No, the fold change is calculated by plotting the logarithm of the abundance values as the y-coordinates on a graph, with the x-coordinates being either 0 or 1 depending on whether the Replicate is in the Control group or the other group.
After these values have been plotted on the graph, a linear regression is performed, and slope of that linear regression is the logarithm of the fold change.
The method was described in this post:
https://skyline.ms/announcements/home/support/thread.view?rowId=43007
There is also an Excel spreadsheet attached to that post which does the calculation in Excel using the "LINEST" function.
If you had only one replicate in each group, of if the "Identity Annotation" was specified, and there was only one unique value for the Identity Annotation in the two groups, then the linear regression method would result in the same number as taking the averages and dividing. But, if you have more than one replicate, then the linear regression yields a different number, probably because the linear regression is trying to minimize the square of the distance of the points to the line.
-- Nick