Because this is top-200 only, they're getting their wins off lower ranks, especially axis in teamgames where the gap between games played is absolutely huge between axis top-200 and allies top-200.
Ah I see, thanks. But then this statistics doesn't say very much, e.g. since slightly more people usually search games as axis, the top 200 of axis will be matched more often against allied players of lower skill while allied top two hundred will have an opponent of equal skill waiting in line more often, making stomps more often for top 200 axis players than vice versa.
The statistic should be filtered to count only games against other top 200 players. Of course we wouldn't have this problem if mirror matches were allowed (screw historical acurateness, we already have USF fighting against the OKW in the east, Soviets in France and so on).