No pH prediction model is perfect. If you want to waste a few days, you can jump over to HBT and read all about the different models as people argue about which one is right or wrong -- complete with many differing theories behind their stances.
My own experience shows my results to be similar to yours. I am typically 0.2 pH units lower than what BeerSmith predicts. In fact, it is so consistent that I can set my pH target in BeerSmith to 5.6 and reliably attain a mash pH (at end of mash) of 5.4 +/- .05 pH units. Being a practical engineer and seeing an offset which can be predicted, I do just that -- aim for the target that I know will end up with the actual number I want.
In the end, as long as you are close to the pH range for optimal enzyme behavior, you will get adequate extraction and conversion of the starches to sugars.