I love stats, and in this post I have done the math and have compared the pick rates (popularity) of heroes in Unranked games to their pick rate in Ranked games. I thought this would be an interesting idea as people seem to be far more interested in winning when playing Ranked, and it would thus reflect statistically in the heroes they play. For this article I use both Raw and Actual percentages. A 2% -> 3% change is +1 raw percent, but is +50 actual percent as the hero is now in 50% more games than before. My data includes all games played in the Month of July.
Mirana is played 62% more in Ranked, while Doom is played an incredible 75% more. So Mirana is in 19% of Unranked games, and 31% of Ranked games. This is a 12% raw increase, but an actual increase of 62% and makes her the most played hero in Ranked, about as popular as Pudge in Unranked. The hidden trend I see most prevalent is that a lot of these heroes have been favored in the pro scene over the last few months. In fact, the only heroes that stand out to me as strange is Witch Doctor, Silencer, Slark, and Spectre. None of these are all that popular in the pro scene, although Slark did have a good run in the pros for Jan/Feb 2014. The other three have never been a Top Pick in the pros, which is why I find them very weird to be played more in ranked. The only thing I find notable is that they all have over 50% win rate; Silencer with an impressive 54.9% win rate.
It’s also interesting to me that heroes like Batrider and Lycan aren’t picked more, I guess people don’t feel like these heroes improve their chances of winning as much as a Mirana does.
Some massive losses here, especially notable from Riki who maintains a very strong ranked win rate of 53%. The trend here seems to be an across the board abandonment of ‘pub stomp’ heroes, primarily roaming gankers that are unpopular in the pro-scene. Win rate doesn’t seem to be a factor here, as even Zeus, Sven and Ursa all have solid win rates above 52% (along with Riki). The issue seems to be that the heroes aren’t popular in the pro scene, and therefore are bad picks? That’s the only thing I can fathom from this data. While it makes some sense that popular heroes like Pudge and Sniper would fall, even very unpopular heroes like Night Stalker and Tusk fell. The pro scene popularity seems to be the dominate force here.
Conclusion and Data Limitations
My hypothesis leading into this was that people would pick a combination of heroes with high pub win rates and high pro pick rate. However this doesn’t seem to be the case. Win rate seems to be an almost non factor in picking. People don’t even seem to pick more ‘dominate mid heroes’ which is often considered a good way to boost mmr. For unpopular heroes, it seems that pro pick rate is the most dominate force here. Increases seems to be mostly dominated by heroes that have been popular in the pro scene recently, but they are still very picky with the heroes. Razor, Batrider, and Lycan are all hugely popular at TI4, but this has no direct surge for Ranked play.
I find Pub Trends very interesting. There are about 30-40 million games of Dota every month, and large scale changes like these tell an interesting story. Some possible external reasons for changes could be the fact that Ranked only offers 3 modes, and newer players can’t play Ranked. These both can impact the data, perhaps lower tier players prefer ‘pub stompers’, resulting in skewed data since they can’t queue for Ranked. For reference, Unranked is still the primary way we play Dota, with about 75% of our games being Unranked (3:1 ratio).
The amount of Ranked games also appears to be decreasing, and I’m not sure if people are unhappy with Ranked or if it’s caused by an influx of newer players, but I’m looking into it. If you’re interested in this sort of Data, this is just my first article of a couple I intend to do on Ranked Matchmaking over the coming weeks, so keep an eye out and thanks for reading.
For updates on articles and a steady influx of Advanced Dota 2 tips, follow @ZotaTips on Twitter.
Previous Stat Article – A complete rundown of the changes in hero pick rates for July, in both Pros and Pubs.
Previous Feature – What I think Ice Frog should do after TI4, and I’m not talking about simple hero buffs/nerfs.