繆震宇 (2005) 提出了一套以SPRT (Sequential Probability Ratio Test) 為基礎所建構而成的共同基金評比系統，該系統以共同基金過去超額報酬為基礎並且對基金進行績效評等。然而SPRT基金評等系統中，參數之設定對於評等結果的產生可能具有相的影響性，因此若能尋找出該評等系統較適的參數設定或參數特性，必能提高該評比系統的效能，使該評比系統能更有效的提供基金資訊給予一般投資大眾，做為篩選基金的參考依據。 本研究樣本資料採自台灣經濟新報資料庫，擷取西元2001年3月到2006年12月間國內外基金月報酬率資料進行分析，並利用一因子變異數分析（one-way ANOVA）以及二因子變異數分析(two-way ANOVA)對SPRT基金評比系統產出之超額報酬進行統計檢定，試圖尋找影響SPRT基金評比系統之主要因素以及較適當的參數設定。本研究之實證結果主要如下所示：（一）總平均超額報酬相當顯著的受到星等改變的影響，並且最高的總平均超額報酬明顯集中於五星等基金，最低的總平均超額報酬則明顯集中於一星等基金。（二）當參數改變時能部分影響SPRT基金評比模型當中最重要的一星等超額報酬與五星等超額報酬。(三)評估期間(n)的設定對於SPRT基金評比系統可能具有較大的影響力。（四）當評估期間(n)為短期(6個月)時評比系統效能最高，因此在設定SPRT基金評比模型的參數時，評估期間(n)應以短期為優先考慮。 Miao(2005) proposed a rating system for mutual fund performance based on SPRT (Sequential Probability Ratio Test). This system rates mutual funds according to past excess return. However, the setting of input parameters quietly affects the results of such a system, and an appropriate setting makes the system more efficiently as a reference for sieving mutual funds.
In this study, return rates of offshore and domestic funds according to database of TEJ (Taiwan Economic Journal) from March 2001 to December 2006 are analyzed. Statistical tests for excess returns estimated by SPRT fund rating system are performed by using one-way and two-way ANOVA in order to figure out the factors affecting the rating system most and to propose best fit parameter settings.
The conclusions are summarized as follows : (1) the average excess return is obviously influenced by star rating. The five-star funds make most of the highest excess returns and the one-star funds make most of the lowest. (2) The shift of input parameters partially influences the most important excess return of one-star and five-star funds. (3) The setting of estimation period (n) has more influences on SPRT fund rating system. (4) The rating system shows best performance when the estimation period is six months. Therefore short-term estimation period is the first priority while setting parameters for SPRT fund rating system.
