Financial websites tend to calculate beta on securities differently for publicly traded companies. Beta is merely an estimable measure of an assets’ risk in relation to the market. Because each website has their own underlying assumptions, they compute beta to have different values. In some cases, the variance in beta is as large as 0.50. To find out what underlying assumptions websites used to compute their betas, we performed a series of regression analysis using the stock return of Compuware against the S&P 500, and Ken French’s Fama-French Model.
We are assuming both the S&P 500 and the stock return of Compuware Corp. values are calculated on the same days as Ken French’s Data from the Fama-French Model. However, this is not true and further research implies subtracting a day, or selecting which day to start the week or month has significant impact on the value of beta. For the purposes of simplicity, we are assuming this factor should be further researched before making any further assumptions. Also, Ken French’s data did not match up with Yahoo finance’s data one for one, so we deleted days or weeks that seemed to be doubles or did not match in order to perform the regression analysis.
Largest Effects on Beta:
Frequency - Basing stock returns on a daily, weekly, or monthly frequency tends to have a large effect on the value of beta. Our calculations displayed beta ranging from 0.8310 to 2.0848, a 1.2538 difference in value. The main cause of this large disparity was due to the daily frequency. Daily frequencies seemed to have significantly lower betas, which can be explained by the sheer volume of data points approaching the market efficient value. For instance, if we exclude the daily frequency, the variance of the betas is reduced to 0.6462.
Time Span - Changing the periods of time on a calculation of beta seemed to have considerable effects on the actual value. This could be due...