By quincy0191
So the aim this time around is to find two groups of heroes: the ones that have a big difference in how well they do when used in different roles, and the opposite. First up we’ll look at heroes that seem to be very dependent on role, then in the second half we’ll focus on heroes that do about the same regardless of what role they play.
Win% of Role-Dependent Heroes
(each hero needs at least five games on a given role in order to qualify for the leaderboard)
The first thing that stands out is that all five of these heroes qualify for the initiator and suicide roles. But it’s not just a straight “it’s easier to play initiator than suicide”, as that’s true for only three of the five. For the other two, Cthuluphant and Magmus, they’re generally better suited to the suicide role. Both also have the biggest discrepancy between their initiator and suicide roles, and Cthuluphant’s difference is the largest among all heroes. He’s also much better on the jungle role than as an initiator, and Magmus is just generally better everywhere else. That’s definitely a surprise as he’s typically considered a mid hero.
It’s also surprising that in general these heroes aren’t that good overall. Kraken, Drunken Master, and Magmus have mediocre win rates in their better roles, and abysmal performances in their worst situations. Collectively they’re just 137-146, which isn’t hugely surprising as this sample isn’t selected for win% so a record around .500 should be expected. But it should be on the other side, as the minimum requirement of ten games overall should encourage better heroes (since good heroes should be picked more often).
The suggestion, then, is that the supposed versatility of these heroes outweighs their issues on certain roles. Picking them early (and an unscientific subjective evaluation suggests they are picked early if at all) doesn’t give away a team’s strategies and allows them to adjust to unexpected picks or bans by the opposing team. However, the reputation of viability in multiple roles may in fact be a farce, since it doesn’t seem to hold up under examination. So let’s check out some heroes that do.
Win% of Role-Independent Heroes
These heroes look much more similar to each other than the last batch did. There’s a common theme: they’re all run as carries in some situations, and as suicides or initiators in others. Bubbles is the only exception to that, though he simply eschews being a carry and an initiator or suicide for an initiator and suicide. Unlike Cthuluphant, Magmus, and Drunken Master, they only seem to fit one of two roles instead of three or four.
Additionally, they’re all generally good heroes, with one exception in Wretched Hag. She hasn’t had quite the strong performance of past events, with both her suicide and carry games not going so well. But the other four heroes have a minimum of 50% win on all their roles. Collectively they’re 99-84, substantially better than the overall record of the most role-dependent heroes.
However, these heroes aren’t quite as popular as the ones in the previous table. Bubbles leads the way with 61 games played, sixth overall. Swiftblade is 11th with 48 games, and Hag is 14th with 38. The other two are outside the top twenty, just 20 games for Chipper and 16 for Revenant.
For Swiftblade a relatively low games played count is expected considering how often he’s banned. But if we consider him and Cthuluphant equals in the ban department (though Cthuluphant actually gets banned more), we still have a substantial pick difference between the two groups, 261 compared to 135. Despite the latter being generally much more stable in different roles, they’re picked nearly half as often.
There are a lot of reasons why that is, mostly connected to the fact that four of the five of them can occupy the carry role. Usually a team won’t run multiple heroes that can be played as carries because the farm requirements are too high, so picking one of these heroes early often gives away their strategy because it’s expected – often correctly – that they will be taking the primary farm. However, clearly these heroes can excel in other situations, so perhaps that logic is somewhat lacking.
Regardless, four of the five could probably stand to be picked a little more, and three of those four are usually available within at least the first six picks and often the last four as well. Selecting a Bubbles, Revenant, or Chipper early on lets a team move in a couple different directions while maintaining a strong presence in whichever role the hero takes. To be entirely fair, only Bubbles holds the holy trinity of being generally available, consistently good, and playable for pretty much every team (as Revenant and Chipper can be considered niche heroes), but I think the lesson there is that more teams need to know how to play the other two.
Finally, it’s worthwhile to mention that we saw one support role and only a couple jungles on these ten heroes. There are two good reasons for that: the first is that the common jungle heroes are rarely placed in other roles, and uncommon jungle heroes are by definition not jungled very often. The second is that the objectives of a support or jungler are often very different than that of a carry or even initiator and suicide. Supports in particular fill a specific and distinct job, and junglers are usually more focused on crowd control than damage output, which is the job of laning heroes for the most part. This means that heroes who do one usually do not do the other very well. There are a few instances of heroes getting enough games to show up as a support (Magmus, Torturer), but there have only been ten heroes overall who have played at least five games as a support.
That’s also a good place to throw in the caveat that the five game requirement let a lot of small samples in, like Revenant. The lack of large amounts of data regarding some of these heroes means it’s difficult to get accurate assessments of their overall power; Revenant could be a very good hero, or he could have had a couple lucky games that vaulted him into rarefied air. Unfortunately the only way to compensate for that is to get more data, which we can’t just create out of nothing. Still, there’s enough here to come to some interesting conclusions and give a couple heroes a spotlight.
quincy0191 is a HoN veteran with a focus on the numbers. Fascinated by quantifying and valuing human beings, he shunned the world of finance in favor of sports and competition. When not on Honcast you can find quincy playing games like Civilization and Pokemon, or watching movies, writing, pondering, or catching a game of baseball.
