The International 2018 Predictions: Heroes
The final part of our prediction series and probably the hardest. With limited information, several game-altering patches since the last international Major and the general meta volatility during the biggest tournament of the year, once again, we are only providing you with our educated guess and reasoning, so feel free to change things if you disagree.
Most Picked Hero
Looking at hero stats from The Summit 9, we can see that 
Necrophos, 
Pudge, 
Crystal Maiden and 
Winter Wyvern dominate the list. Given how the first three received some heavy nerfs in 7.19 and how Winter Wyvern was by far the most successful hero of the tournament with 70.59% winrate, it might seem rational to assume her as a winner in this category.
But there are some caveats. First of all, The Summit 9 was more or less a local tournament. It had no proper EU representation, no CIS or Chinese teams and the only SEA team at the tournament is hardly an SEA team, with three Western players and a Western coach.
Secondly, Winter Wyvern also received nerfs in 7.19. They weren’t as hard, but they are still nerfs. Even if we assume that she is going to stay popular, she will start to get banned and that will cut directly into her pick rate.
Finally, the International is always the time when professional teams tend to go back to the basics, prioritizing comfort picks, such as 
Enigma, 
Batrider and 
Tidehunter. Regardless of meta, these heroes always make a comeback closer to TI.
Of all the heroes who can fit into this category and were recently popular, 
Beastmaster probably fits the description the most. It is a comfort pick, he is a good laner, has global presence and provides his team with vision. He was also the most picked hero of the China Supermajor and isn’t popular enough to be banned consistently.
The nerfs the hero received in 7.17 and 7.18 are noticeable, but not nearly as heavy as what Windranger, Warlock and Skywrath Mage received since the end of the TI8 qualifiers. Beastmaster was the 4th most picked hero across all qualifiers and there is a good chance he will be the most picked hero of the International 2018.
Most Banned Hero
Io is almost always the most banned hero during the International, unless he is completely gutted with nerfs. If you believe that Io is not a complete and utter mess, going with him might actually be a good idea—he is one of the staple TI comfort bans.
If, however, you think Io is going to be completely out of the meta based on him no longer being able to save his teammates as reliably, there are some alternatives.
There is an argument for 
Enchantress, who filled the 
Chen niche as a lane bully with summons. She was the most banned hero of the Summit 9, but, once again, small and heavily restricted sample is non-indicative.
There is also Lycan, who was the 3rd most banned hero during the TI8 Qualifiers and the second most banned hero in China Supermajor. His nerfs were limited to armor and base damage reduction, so he is still an early game threat that can finish the game against you before you get a chance to play Dota. He is also heavily favored by the Chinese scene and there are 6 Chinese teams at the tournament. And he also has a history of making comebacks during the TI.
Hero with the highest winrate
When looking for a hero with the highest winrate with at least 5 games in a tournament that will have almost 200 games, we need to think of some cheesy last pick that will be only picked five to ten times, but will leave the enemy absolutely no chances.
Broodmother would probably be a good guess, but currently pretty much every team either knows how to play against her or ban her early, if they see their draft is susceptible to being Brooded.
Huskar is also in the category of cheesy heroes, but, once again, professional teams are generally aware of the hero and frequently won’t let the enemy team to draft 
Dazzle or 
Oracle with Huskar, limiting how aggressive and dominant he can be.
There is one hero, however, that is slowly creeping out of the shadows. 
Arc Warden is a cheesy hero with an extremely high skill ceiling that is best when picked last. He is versatile enough to adapt to different game situations, can stall the game if necessary and has 75% and 80% winrate in the Supermajor and the Summit 9 respectively.
Hero with highest kill average
There are two ways to think about this category. We either once again look for a cheesy or off-meta hero that will be picked at least 5 times and will have incredibly good games, being the last pick. Or we think of some hero that is able to at least trade his life for a kill consistently and has high burst damage.
If we believe 
Phantom Assassin will be picked at least 5 times during the tournament, she can be a good choice. She was the best killer in China Supermajor and had the highest kill average in the last TI and second highest in TI6, while being buffed consistently throughout the year. She also kind of fills both categories, being an off-meta pick and a hero capable of suicidal trades.
Ursa can also be a good guess, once again filling both categories and even receiving some buffs after the Qualifiers. Huskar is probably in the same boat.
Tinker and 
Shadow Fiend were the best killers in the Summit 9, but the problem with both these choices is that they are highly inconsistent and given a bad enough start can become a non-factor for at least half the game, dragging the average down.
Given TI history, recent trends and all the buffs, we will bet on Phantom Assassin as the hero with the highest kill average in at least 5 games.
Hero with highest assist average
Ancient Apparition was the third in the Summit, first in ESL One Birmingham and in top2 of both TI6 and TI7. All his abilities qualify for assists and two of them can grant him an assists from across the map, while Ice Vortex has an incredibly long cast range.
With the amount of games we can be almost sure that AA will be picked at least 5 times, and regardless of how the game will play out, he will most likely have a decent amount of assists, as one of the stronger kill-enabling heroes that rarely get the actual last hit.
Hero with lowest death average
Weaver was one of the more popular meta heroes in the Summit 9, with strong lane presence and decent core potential. Moreover, he can also be played as a utility/damage hybrid with the help of 
Aghanim's Scepter, so he will be played at least 5 times, while being a highly elusive target. When played as a hybrid, he is also more likely to stand in the backlines, providing a safety net for his other core, so is less likely to get initiated on.
Another good option is 
Phantom Lancer, who is very elusive and hard to kill. Similar logic can be applied to 
Morphling. Our bet is still on Weaver, since he was starting to get momentum as an offlaner and in the current dual lane meta the offlane is generally the safest and most farm-rich lane to be in.
Hero with highest last hit average
Anti-Mage would be the go-to choice in the last two year and is still a rational pick, but with the changes to the meta he isn’t as clear of a winner as he used to be.
Recently popular heroes like 
Meepo and 
Templar Assassin actually consistently overtake him in the last several tournaments and Meepo is one of the heroes that will probably be played at least 5 times, but won’t be played too much to drag the average down. Both him and TA also generally start flash-farming much earlier than Anti-Mage and with games ending sooner and Anti-Mage being a pretty weak hero in the current meta, either choice can be rationalized over AM.
The problem with Meepo and TA is that their games are generally shorter, with them either becoming dominant and finishing the enemy, or getting crushed and becoming unable to defend their own highground.
So when taking into account the average game length being a major contributing factor to the average amount of last hits, for this year we suggest going with either 
Terrorblade or Phantom Lancer. They are flash-farmers, are not complete creeps in lane, can relatively safely cut waves and are more than happy to take the game late.
Statistically, Phantom Lancer slightly edges out Terrorblade in the Summit and was more popular during the qualifiers, however Terrorblade feels more similar to Anti-Mage, so personally, I am going with TB.
Hero with highest XPM average
Hero with most kills in a game
Any core hero can have this one amazing game where they go completely out of control and get kill after kill, so this one is going to be impossible to rationalize fully.
But there are ways to make your guess more educated. What hero is more likely to get the last hit on the enemy? What hero, when given advantage, starts hunting for more and more kills? What hero can’t necessarily end the game, despite getting kill after kill? Finally, what hero can start getting kills very early?
Essentially, we are looking for a mobile mid lane hero with good lane presence, lower flash-farming capabilities to encourage ganking, high burst damage and weak objective taking.
Last year this category was taken by 
Queen of Pain, the year before it was a tie between 
Invoker and 
Mirana. Throughout the season, 
Storm Spirit was the hero with most kills when played by Mineski. All these choices are viable, make sense and can be justified. Considering we believe the “Most Kills” in a game will come from a PSG.LGD vs. Fnatic bloodbath matchup, with Maybe being the slayer, we will go with his most consistently successful ganking hero and pick Queen of Pain for this category.
Hero with most last hits in a game
We might be risky by going against the grain in our last hit average prediction, naming Terrorblade instead of Anti-Mage, but there is going to be this one Anti-Mage game where will get absolutely out of control. There is always this one game.
It won’t be enough to make up for all the times the hero is going to be crushed in lane and be a game-losing creep in a sub-25 minute match. His average is still going to be pretty low. But this one time he is going to show us all non-believers that he is a strong hero, maybe not statistically, maybe not logically, but it will be enough to convince half the Dota community that first pick Anti-Mage is a good idea.
thanks for making these predictions, i dont have so much clue/knowledge about pro play but i want to earn some pts :D
Nice article
I doubt Meepo will see 5 games played. People know who the Meepo teams are, they’re either going to show a draft which is anti-Meepo, or straight up ban him.
edit please, where there are fewer deaths on average, the hero PL is rooked
""but it will be enough to convince half the Dota community that first pick Anti-Mage is a good idea.""
Ruddy brilliant stab at all the low level players out there who are trying to learn!
Nice description of HERO WITH HIGHEST XPM AVERAGE
Meepo
XD
most banned dark willow mb?
I liked the highest xpm section, concise and to the point :D
U missed spirit breaker as the hero with most assist overall
And broodmother on lowest dead
And summit 9 is the most worthless tournament to base predictions on literally the smallest which u did anyway
@EtdAwesome, if meepo gets 5 games, bet seems safe. if meepo doesn't get 5 games, who do you pick?
There are a ton of hero that can get a high xpm.
So lack of better choices, i go with meepo, hoping he gets 5 games, and decent xpm even in losing ones.
Hmm meepo is op this patch, he can also work against axe or something.
is there any team play Meepo besides Secret and Fnatic? i doubt he will be played 5 times tho
For xpm i will guess storm spirit since he's more reliable than meepo for picks and one of my personal highest xpm
Ban will be very diverse and they said phoenix worthy of a ban.
Assist is very diverse (usually supports who can cast long range and have less damaging spell). I personally go for either disruptor or SB
Lowest death will probably morphling, especially rtz's and miracle's
For winrate, my best hope probably visage considering 33's visage and w33's visage
Good choice on assist but I think pos 4 Zeus or core Zeus for the most kills would be ideal as well..
Highest XPM is definetely weaver with his 35% XP gain talent on lvl 10.
@ AYAYA CLAP
almost nobody goes for the xp talent for weaver. Mana Break too good
tnx:DD
veri interestinga (no)
tyyyyyyyyyyyyyy
And for fantasy?
thanks <3
I think zeus is worth considered as highest assist avg, huge aoe of static field and ofcourse the ult. in a teamfight. Instant 5 Assist each full teamfight.
I really liked this article for a very detailed analysis of who will have the highest experience per minute in this very tournament. All the factors have been included and precisely calculated.
I laughed and stop watching when I see top pick is Beastmaster and top ban is Lycan. LMAO PREDICTION
what a noob predicksion
Beastmasterm, lycan, phantom?......What The Fuuuuuuuuuuck.........
Noob prediction ever
Is there any site where I can check current hero/tournament statistics?
lulPredictions
gg predict
\
Lol this brewmaster prediction is NOT paying off ^^
KawaiiSocks, you are so idiot that I will report you. I never checked your stats until now cuz I see that my predictions are wrong, but 4-5k player shouldnt predict. Ive done all of your predictions with rush like saying "this guy is posting on dotabuff on the main page, it must be sth." You idiot made all of us wrong and thats your fault. WE SHOULD NEVER TRUST PEOPLE LIKE YOU AND WHO LETS YOU TO DO POST THIS IS ABSOLUTE MAD! I am so mad at you, you will get reported. You just used our rush of make predictions when time was running up, but you made our predictions false. Thx for your brain. Thx for your "CHINESE TOURNAMENTS ANALYZES". IDIOT...
Trash predictions , these heros are not in meta ... Waste of points
Yeah what a fail