CAPM beta of a stock mismatch

hi,

i was going through the CAPM with missing data example to find alpha and beta of individual stocks. i was getting the data through money.net. i first got daily data for 2 years for 5 stocks , s&p500 and 1 yr us bond rates. i converted the daily prices to log returns and the us bond rate to daily return as well. following the example as on the website i found the excess return like this:

TestData(:) = Data(:,i) - Data(:,14);

TestDesign(:,1) = 1.0;

TestDesign(:,2) = Data(:,13) - Data(:,14);

% Estimate CAPM for each asset separately

[Param, Covar] = ecmmvnrmle(TestData, TestDesign);

most of the program was exact copy of the example as per mathwork.

now i got the alpha and beta estimate for the five stocks.to my surprise i checked the values of the beta i got to be very different from those i found online. for eg. my beta for the company amd is 2.0316 but on yahoo, zacks and reuters it is around 3.05 to 3.68. some of the other values i found is amgn(my beta 0.9855, zacks value 1.158),aapl(my beta 1.3717, nasdaq value 1.02),ccl (my beta 0.907, zacks value 1.09) etc.

to recheck if i'am doing something wrong i even used the following formula covariance(stock,s&p)/variance(s&p) and i got the exact same values as mine. i tried taking longer data of 3 years but not much change in values. could anyone help and inform if there is any kind of error in my calculations? can anyone inform why the discrepancy in the values. though some values are close values like that of amd have a huge difference. any help on the matter would be highly obliged.

thanks in advance

azim