I am using linest to calculate a simple y = mx + b regression with an n of
57 values. I also have it return the r^2 value. Linest gives the same
values as the chart trendline r^2, and my manual calculation of r^2, when I
set constant to "TRUE". However if I set constant to "FALSE", and thus force
the line through zero, linest returns an r^2 value that is higher than the
chart trendline value (0.9909 vs 0.99750).
Excel Help mentions that the degrees of freedom change when setting
constant to false, but my manual calculation of "Degrees of Freedom Adjusted
R-Square" does not give the linest value, it gives 0.099746.
Is there a bug in linest with respect to r2 values of forced lines?
Thanks.

-Filtration Guy