The purpose of this study is to build and test a statistical model for the dynamic estimation of portfolio Betas. Of particular interest is the quality of Beta estimates obtainable from relatively small samples of daily return data. Also of particular interest is an assessment of the relationship between the quality of these estimates and the degree of portfolio diversification. For obvious reasons it is desirable for a mutual fund manager to have the best possible estimates of the ongoing (and possibly changing) Betas of competitive funds. These estimates together with estimates of the degree of diversification will allow a portfolio manager to develop investment strategies relative to the expected performance of his own portfolio and his competitors in the market cycle ahead. These estimates will also allow inferences to be made with respect to the current market outlook of each individual competitor. For example, a gradually increasing fund Beta would indicate a bullish outlook on the part of a particular competitor.