Axe

Axe declines from 51.8% in Herald to 48.9% in Divine, crossing fifty around Legend. On its own that is an ordinary result — a mild version of the low-bracket pattern that Bristleback, Ogre Magi and Sniper show more sharply. The matchup table explains it better than the brackets do.

Look at who Axe beats: Bristleback at 59.5%, Silencer at 58.8%, Crystal Maiden at 58.5%, Luna at 57.1%, Ogre Magi at 56.9%. Now look at what those five heroes have in common on their own pages. Their average bracket skew is −3.8. Across the dataset as a whole, the opponents heroes beat average −0.4. Axe’s victims are falling roughly ten times faster than the typical opponent.

The reverse holds too. The four heroes that measurably beat Axe — Keeper of the Light at 38.5%, Abaddon at 40.4%, Ember Spirit at 41.6%, Centaur Warrunner at 43.5% — average +1.8, comfortably on the rising side.

That gives Axe’s own decline a straightforward explanation, and a more interesting one than “it is a low-bracket hero”. Axe is not getting worse. Its food supply is thinning. The heroes it reliably punishes are themselves played less, and played worse, as brackets rise, so the matchups that produce Axe’s wins simply stop being offered. A hero whose strength is conditional on the opponent inherits the opponent’s trajectory.

It also suggests what to do about it, which most pages on this site cannot. Axe in a high bracket is not a hero that needs to be played differently so much as one that needs the draft to cooperate: against Bristleback or Ogre Magi the numbers are as good at Divine as anywhere, because that matchup does not care what bracket it happens in.

As always, read the margins. These pairings rest on 104 to 232 games with five to eight points of uncertainty each. The pattern across five of them is firmer than any single figure in it.