Throughout the American TI4 qualifiers, I heard a lot about how many of the North American teams were drafting like Evil Geniuses (EG). After all, the teams were competing for a spot in the largest Dota 2 tournament ever, so wouldn’t it make sense to emulate the most internationally successful American squad? This question eventually led me on a lengthy study of drafting behavior across all the major Dota scenes and some interesting insights on what we might see leading up to the The International.
The Americas qualifier
My first step was to compare the picks of the top five teams in the North American TI4 Qualifiers to the drafts of EG to see if there was any actual statistical proof backing up this theory. The first step was determining EG’s most commonly drafted heroes in both 6.80 and 6.81.
I then compared these picks to the drafts of North American Rejects, Team Liquid, Sneaky Nyx Assassins, Union Gaming and Team eHug. The results were quite interesting as I saw a significant increase in draft similarity with all the teams except for North American Rejects. Overall, Team eHug and Team Liquid drafted most similarly to EG - both teams drafted one of EG’s most commonly picked heroes around 70% of the time.
Essentially, EG’s drafting style could have had a profound effect on the other top teams in the American scene. Considering that EG is one of the few American teams capable of competing on an international level, it only makes sense that other American teams would try to emulate their success by adopting similar strategies. Yet this is only a correlation and does not imply causation, so I decided to dig deeper.
An international trend?
This data could also suggest that the patch was the root cause of this increasingly similar drafting behavior, so I decided to compare EG to other successful teams around the world. EG has 32 wins and 5 losses in the current patch, so I wanted to compare them to other teams who have had at least 30 games played and over a 50% win rate in 6.81. These teams were LGD, Mineski, Empire, ViCi Gaming and Fnatic.
The investigation showed that these teams drafted far less similarly to EG than the American teams. The average 6.81 draft similarity to EG was only 38%, far lower than the 56% of the American teams, and the average growth in draft similarity between 6.80 to 6.81 was under 10%. In fact, most of this growth can be attributed to one team - LGD, who increased their similarity to EG’s picks by roughly 120%. The rest of the teams saw negligible increases, and ViCi Gaming actually decreased their pick similarity by almost half. In fact, if we remove LGD from the calculations, there is actually an average decrease in draft similarity.
Overall, the numbers seem to indicate that there is an above-average correlation between the draft behaviour of the American scene and EG’s picks.
Regional game flow
Next I wanted to look at what else made EG as successful as they’ve been besides their hero picks. To explore this I decided to compare EG’s game time and kill totals to the various TI4 qualifiers.
The things that stood out to me most was that EG’s average game time was much faster than other teams and that their average kill differential is significantly lower as well. This led me to believe that EG’s style of play allows them to gain a farming advantage early in the game not only through kills but through their efficiency.
I decided to look into this further by breaking down EG’s farm efficiency and priority.
Farm efficiency and priority
EG and Arteezy are known for introducing the farming mid into the meta. In the farming priority breakdown, I found that EG generally gives most of the farm to their mid player, Arteezy, who on average receives 28% of the team’s total GPM. Team Liquid, who had the highest increase in draft similarity to EG, was the only other team to prioritize their mid player’s farm like EG, giving 30% of their total GPM to qojqva.
I found that only Arrow Gaming exceeded EG’s average total GPM and XPM. Most teams hovered around 2,000 GPM/XPM, while EG averaged at roughly 2,150 GPM/XPM.
All the other teams gave the highest farm priority to their carry player. CIS.Black^ saw 32% of the team’s total GPM, the highest among all these teams. These results indicate that CIS has the closest to a 4-protect-1 strategy of the teams that qualified for TI4. On the other hand, Virtus Pro, MVP.Phoenix and LGD tend towards triple-core lineups, with only a difference of about 3 to 5% between their lowest-prioritized and highest-prioritized core player.
Overall, there seems to be a wide variety of trends in Dota 2 currenty, and they aren’t particularly tied to a region. We are seeing farming mids with teams like EG and Liquid, triple-core lineups with LGD, VP and MVP, and more traditional dual-core drafts with the rest of the teams. However, no matter the amount of cores, support roles typically share less than 30% of the farm.
Conclusions and takeaways
As teams begin playing more official matches in this patch, we will be able to look at these trends over time to see what works and doesn’t work for certain teams and how they adapt to the metagame prior to The International.
In the end, the teams that began drafting more like EG performed better than usual. LGD and NAR both qualified for TI4. Liquid qualified for the play-in series and Team eHug did far better than most predicted in the American qualifiers. However, my analysis indicated that this was a more localized trend in the Americas, and that teams in other regions, like CIS or MVP, also qualified for TI4 using quite different drafting styles and strategies.
It seems that the recent patch has ushered in a metagame centered around experimentation, an example of this being the Third Place Decider match in WPC-ACE where Newbee and Alliance combined to draft 23 unique heroes out of 30 picks. Teams are finding more and more viable strategies in the new patch, and are experimenting with what suits them best. The next several weeks will be an important time for teams to find their identity and strategy for The International.
NB: Statistics last updated on 30/5/2014, and are courtesy of DatDota.com.