Alpha-Beta Test Run Results

We have estimated the alpha-beta values for 51 mutual funds for 1-year (brown), 3-year (green), and 5-year (blue) ending 9/30/2006. Our estimated values are consistent with the published values at Morningstar, Yahoo, and Google. However, there were some values which were not the same as the published values due to different reference indexes other than S&P 500 were used for the published data. In our calculations, we use VFINX returns as the market returns. We believe this is a much better market index than the theoretical S&P 500 index.
We made three observations about this test run. First, our calculation methodology must be consistent with the methodology used by Morningstar since our estimated values for alpha and beta are consistent with the published values at Morningstar web site. We also checked our values with the published values at Google Finance. We have discovered the same consistency. So we believe that our data source and calculation methods are reliable.
Secondly, the 51- funds used in our test run have a wide range of alpha-beta values. The beta values range from 0.25 to 3.00 while alpha values range from -14 to +41. We believe these ranges cover a wide spectrum of data. So the alpha-beta values are not concentrated closely to its origin (alpha = 0 and beta = 1). There is a great possibility for someone to select overly under-valued or over-valued funds among the funds. We believe there is greater potential if all mutual funds are considered for this project.
Thirdly, the values of alpha and beta are not very stable over different time periods. The “best” funds (low beta with high alpha) are changing places among themselves. This may be the reason why some sector rotation funds can outperform the overall market for an extended period of time. This characteristic may be a good treasure place for swing traders to harvest the change tides of sectors or styles.
Our next project is to study the time series characteristics of alpha and beta value for a few selected funds.