Covariance

I am building stocks Portfolio model . I have to use the Covariance . so i have to use matrix to get the covariance between the 20 stocks model. I have 60 HISTORICAL data input per each stock.

My problem:

i used Covariance under data analysis icon and got the results. it is okay with me and had no problem. i needed to use more straight method as i am using MACRO in the model. So i tried to use the function " {mmult(transpose),(xxx : yyy), xxx : yyy)/60} . supposedly the formula should give me the covariance matrix i need for my whole data, and should produce the same result as 1st method. Actually it is not and the mmult produced different results. for example Covar under 1st method produced 0.2872% vs. 2nd method which is 0.2994% . Why this difference ? where is my mistake if any ?