Ursa

Ursa climbs 4.9 points across the brackets, from 45.4% in Herald to 50.3% in Divine, crossing fifty only at the very top. That alone would make it an ordinary entry in the “rewards practice” category. What makes this page worth reading is that the matchup table flatly disagrees with the trajectory.

Nine pairings clear the significance test against Ursa and only three in its favour. Sniper takes it to 38.9%, Puck to 39.7%, Lich to 40.4%, then Beastmaster, Shadow Demon, Phoenix, Ember Spirit, Windranger and Templar Assassin fill out the low forties. On the other side there are three: Underlord at 62.8%, Dark Willow at 59.4%, Timbersaw at 58.6%.

So Ursa is a hero that nine opponents measurably beat, and it still improves steadily as the players holding it get better. Those two facts are not in conflict, but the tension between them is the useful part. A losing matchup profile describes vulnerability to the enemy draft. A rising bracket curve describes something the player contributes that the draft does not determine. Ursa has a lot of the second and a real shortage of the first.

That suggests a specific reading of the counters. All nine of them share a quality — range, mobility, or a way to refuse a straight fight — and none of them is a hero that out-brawls Ursa on its own terms. They do not beat it by being stronger, they beat it by not being caught. That is a positioning problem, and positioning problems are exactly the kind that improve with practice, which is consistent with the climb.

Two cautions. The matchups here rest on 101 to 179 games apiece with six to eight points of uncertainty, so the ordering within that list of nine is not meaningful even though its direction is. And the climb, while steady, only just reaches an even win rate — Ursa at Divine is a 50.3% hero, not a dominant one.