Published in


The International 10 Compendium Predictions

Hey, it’s time for some TI10 predictions. There’s no public announcement regarding stuff like tiebreakers or which teams start in which group — so it’s possible I’ll change some of these if any of these are crazily different to what I expect.

Remember that last TI there were 193 games: 144 in Groups and 49 in Playoffs. Playoffs are 4x bo1, 17x bo3, and 1x bo5 — so the number of playoffs games is possibly 41 to 60. The overall number could be higher if there’s tiebreakers in the Group Stage. These numbers are discussed here because they affect various other categories.

Also to note that there’s been Dawnbringer added to pro Dota, and minor patches since we last saw the ‘big’ teams play — however there’s been some lower tier pro matches which often represent the top level well in a broad way.


We’ll start with the Tournament section, since it’s generally the easiest to predict.

Total number of games played in the Main Event. The Singapore Major had 2.52 games per bo3, the AniMajor had 2.71. This very inconveniently puts us just on the border between 45–49 and 50–54 — which I’ll take the upper side on. 50–54.

I’d be very surprised if we don’t see 101+ total picked heroes. 102 have been picked twice or more in 7.30d in 211 games and the average TI10 attendant is much more experienced than active non-TI 10 Dota players who are still playing matches — leading to a wider hero pool to pick from.

Total number of heroes Banned will either be 101+ or 91–100. I’m going under (i.e. 91–100), simply because there were 99 heroes banned 2+ times this patch, and there are fewer teams participating in TI than in general play. This really is highly dependent on which teams make the biggest loser bracket runs since they play the most games.

Most Combined Total Kills in a game depends on how long the games will be, coupled with average kills per minute. The two categories which make the most sense here are 91–100 or 101–110, but there’s been only 5 out of 4449 games this year in the latter category and 40 in the former. I’m going with 91–100 but wouldn’t be surprised if we saw one category above or below that.

Longest game of the tournament. As expected, The International often sees slightly longer-skewing matches simply because of the preparation and close skill of the teams. There have been some long games this year, the longest being 107:20 between Elephant and RNG and just two more > 100 mins. 3 out of 4448 is a bit rare for my liking though and so is 90–100 mins (funnily enough hasn’t been one this year). I’m torn between 80–90 or 70–80. I’m going to go with 70–80, simply because of the added influence of Tier 5 Neutral items at 60 mins will break stalemates, and hoping that there’s not one game that blows me up.

For the shortest game of the tournament, Aster vs IG was 14:57, the only sub-15 minute game this patch (~700 games). There were only 19 sub-15 min games this year (~4448 games), and Aster/LGD was the only other one which was between two TI10 teams. 2.09% of games this year have been < 20 mins, but just 0.71% on this patch. I’ll go for value by playing 15:00 to 19:59 — however it could reasonably be either side (one super short gg, or just tighter matches all round).

In 7.29/7.30 there’ve been 2862 pro games. In just 16 (0.56%) did a player have 26+ kills, however over the expected ~193 games this represents ~66.2% odds so I’ll take it. 26+ as most kills in a game. With similar logic it’s very close between 15–17 deaths or 18–20, but I’ll go under with 15–17 as most deaths in a game. Most Assists in a game is one of the easier predictions, it’s happened so many times this year that it’d be shocking if we don’t get 36+.

For Highest GPM it’s always a tossup between 1000+ GPM and 900–999. Alchemist is not super relevant in the meta, but right now it’s possible for so many heroes to reach this goal that the favourite surely is 1000+.


There’s not a huge sample of pro matches on 7.30d (~100 games), however there are a few heroes which appear way more commonly as picks/bans than all others: Beastmaster, Tiny, Monkey King, and Invoker. A lot of people are trying to use high level public data to supplement this, however public data is without bans — and I expect Beastmaster to tank most of those bans and be the most banned hero. This opens the door to Tiny or Monkey King to be more the most picked up — both are flexible heroes which don’t expose most of the draft. I’m going with Tiny — even if he does get banned in games his pick-rate should be through the roof..

The hero with the highest winrate is generally a niche 3rd phase pick. I’m going to go with Troll Warlord for two reasons: the hero is 10–4 in this patch (~211 games), and it’s a comfort pick of past patches which teams could pull out of the bag. I would go Anti-Mage, but I think it might be something which SA teams pick (they do have some excellent AM players, but I’m not overly confident in their prospects at this event). I’d also consider Lycan, but I think it’ll be picked a bit too much.

For highest / lowest categories I’m just going to look at stats of 7.30d (despite the small sample) and 7.30 as a whole and list my pick and other considerations.

  • Hero with highest Kill AVG: Ursa, Morphling, PA?
  • Hero with highest Assist AVG: Ancient Apparition, Earth Spirit, Silencer
  • Hero with lowest Death AVG: Anti-Mage, Lycan
  • Hero with highest Last Hit AVG: Medusa, Sven, Terrorblade
  • Hero with highest XPM AVG: Anti-Mage, Templar Assassin
  • Hero with most Kills in a game: Ember Spirit, so many viable options here.
  • Hero with most Last Hits in a game: Naga Siren, Gyrocopter, Terrorblade.


I think LGD will win The International 10. It’s always so hard to predict how a metagame will develop — so the safest option here is a team which has been consistently good. This basically means LGD or EG, however Secret were excellent until they qualified for TI (after which they looked mediocre).

For highest / lowest categories I’m just going to look at stats of 7.29/7.30 and see which teams are looking the best at these stats, and adjust them based on my expectations for the teams.

  • Team with most Kills in a game: Team Secret, PSG.LGD Vici Gaming
  • Team with highest Kill avg: PSG.LGD, EG, VP
  • Team with fewest Deaths in a game: VP, PSG.LGD, EG
  • Team with most Assists in a game: PSG.LGD, EG
  • Team that wins the Longest game: Team Secret, Vici Gaming
  • Team that wins the Shortest game: Virtus Pro, EG, PSG.LGD
  • Team with highest Game Length AVG: T1, Vici Gaming
  • Team that picks the Most different heroes: Team Secret, PSG.LGD (notes)
  • Team that picks the Fewest different heroes: Thunder Predator, Fnatic


Looking at player stats for 7.29/7.30, and adjusting those based on my expectations for the event.

  • Player with highest Kill avg: Abed, NothingToSay, Nisha
  • Player with most Kills in a game: Abed, NothingToSay, Nish
  • Player with lowest Death avg: Ame, Arteezy, NightFall (?)
  • Player with highest Assist avg: Y`, Dy, XinQ
  • Player with most Assist in a game: Dy (Ancient Apparition!), Faith_bian, XinQ
  • Player with highest Last Hit avg: 23savage
  • Player with most Last Hits in a game: Poyoyo
  • Player with most GPM in a game: Arteezy, Ame
  • Player with highest GPM avg: Ame, Arteezy, Poyoyo
  • Player that plays the Most different heroes: Zai, Faith_bian

Feel free to send me comments here, or on Twitter on how you think the event will go down!



Professional Dota 2 statistics

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store