Image by Wykrhm
Trying to understand TI8 Meta is problematic, to say the least. On one hand the game is still incredibly focused on the laning stage. On the other—most teams attempt to deviate in one way or another, trying to slightly outgreed their opponent on one of their lanes.
The reason most teams are playing 2-1-2 Dota currently is the deny XP change—it went from 70% to 25% in patch 7.07, making suicidal offlanes a lot less viable. The enemy doesn’t need to kill or even zone you out anymore, since they can simply deny every single creep and leave the offlaner without any experience.
From this we got high emphasis on the laning stage—stronger lane, which can get momentum going their way in the first couple of waves, will quickly outlevel their opponents and gain further control over the lane. Getting a bad start is incredibly punishing and if previously lost lanes at least got some levels, currently even experience is highly contested.
This was the general consensus going into TI—pick strong lanes, dominate the first 8 minutes of the game, gain map control, farm bounty runes and win through economy and item timings. However, despite these strict rules and restriction, the meta somehow ended up being impressively diverse with 109 different picks.
The Big Three
Vengeful Spirit, Mirana and Tiny top the most picked chart. All three heroes were picked more than 60 times during the group stage, seemingly coming out of nowhere—there were glimpses of Tiny and Mirana during the Summit 9, but Vengeful Spirit’s popularity is certainly baffling.
She does fit well into the category of comfort heroes and the hero did receive several important buffs in the last several patches, but no one expected her to be the biggest hero of the tournament. 69 games, 50.72% winrate—the hero is far from overpowered, but she is reliable, lanes well and can provide some necessary initiation from a support position without any economic investments.
Mirana is interesting primarily because of her winrate—winning 38 out of 61 games is very impressive. Capable of going to any lane and being played in any position makes her an extremely flexible pick, though she was mostly played as a core. Her main strength lies in her ability to pose a kill threat at all stages of the game, while also having decent scaling from the level 15 talent.
The last of the big three, Tiny, looks like a slightly overvalued hero. He lanes very well and has a lot of burst damage, but his economy heavily relies on being able to snowball—something harder to achieve against stronger teams. He has a 41.67% winrate across 60 games, but it doesn’t mean the hero is weak and we don’t expect him to lose popularity during the main event. He might not be doing well in pubs, but he excels at creating the necessary space for some of the greedier heroes in highly coordinated environments and will probably remain highly valued because of it.
Annoying to deal with
Enchantress tops the most banned section, closely followed by Io. While Io is in fact much weaker after the nerfs he received in the last couple of patches and is mainly banned to prevent certain powerful combinations, Enchantress looks like a very strong hero no one wants to deal with.
66.67% winrate across 21 games is an impressive number, but it is understandable. The hero is very hard to punish during the laning stage, courtesy of her high movement speed, untouchable passive and a healing ability. She does a considerable amount of damage from almost full range, can summon the help of tough-to-deal-with neutrals and scales incredibly well into the late game with her spell-immunity piercing damage auto-attacks. She is also very independent, capable of standing her ground even if her support rotates out, allowing for higher aggression across the map.
Silver bullets
Venomancer, Batrider, Kunkka, Arc Warden and Lycan all have extremely high winrates, without getting much attention. Venomancer is particularly impressive as a hero who won 6 out of 7 of his games. He doesn’t fit all drafts, but he can be oppressive if utilized correctly.
Batrider is also very interesting, being a TI specialty. The hero is almost completely forgotten during the season, but always makes a comeback during the biggest tournament of the year. Prevalence of first phase Vengeful Spirit will probably cut into his popularity this TI, but he is still the most reliable initiator in the game.
Highly overvalued
Rubick, Windranger and Huskar are the most overvalued heroes of the tournament so far with 10+ picks each and abysmal winrates. Rubick is understandably weak—hero does very little in lane and has to be played almost perfectly to have an appropriate impact. He can still turn fights and do amazing flashy plays, but most other supports can provide similar levels of utility with a lot less effort and much higher consistency. It is not a good patch for Rubick.
Huskar is kind of surprising, but also isn’t. Previously, the hero was mostly picked in a duo with either Dazzle or Oracle, but teams started experimenting with Huskar picks without his usual pairings—neither of these two supports fit the meta with a strong emphasis on the laning stage. Experiments proved unsuccessful for teams trying to play Huskar, with 28.57% winrate across 14 games.
Windranger was picked 19 times and won only four games, for a 21.05% winrate. There was always rivalry between Windranger and Mirana, with the rise of popularity of one hero resulting in lower popularity of the other. They kind of want to do similar things and one is always better than the other, and it seems Mirana is substantially better this patch. It is worth noting, though, that while Mirana was mostly played as a core, Windranger had a fair share of support games.
Making sense of it all
Looking at the stats, the diversity of the tournament starts making a little bit of sense—teams are experimenting with off-meta picks, but they really are not that successful. Weak lane supports like Disruptor are game-losing—he lost all four games he was in. Supports with wavelclear are a lot more reliable. Weak laning cores are also a problem.
Going into main event and elimination matches specifically we expect teams to tighten their picks significantly. We probably won’t see any more Monkey Kings, Phantom Assassins or Lunas—they are not good fit for the current meta and can’t contest their lane well enough, getting last hits, sustaining themselves and harassing opponents at the same time. And even when given some space come midgame, their impact is already limited.
There are a couple of interesting exceptions, however. Spectre remains a very popular pick with a decent winrate of 44.44% across 27 games. She often uses a 2-0-2 build for lane sustain and survivability and ends up with a single point of Desolate by level 10, but at least she is not dying and can turn fights with her ultimate.
Terrorblade was picked in 9 games, winning 5 of them. He is a strong laner for 40-52 seconds every 2.5 minutes and can be quite resilient against a low magic damage teams. He is also good at stalling the game and creating space for himself with the help of illusions.
Medusa somehow managed to win three out of five games she was played in, once again stalling and making sure she gets to a point where she can be a dominant force.
These exceptions to the lane-dominant meta leave us with an interesting conclusion: this tournament is about going all-in. You either all-in with strong lane and early game pressure, with heroes like Mirana, Necrophos or Gyro. Or go all-in ultra-late, stalling as much as you can and defending your highground.
Midrange carries don’t work too well—with lower tower bounties and lower kill bounties their timings are a little bit off and faster cores simply outpace them and can play aggressive, while slower carries generally have enough time to get items and levels for holding highground and eventually outscale.
It is going to be interesting to watch how the meta develops on the main stage and whether we will see the last six heroes being picked, but right now it doesn’t look like it—teams have learned their lessons in one of the most intense group stage in TI history and will probably adjust towards less experimentation and more safety.