With the International over and 6.79 on the theoretical horizon, it’s a good time to look at how the general playerbase has responded to the balance shifts in the current patch. To do this, I’ve put together four samples of public matchmaking games of roughly 35,000 games each. The sample comes in two pairs. The former was collected from the months of January through March of this year, and covers a large stretch of 6.77 (and no games from 6.77c). The latter is made up of games that took place between June 4th and is split between 6.78 and 6.78c. Each pair is then further divided into games that fall under the Normal bracket and games that fall under the Very High bracket as defined by the Dota 2 client.
So let’s start by looking at the heroes with the greatest increase of usage between the two patches in Very High:
Let’s start with the big one, Treant Protector. As I’ve covered elsewhere, 6.77c was a huge patch for Treant, and public matchmaking has certainly caught onto it. His 164% surge in usage almost doubles the next largest increase. Interestingly, Treant was also the #1 increase in Normal, but at a more modest 54%. Normal trends are often a faint echo of what we see in Very High. While Treant’s nerf in 6.78c has brought him back down to earth competitively (he was a relatively rare pick at TI3 with a 6-7 record overall), it hasn’t appeared to have effected him much in public play.
Spiritbreaker is our second biggest mover, and while his 6.78 buff to Charge of Darkness has only helped him to see a tiny bit of competitive play, it has succeeded in turning him into the preeminent pubstomper of the patch. Due to sample size issues, the win rates I have aren’t super precise, but based off what I do have it’s rather likely that Spiritbreaker is at least in the top 5 heroes in Very High rated matches.
Moving down a bit, we have Huskar and Timbersaw. Both heroes saw a substantial set of changes in 6.78, and both heroes have seen their win rates go up as a result. Incidentally, this sample independently confirms my previous finding that Huskar’s rework has turned him into one of the highest win rate heroes in Very High games.
Besides win rate increases, one of the biggest movers in usage trends is players imitating competitive play. Weaver, Alchemist, Spectre, and Visage are I suspect a result of this, as none of these heroes saw major changes in 6.78. Visage and Alchemist have obviously been blowing up for months in the Eastern scene, and Weaver’s popularity surge was demonstrated during the TI3 prelims. Spectre is a hero that has been held back competitively by aggressive trilanes, but her public win rates over the last couple patches have been consistently high, which I suspect is a result of people finally catching on to the buff Desolate saw back in 6.75. Clockwerk might also be a case of imitation, but I had thought his competitive usage had fallen off a bit recently.
That leaves us with Outworld Devourer, Doom, and Mirana. All three of these heroes saw noteworthy buffs in 6.78 that while not as extreme as some of the others have still boosted their usage. For Outworld Devourer, the 5 extra movespeed he received in 6.78 has had a noticeable effect on his overall win rate and likely contributed to the attention he received during TI3. Mirana received bonus damage on her Arrow and a reduced cooldown on her Ult. Neither change was earth-shattering, but they’ve definitely helped her out and Mirana is the kind of hero that Very High players love to pick. Doom, well, some of his surge might be attributable to people playing around with his new Aghanim’s effect, but I suspect that there’s actually something more long-term going on here. His Very High winrate is up this patch without a corresponding shift in Normal, which could just be sampling error, but it’s impossible to reach a solid conclusion without getting more data.
Meanwhile, on the bottom end of usage rate trends:
The bottom four are all heroes that got hit hard in 6.78 (or 6.77c) and saw big drops in their win rates as a result. Granted, all four had been +55%, so the nerfs weren’t exactly unwarranted.
Phantom Lancer is the biggest loser in terms of usage due to the damage nerf to his illusions. He’s still very much viable in pubs as well as competitive play, but it seems like a lot of the attention he was receiving was driven solely by the desire for easy wins.
Drow Ranger is a similar story, except that she never made it into Captain’s Mode before the nerfs she received in 6.77c. She’s been a bit of a problem child as far as balance goes, as any change that might make her competitive in professional play seems destined to make her too mindlessly successful in public games.
Keeper of the Light’s decline came as a two part story, with Illuminate nerfs in 6.77c and a small Base Strength nerf in 6.78. Like Phantom Lancer, he’s still viable, but he’s not the +95% Pick/Ban hero in competitive games that he was during the first quarter of the year.
Finally we have Troll Warlord. While he did see a nerf to his ultimate in 6.78, the bigger story is the bugfix that restored a cast time to the melee version of Whirling Axes. The net effect was similar to what happened to Ursa in 6.75 when Overpower was given a cast time, and it also explains why Troll’s Dota2 win rate was much higher than many Dota1 players had predicted. Troll was recently added to Captain’s Mode, and he certainly has a chance of becoming at least a niche pickup in competitive games, but his odds of showing up are certainly lower than they would have been in 6.77.
The rest of the list is tricky, and in general they’re smaller movements that don’t constitute major trends. Undying and Centaur Warrunner are both heroes that have seen substantial downwards tuning, so their decline in 6.78 might just be the residue of larger declines in previous patches. It’s also the case that many of the heroes on the rest of the list, such as Brewmaster, Tidehunter, and Enigma, are heroes that are largely dependent on teamfight ultimates. Their decline could be a result of the public meta imitating the recent professional meta that places a greater emphasis on multi-core lineups.
If you’d like to see the complete findings, you can find a spreadsheet containing the data here. To close things off, I present to you without comment the hero usage trends in 6.78 between Normal and Very High.