应用基因演算法於共同基金绩效指标一致性之研究.docVIP

應用基因演算法於共同基金績效指標一致性之研究 黃于珊 元智大學資訊管理學系 s922640@.tw 蔡佳倫 元智大學資訊管理學系 s922522@.tw 張書勳 元智大學資訊管理學系 s922549@.tw 邱昭彰 元智大學資訊管理學系 imchiu@.tw 摘要 在基金市場當中，有許多績效指標可供投資者作為投資參考，如Sharpe Index、Jensen Index、Beta Coefficient、Information ratio等指標，然而，各投資公司所提供的績效評估結果往往只採上述其中一個指標作為評估標準。但是對於一般投資者而言，當他們看到多種績效指標評估後，常常很難從各基金績效評估中了解到一個更具整體性的績效評估結果，進而選擇較好的幾檔基金作為投資選擇，換言之，因為有太多基金指標的評估參考，使得一般投資者難以從中得到一個較整體的參考依據。 為了讓一般投資者能夠更容易地從績效評估中選擇基金進行投資組合，本研究試圖透過基因演算法及台灣共同基金做實證研究，探討基金各績效指標間是否具有一致性，以幫助投資者作基金的投資選擇。而所謂的一致性，則是代表此檔基金的各項指標評估是否皆認為此檔基金是值得投資或是不值得投資。本論文利用基因演算法將每檔基金的各個指標評估結果做切割，並分別從切割的結果做評等(極好、好、普通、差、極差)，再去尋找指標之間的一致性。例如：當Sharpe Index評估出此檔基金為極好時(值得投資的)，Jensen Index或其他指標是否也會顯示出此基金為極好。經實證結果發現Sharpe Index、Jensen Index和Information ratio彼此間對於基金的表現具有高度一致性的看法，而Beta Coefficient和其他指標間一致性較低，投資者可以容易地根據研究結果來做更準確的投資評估。 關鍵字：基因演算法、績效指標 Application on Consistency of Mutual Fund Index Based on Genetic Algorithm Abstract In the fund market , numerous fund indexes are suitable for general investors as the investment reference, such as Sharpe Index, Jensen Index, Beta Coefficient, Information ratio ,etc., however, every investment company often assesses the result and adopt one of above-mentioned which is an index among them as assessing the standard in performance offered. But as to general investor, after seeing many kinds of indexes performance to assess, perceive that it is very difficult not only to choose several better funds as investment but also to assess from one fund performance that assesses the result in a performance that has globality even more. In other words, because the assessment of having too many fund indexes is consulted, make general investors difficult to receive a more whole reference basis from it. In order to let general investors choose the combinations of funds more easily while assessing from a large amount of funds. This research attempts to do the research through genetic algorithms (the heredity of artificial intelligence) and Taiwan Mutual Fund, assisting investors to make the investment