honestly, that's a rookie move. Averaging download speeds over a handful of tests is barely scratching the surface. You need to look at consistency over a longer period and under different loads. Plus, proxy performance can fluctuate because of so many variables - server load, IP reputation...