=$W$1+$V$1*A2+$U$1*B2+$T$1*C2

where W1, V1, U1, T1 are the B0, B1, B2, B3 values ... and A2, B2, C2 are the X1, X2, X3 values respectively for the first observation. And then I fill down so each observation has its matched y-hat value.

That works efficiently for me when there are a few predictor variables. But I am working with a dataset now with about 100 predictor variables, and it is a real pain to type in an extended formula. What is an easier way for me to create a column of y-hats?