The tests were with a non-parametric test (Mann-Wilcoxon rank sum test for those who know what that is)
The armies in the late period consisted of 4 cavalry armies (Merovingians etc), 4 Eastern Knight armies (knights with LH and LF such as Hungarians), 4 Heavy Foot armies (Swiss etc), 7 Longbow armies, 5 Medium Foot armies (Catalans etc), 8 Shooty armies (inc Ottomans, Mamluk etc) and 6 Western Knight armies (i.e. Condotta Italian).
None of these army types did significantly better than would be expected by chance, though a couple of types did significantly worse (Western Knights at 1% significance, Cavalry armies at 5%, though with the caveat that we are doing 7 tests, so there is about a 30% chance of at least one by chance alone).
Ottoman taken on its own did do signficantly better at 1% - they were 1st, 4th, 7th and 11th out of 38. So maybe Janissaries are really the problem - or Pete
