Windranger

The skew figure on this page says −0.4 points, which makes Windranger look like one of the steadiest heroes in the dataset. Read the bracket column instead, because the summary is hiding the shape.

The win rate runs 49.5, 48.3, 48.3, 48.0, 48.2, 48.3, 49.1. It falls for four brackets, bottoms out in Archon, and then climbs back almost to where it started. Herald and Divine happen to land close together, so a figure built from the two ends reports almost no movement across a curve that moves twice.

This is a limitation of the metric rather than a quirk of the hero — most of these curves are not straight lines, and any single number taken from the endpoints will mislead on some of them. It is worth saying plainly on the one page where it matters most.

What the curve does not do is cross fifty. Windranger never averages a win at any bracket, peaking at 49.5% in Herald, and it is the second most-played hero in that group behind Lion. The trough is shallow — a point and a half from best to worst — so none of this describes a hero in trouble. It describes a hero that is slightly behind everywhere, least so at the extremes.

The matchup table is much more decisive than the brackets: thirteen favourable results against three. Zeus at 61.5%, Viper at 58.3%, Necrophos at 57.5%, Underlord at 57.4%, then a long tail in the mid-fifties against Slardar, Ursa, Templar Assassin, Tusk, Lion and Mars. A hero that beats thirteen opponents measurably while sitting below fifty overall is winning its specific fights and losing the average somewhere else.

The exception is severe enough to flag on its own. Treant Protector takes Windranger to 31.5% over 108 games — the heaviest single result on this page by a wide margin, and well outside its own uncertainty of ±7.3. Ember Spirit at 40.2% over 301 games is the firmer of the two bad matchups, with less margin to argue about.