說明本研究所提出的「啟發建構模型」投資組合可以 有效的模擬指數現貨，其績效要優於傳統的「市值加權模型」投資組合。 另外建構策略選擇方面，建構期以1500筆到2000筆的投資組合在1800筆資 料的檢驗期模擬績效最優，投資組合家數只需10到20家便可有效模擬。 In order to archive internationalization and liberalization, Taiwan gradually permit some transactions of modern financial merchandise. There among, the issuance of Taiwan Stock Index Future Makes the whole financial market more integral. The three major function in future market, speculation, hedge, arbitrage, offer investors the pluralistic transaction styles. In contrast, the risk management of investment become more important under the circumstance of complicate financial market.Therefore, we try to propose a new portfolio construction model. for the spot portfolio formation in arbitrage transaction. This new model apply the heuristic algorithm to establish a portfolio which can efficiently simulate the spot index in acceptable time scope. According to our research ,the heuristic model can efficiently simulate spot index, and the performance is better than the traditional model , Capitalization Weighted Model. Besides, portfolio will have best simulation performance with the records between 1500 and 2000 during the construction period. The portfolio only need 10 to 20 stocks to simulate efficiently.