Meta of all previous The Internationals
So… Hey, it’s been a while. Initially I wanted to make a series of short posts about meta at different TIs starting with TI7, but then the idea evolved, and now it seems easier to make one big post about all TIs, starting with TI1.
In this post I’ll go through every TI with infographics, then through all-time TI stats and some of the things I encountered while going back in time.
This post is massive, be aware of that.
Oh yeah, all of these stats are now available in my stats hub. I’ll link reports for every specific TI in its category below. Another thing to note: for every of these reports I manually edited player and team names, as well as updated team logos, to look more consistent (and better as a whole). If you want to check all the TI reports, you can check out my The Internationals section following this link.
The International 2011
The first TI, the one that started it all.
Bear with me here, it will be a long story.
Back when it happened, I just heard about DotA once and didn’t even follow the game, so naturally I didn’t even know what was happening there. For me, a “newbee”, TI1 was “the TI with fountain hooks and Free to Play movie”. In reality it was a bit different.
First of all, the game didn’t launch yet, and it was probably the first public appearance of Dota 2. The game was missing a lot of DotA heroes, and the players, while being the best and most famous in DotA All-Stars, were seeing the game for the first time.
The thing that surprised me the most was that the event didn’t have its own in-game ticket. And it kind of makes sense, the game didn’t have the tournament system, and it was kind of unpolished at the time. So not only it was using “The Internal”, a test ticket, but the players were using different accounts as well. The ones that were made specifically for TI1, it seems.
That’s why you can’t really find “The International 2011” on Dotabuff or TI1 matches in Dendi’s profile. They are in a different place, that’s all.
While I was looking for the matches, I tried every obvious approach, even checking out Liquipedia, but there were no mentions of the matches or their IDs.
But in the end I found this page. It’s the official page for TI1 made by Valve. It’s very old and it sometimes “breaks” (you need to manually replace HTTP with HTTPS every time in your address bar), but there I was able to find the match IDs.
That’s how I found the matches, and figured out they were missing information about teams/players. But (aside from “double” matches that weren’t actually played) there was another thing that required attention: the matches did not have draft data.
Early tournaments were using the “old Captain’s Mode” and it seems like draft order data wasn’t stored anywhere for these matches. In case of TI1, there were no replays to parse (and they wouldn’t be supported anyway).
So I did the next best thing I could think of: I manually edited team and player IDs and names (replacing them with the “real” IDs of these teams and players), as well as rewatching every single game at TI1 to insert the draft order data.
Most of the matches could be found on YouTube, in the official TI1 playlist, but one match was very elusive and didn’t have an official VOD. I managed to find it on some random Chinese website in the end, so every match has draft data now.
Fun fact: at TI1 and TI2 teams were not choosing side and first pick independently, but rather were swapping sides every match (or so it seems, I might be wrong)
And now I present to you: the TI1 meta.
One thing that I noticed immediately: the meta is very similar to what is commonly associated with “the competitive meta”.
Another thing that’s hard to miss: overall skill level in these games is much lower than it is today. It’s common to say that “the best players back in DotA/TI1/TI2 days were like 2–3k players today” and this is very noticeable when you are watching these games.
What’s interesting to me is that a lot of the most popular heroes at TI1 are more or less popular at 2–3k bracket even to this day. But, of course, TI2 meta is much closer to what’s popular among the players.
Another thing that I found fascinating: out of all the heroes that were available at TI1, every single hero was contested, but only one hero was banned once, but never picked: Axe.
And, yes, for every single TI report I manually updated the list of heroes available in Captain’s Mode.
And now I’ll leave you with some cool graphics about TI1, including the meta graph and records.
That’s about it with TI1. You can check the stats yourself in The International 2011 report.
The International 2012
TI2 had much more heroes available, had a proper teams and tickets system, had a much higher skill level (since players were now more familiar with the game).
A lot of the moments that are commonly associated with “the first TI” are actually related to TI2. It was “THE” TI most people at least heard about.
For me personally it was TI I have some memory about. A friend of mine gave me an invite to Dota 2, back when it was relatively rare. I didn’t really stick to the game, but I remember playing occasionally during some events, and when TI2 happened, there was a custom loading screen (which was a big deal back then).
It was also the time a lot of new players started to show interest in the game. People were watching TI2, they heard about this new cool game, everyone wanted to play it.
In case of finding matches it was relatively more simple. Unlike TI1, it’s actually possible to download TI2 replays and even watch them using Source 1 version of Dota 2 client. However, it’s harder when it comes to parsing the replays: some of the additional features (including items purchase details) are not working properly, and TI2 was still using the “old” version of Captain’s Mode. It should be technically possible to at least get the draft order (and bans) data from the replays, but I only got to test this approach after I’ve got through VODs of all the 154 of TI2 matches to manually insert draft data (and recover hero/tower damage and healing numbers).
So as you might imagine, meta at TI2 was quite weird, with everything pretty much based around Lycan, Naga and Dark Seer. And Rubick. And Enigma.
You remember the famous Patience from Zhou moment, right?
Most of the games I watched had 5 heroes banned in some order: Lycan, Naga, Dark Seer (these three were usually first stage bans), Chen and Leshrac.
Surprisingly, there were 24 heroes uncontested and no heroes were only banned, which I find surprising.
The International 2013
Ah, the legendary TI3. It was the year many started watching competitive Dota. And it was also the first TI ever to have a “compendium”. In terms of stats, it was also the first TI that would have a proper “analysis” for the matches, including lane roles and stuff like that, as well as having proper draft data.
For the “modern” Source 1 era (TI3/TI4/TI5) there weren’t many issues regarding draft data or player/team names. Although it was a bit tricky in terms of additional data (warding stats, roles, item purchases and stuff like that), but thanks to Dotabuff I was able to fill in most of the gaps.
One thing I’d like to point out is peculiar the list of most popular heroes is. In a way, this meta was truly a dream for Alliance, coiling into the first 5 games series in TI history and one-million-dollar-play.
It was also one of the most popular, most loved and memorable TIs in history. Aside from “the play” and the story of Alliance getting the victory, it was also the TI that was full of interesting highlights, like the infamous Fountain Hook.
There’s not much else to say, so I’ll just leave you with the graphics. And if you want to explore the reports, you can check TI3 event here or TI3 qualifiers here (qualifiers are missing players/teams data corrections and additional analysis).
What I find peculiar is how strong Chen is at every TI by this point. The hero was in top-10 by rank every single time by now.
Since I’m including these “Item Overviews”, to prevent further questions: winrate difference is based on the item’s winrate when purchased with an early to median timing, compared to games when this item wasn’t purchased at all. It works the same for cases with “item-hero” combos, but here items stats are even more skewed.
So “100% winrate difference” for Antimage with Power Treads means “this hero was never played without Power Treads, and early/median timing led to winning conditions”.
The International 2014
The International 2014 is commonly called “the worst”, but to me personally it was the first TI I watched. And it was TI that got me into the world Dota 2 (I only started playing the game actively after this).
While it’s common to say that TI4 “was not diverse” and “had short and boring games” (and it might be somewhat true for the finals, I admit), but surprisingly TI4 has less uncontested heroes, lower median of picks (practically meaning more diverse meta) and higher median game duration.
Interestingly, TI4 meta looks a bit similar to a push-heavy TI2 in a way. But considering the popularity of Razor and Doom, it’s not surprising to see Weaver being one of the most effective heroes during the tournament, and Linken’s Sphere being one of the best items.
The International 2015
TI5 was the last one played on Source 1 engine. And it was the last TI with the number of uncontested heroes being that high.
It was also the TI that showed just how important map control and information are in Dota, leading EG to victory with Techies, Tusk and Bounty Hunter and shaking the meta around them, like a disaster. It was also the first TI to have bounty runes, introducing the roamer meta into the game.
The International 2016
Coming after the engine update, TI6 was the first one to be played on Source 2 (and my personal favorite). Engine change also led to increase in data available for analysis, especially with the update to ever-reliable Clarity parser by SkadiStats.
Because of that, every TI in the Source 2 era has a lot of accessible data to work with (and there weren’t really any challenges).
Because of the engine change, it became easier for the Dota 2 team to release new patches, which ultimately led to massive increase in the quality of balance in Dota 2, and diversity of the meta. Well, at least it’s the first time the number of uncontested heroes was below 10 (7, to be exact). Most people agree that it’s also one of the major reasons behind the stellar performance of Wings, the champions, and Digital Chaos, the runner-ups, at TI6.
The International 2017
TI7 was the last one (since TI1) to be held in US. And aside from that, TI7 was an improvement in every aspect of TI6. Even the meta is just as diverse, but now there were only 5 heroes uncontested, 1 was only banned and only 6 heroes haven’t won a single match.
And the meta became really different (both in terms of heroes and playstyle), compared to TI6 (and the reason for this difference is, once again, happened mostly because of the engine change). One thing that makes this TI really special: it was the first one played after the 7.00 update, starting the new era in Dota (and even having the completely new hero available).
The International 2018
And just like TI7 was the last one held in US, TI8 was the first one to move outside of it. This cavern-themed tournament felt very fresh and entertaining, making games more action-packed with the diverse set of heroes.
And similar to Wings, this kind of meta worked well with OG’s playstyle. It led them to victory and gave us a lot of memorable moments we commonly associate with THE International.
Meta-wise, it’s “just another TI after 2016”: diverse meta, completely different from the previous one. With TI8 came two new heroes (Pangolier and Dark Willow), and it also started the “2–1–2” era in Dota. Before patch 7.07 it was common to have either a trilane (on safe lane) or a dual lane with a roamer.
With the game being more action packed, laning stage being even more significant, while reducing the value of early jungle farming or bounty runes, it became natural for teams to have one support on each side of the map early on. Today it became an unwritten rule to have a duo of Safelane Support (position 5) and Offlane Support (position 4), who start roaming together later into the laning stage, while also securing laning for their respectable cores.
As a side note: while at the time it was very rare to see position 4 to be played outside of offlane, today it’s rather common to see “Offlane Support” being played as a “mid support”, roamer or a jungler (just like before). However, ever since TI8 the early laning became so important, that to this day it’s rare to see a hero alone on lane early on, even if position 4 is a roamer/jungler.
It was also the first time I expanded my reports and made a special category for TI8. It includes daily stats, separate stats for qualifiers, main event and group stage, as well as complete report and pre-TI8 stats. You can check all of it here. And you can also check my texts about TI8 here.
Say goodbye to Ring of Aquila, fellas. We won’t see it again until TI10.
The international 2019
And the last pre-COVID TI we had. What a journey the last two years were!
While TI8 gave us the “cinderellas” story of OG, TI9 expanded on it. While Liquid and OG in turn showed that in Dota everything can work.
This time every hero was either picked or banned, with only three heroes (Keeper of the Light, Ursa and Clockwerk) remaining unpicked throughout the tournament.
But it was not the only meta trend, going into TI9. After “The Aghanim’s Update” every hero got some kind of improvement. On the other side of things economic changes accumulated through the years (free wards, reduced cost for sentries, stacks bonus gold and more) made it easier to scale as a support. Both of these factors helped to create the most diverse TI meta, with heroes often being played on a completely new role. One such example would be the famous Carry Io by OG.
And now the only thing left is to wait for The International 10 to happen.
TI10 also the biggest one in terms of changes that happened since the last time. Including (but not limited to) Aghanim’s Shards, Neutral Items and personal couriers.
But since it’s the 10th event in the series, let’s look back at the history of TI as a whole.
And some notable milestones
I’d like to end this post with some notable milestones throughout all The Internationals.
There were total of 1435 matches played. 342 players played at TI, in 71 different teams. Duration of all the matches combined would be (almost) 941 hour. Combined time spent dead by all players is 575 hours.
The longest TIs happened in 2018 (195 matches) and 2019 (193 matches).
During these games there were 66255 kills, 2362129 last hits, 116299 creep denies, 566 courier kills, 2527 Roshan kills, 38800 observer wards placed (10034 destroyed).
Players dealt total of 149 million damage to heroes, 18.8 million damage to towers, healed for total of 9.6 million points, stunned other heroes for total of 58 hours (starting with TI6). They also made 476k map pings, 10.5k neutral camp stacks (starting with TI6), 5309 buybacks, 578 Godlike sprees and 21 Rampage.
Players who played the most matches at TIs: KuroKy (203), Puppey (198), ddc (154), s4 and LaNm (153).
Miracle- killed Roshan 69 times throughout all TIs. MinD_ContRoL killed 20 couriers total. Arteezy killed 39.3k creeps, while Dendi denied 2.3k creeps.
KuroKy participated in 2894 kills and placed 1983 observer wards.
Teams wise LGD played the most matches at TI (194), followed by EG (161) and Liquid (148). However, Natus Vincere played the most matches in TI Finals (12), followed by OG (9), then by Liquid and Newbee (7).
OG and LGD are the teams who did the most Rampages during TIs (3). However, only OG have the Rampage made during TI Finals by Topson.
And that’s about it!
Until I finish my Pre-TI10 teams report, there’s not much left to say. Let’s enjoy this TI together!