6.80 Heroes: by the Numbers
Starting with the most popular heroes of 6.80, in terms of their overall percentage of appearances in drafts (as picks or bans), we see the following breakdown.
It is unlikely that anyone will be surprised to see Invoker taking the top spot in terms of popularity. Anybody who has been following the competitive scene in the past 2 months will be fully aware of the status this hero has achieved. Widely understood to be the strongest hero in Dota at the moment, and being piloted exceptionally by Resolut1on in particular during the beginning of Empire's rise to prominence this year, Invoker is a no-brainer. Below is a graph showing how the popularity of Invoker is distributed across the various positions. Positions 1-20 suggest the 20 positions of a draft: 4 bans, 4 picks, 4 bans, 4 picks, 2 bans and then 2 picks.
Also unsurprising is that the overwhelming majority of Invoker's participation in drafts has been either in the first ban phase or as the first pick of the draft. This cleanly corroborates the idea that teams see this hero as the best hero around. The point is further made by referring to our original table and noticing that Invoker is the only hero with a ban percentage greater than 50%. Teams generally don't like to face this hero and when it does make it through the first ban phase, it is almost always first picked. Presumably, these cases are explained either by the fact that an opposing team has a specific plan to deal with Invoker or by the fact that they are even more worried about other heroes given a particular weakness they might have or a particular strength their current opponents might have. Notice also that, where banned in the first phase, Invoker is banned more in positions 2 and 4 than in positions 1 and 3, suggesting that teams which have first pick will generally force their opponents to ban the hero rather than risk allowing it to be first picked.
Next on the list is Nyx Assassin. Neither Invoker nor Nyx were discussed at length in either of my previous articles owing to the fact that those articles focused on non-solo roles and these heroes are primarily picked as solos. That said, when discussing supports, I did point out that Nyx's 33.3% pick-or-ban percentage in 6.79 had quickly transformed into 82.4% in the early stages of 6.80. As discussed in the previous article, the increase in popularity can be explained by the fact that this hero has started seeing a lot of play as an offlaner, making it pickable for three roles, the fact that 6.80 gave it a minor buff (matching the cast range of impale to its travel range) and the fact that teams had been allowed some time to get used to Impale now being a skillshot. As with Invoker, it is unsurprising that Nyx has maintained its place at the top one month down the line - with what is actually a slightly increased pick-or-ban percentage of 83.8%.
Nyx's draft position follows a similar pattern to Invoker's, with a notable emphasis on the first ban and pick phases. That said, while both heroes, where picked, are predominantly picked at the very first chance, Nyx sees only a slightly lower number of picks in the 2nd, 3rd or 4th pick position than it does in the 1st position. Nyx is also banned a bit less and (thus?) picked a bit more than Invoker is. This provides a nice contrast to Invoker. While Nyx is a highly prioritized hero, it doesn't seem to be viewed in as absolute terms as Invoker. It normally gets banned or picked early on, but it's not necessarily a case of 'ban this hero or you better have a plan for dealing with it'.
Following Nyx is the hero that many Dota fans may now be starting to tire of seeing, Visage. Ever since Visage started being a big deal, it's pretty much continued being the biggest deal of a support hero there is. However, those who are bored of Visage might take comfort in the fact that the hero has gone down in popularity in 6.80, moving from the 92.5% figure it had when I did my early analysis of the patch to the (still pretty convincing) 76.9% it sits on at the moment. This is a hero that requires very little analysis to explain its strengths, though one interesting point that EternaLEnVy made in my interview with him last week feels especially pertinent here. EE suggested that sometimes support heroes can farm in more dangerous positions as the cost of their being picked off is a lot lower than the cost of a 'true core hero' being picked off. In Visage's case, this is an especially strong justification for seeing people like Aui_2000 farming up a storm on the hero in the midgame because, for this hero, not only is the cost of dying reasonably low, but the benefit of farming is extremely high.
In keeping with the pattern established by Invoker and Nyx, Visage is drafted mostly in the first ban or first pick phase. Regarding the distinction made above between the former two heroes, Visage fits more of a Nyx profile than that of an Invoker. Sure, it's picked more as a first-pick than anywhere else, but the difference is not that huge. Like Nyx, Visage is clearly a high priority to teams during a draft, but it's unlikely to be given the kind of special status Invoker has.
If 6.80 was a movie, the underdog-cum-hero of the story would definitely be Ancient Apparition. Back in February, I discussed this hero as the top up-and-coming support. At the time, its 5% pick-or-ban percentage from 6.79 had already transformed into a whopping 35.7%. Now, 5 weeks later, that figure is 62.8%. I think it's appropriate to compare this hero to Visage in understanding why it has overtaken every other support in terms of popularity. Just like Visage, AA is a dominant presence during the lanestage providing insanely high damage output even at level 1. In fact, Chilling Touch is arguably a much more reliable damage source at level 1 than Soul Assumption is - and, Chilling Touch is a useful tool to zone out enemy heroes while a Visage who has to zone out an enemy has to level Gravechill at level 1, thus reducing the chance of making a kill before level 2. Moreover, while I pointed to Visage's ability to scale well with farm above, AA is in the same boat - since the recent addition of Aghanim's Scepter for the hero. And again, it could be argued that AA is even more suited to the role of a 'often-farming-alone' support than Visage is. While Visage can use familiars to help the team while the hero itself is farming, AA's main contribution to the midgame is Ice Blast, which actually benefits from being far away. AA can also cast Ice Vortex from very far away, and can apply Chilling Touch to allies in advance - whereas a Visage wanting to slow or nuke enemies, will have to be close to the action itself.
While sticking to the established trend that very popular heroes normally get picked or banned early in a draft, Ancient Apparition is unique in the top 5 in that it's the only hero there which is being picked more than it's being banned. While the difference between the two is very small, the fact that the pick % is even near the ban % is significant, given that all the other heroes in the top 5 are banned substantially more than they are picked. One question which will inevitably be asked soon is whether there is room for this hero as a solo. Those who watched Free To Play recently will have been reminded that AA's introduction to Dota 2 saw it being used even more as a mid solo than as a support - and this was before AA could scale well with gold by buying a Scepter. Might we be in for more Dendi AA's in the future?
Rounding off our top 5 most popular heroes is Doombringer. Like AA, this hero was discussed in one of my posts about early trends in 6.80, where it was showing early signs of increased popularity, sitting at 55.9%. This figure has since moved up to 61.5%, which marks the last of the heroes currently above 60% in terms of popularity (though these figures can change fairly easily, and 6-10 are not too far behind). As previously discussed, this hero's popularity can be partially attributed to its versatility, being pickable in any position from 1-4 (farm priority positions). But other than being versatile, the hero is also very gold-efficient, creating more gold out of the same number of creeps (somehow, eating them does this?) and provides an insane presence in the mid and lategames - not only does Dooming an enemy lock that enemy out of a fight or enable your team to kill a hero that otherwise would be near impossible to kill, but Doombringers also carry with them some kind of 'Presence of the Trucklord', walking around with an air of unkillability and usually providing several auras while doing so.
Doombringer's Draft Position Graph provides some very interesting data. Firstly, unlike all the heroes above, this hero is not being drafted mostly in the first pick or ban phase. Secondly, the hero's least common position in a draft is the position which was for the above four heroes their most common position - first pick phase, first pick. Furthermore, where Doombringer is banned in the first ban phase, it is mostly banned in positions 1 and 3, suggesting that the team with first pick is more likely to ban it. This sets up Doombringer as a hero teams might want to avoid their opponents picking as a response to their own first pick. Thinking along similar lines, it might be that first picking Doombringer is not very popular because opposing teams then are less likely to pick those exact heroes that Doombringer picks might be good responses to.
But does this mean the hero is only being picked as a counterpick? Surely if it was being picked for its versatility, wouldn't it be getting picked mostly at the beginning of a draft? I think there is no simple answer here. Most definitely, the hero is not exclusively being picked as a counter. And most definitely, it gets picked at least sometimes just for its 'Scorched Earth + Auras = Presence of the Trucklord' effect. However, it looks as if, regardless of other virtues of the hero, a Doom pickup becomes substantially stronger if you know there will be a specific target that you'd love to Doom. So much so, that teams will avoid first picking it, to increase the chances of enemy teams choosing a hero that is weak against it.
Out of respect for my readers, I'm not going to go into detail about all these heroes here. Nature's Prophet, Clockwerk, Mirana and Storm will not surprise anyone by being on this list. These are all heroes which have already excelled a lot in the competitive scene since 6.79 or, in some cases, even earlier than that.
Batrider, however, is a more interesting case, and might warrant its own article at a later stage - since this is a hero that for years would not go away, no matter how much it got nerfed, reaching as high as 99.7% pick-or-ban percentage for 6.78. And then 6.79 arrived with these changes:
- Base damage reduced from 48-52 to 38-42
- Vision reduced from 1400 to 1200
- Flaming Lasso cast range reduced from 175 to 100
And, just like that, the hero's popularity went down to 35.5%. While this is not an especially low percentage for any given hero, for Batrider, after being near 100%, it's clear that the hero was quite consciously ignored by most teams for a while. Then, in 6.80, Icefrog gave Bat a small nudge in the forward direction:
- Attack animation improved from 0.5 to 0.3.
And now it seems like the Bat is back, very quickly becoming more and more popular. Currently at 55.5%, it won't surprise me if this hero reaches 70% or higher in the next two weeks.
But which heroes are most successful?
It will perhaps surprise you to read that only two of the top 10 most popular heroes in 6,80 make it onto the list of top 10 most successful heroes in 6.80. In collecting the data on success rates, I ignored all heroes with less than 100 games played in the current version - except for Shadow Fiend, which is currently very topical, and might very well be past 100 by the time I post this article. (data is accurate for the end of Monday the 24th of March, 2014)
Other than the fact that this top 10 diverges so sharply from the top 10 in terms of popularity, there are two very interesting things to note about this list. Firstly, Lycan, Shadow Shaman and Centaur Warrunner, which have unquestionably been favoured more by the Chinese scene than anyone else, are all right at the top. By no means are these heroes only being picked in China, but it's the Chinese who are drafting around them almost every game. Perhaps they are onto something?
Secondly, my first instinct when looking at this list was: What the hell is TA doing there? Templar Assassin is a hero that has been on the fringes in recent times. Indeed, it was speculated that one reason that Empire might have improved so substantially after Scandal's departure saw Resolut1on taking his place was that the former, and not the latter, was 'maining' TA at the time. However, although the hero's win rate has recently gone up, its win rate in 6.79 was already comfortably at 55%. No, the significant change is that it's starting to get picked a bit more again now. At the end of 2012, we were still playing 6.76, a version in which TA's popularity was at 91.5%. 6.77 brought a minor increase to the cost of Refraction (and various other changes to the map). TA's pick or ban rate dropped massively to 31.5%. In 6.78, it dropped to 19.1% and in 6.79 it reached its all-time low at 12.3%. In 6.80, the hero is back up to 22.3% so far, being picked-or-banned more than it was over the last two versions. This may be an early sign that teams are starting to find it more natural to fit TA into their strategies again - where it fits, however, it has always been highly successful.
Some readers will be disappointed at the abrupt ending of this article. There is certainly a lot more to say about the win percentage table, and I would have loved to do more in depth comparative analysis between the two tables. It might be that the most popular heroes struggle to have the highest winrates because they are played more often than others. It might also be that their superior popularity means that they don't actually make it past the ban phase except in cases where opponents are prepared for them, thus reducing the extent to which their level of success can be a measure of their strength as picks. Therefore, it's not clear what to make of the divergence of the two top 10s. Lacking the resources to do the appropriate comparative data-level analysis (and skeptical about whether such a thing is even possible in a meaningful way) I decided to rather just introduce the notion here and leave it at that. That, and this post is really long and I do not want to be responsible for your poor posture.
As usual, any and all feedback is extremely welcome. All data used comes from datdota.com. Oh, and you can follow me on Twitter if you like.