Caution: Technical GPS stuff inside.

My initial testing on Alpha runs with the GT-31 produced some pretty strange speed graphs for the Alpha's, especially with trackpoints. (Which was disappointing as with the GT-11 the trackpoints seem to produce the more realistic results)
I had no evidence for an explanation until now but we thought it would be to do with the sirf3 chips internal filtering.
Testing on the weekend with one GT-31 with a new firmware that allows 'smoothing' to be turned OFF in the chip shows our hunch was correct.
With the smoothing turned off the results from the GT-31 were very consistent with the trackpoint results from the GT-11 also worn at the time and don't show the strange artifacts that the GT-31 with the smoothing ON shows. See this screenshot of two GT-31's side by side. One with smoothing turned off (Top):

That application is GPS-Results. Below is the same run show in RealSpeed which has the advantage of being able to import both files and superimpose them in the graph.

It means that using the GT-31 for Alphas with the current firmware is not good. If you MUST use the GT-31, use the doppler speeds for Alpha.

Best to use the GT-11 for Alpha runs at the moment and use the Trackpoint speeds from the GT-11.

Hopefully, the new version of the firmware will be released soon for the GT-31 and we can turn the chip 'smoothing' off. It should be fine for Alphas then. According to Tom, it also improves normal straight line speed accuracy to switch of the smoothing function.