Hey it’s that time of the year again when compendium predictions come out and everyone gets flamed for how badly they did last time. Remember that these are very very difficult to get right - most of the prediction categories are basically just crapshoots, with a single unusual game able to knock many predictions out of the park in one fell swoop. Here’s my predictions, with some rationale for a few select ones.
No surprise to see Sand King as the most picked hero—he’s picked 25% more than the next most picked hero this patch in pro matches. Nightstalker is the most banned hero in 7.06 pro games, but if you filter to only the 18 invited teams it gets a lot closer for Io. I’m going with NS, but I’d not be surprised by Io trending up as I expect it to be permabanned against a few of the top 6 teams (who’ll play more games).
Highest winrate is some niche hero you think is going to be picked just enough times and still perform well—I went with Nature’s Prophet. Queen of Pain and Zeus currently average the most kills and assists respectively so sticking with them seems appropriate (and Queen of Pain to get most in a single game). Last Hit Average I’m going with Anti-Mage over Naga and Medusa because I think it’ll be a hero which definitely gets picked 5+ times. I think there’s a real chance Naga and Medusa fall short of this despite some teams using the heroes recently. For similar reasons, going to take Anti-Mage for XPM and lowest Death averages. I do believe we’ll see one Medusa that gets 800+ CS.
Average Kills and Deaths are all based on the 7.06 averages in premium events with some normalization for regional strength and also for team strength (teams which are strong are likely to play a few short games against the weaker teams). Most Kills are a gut feel on PSG.LGD playing a very long stagnated game at least once since they average the longest games despite a 58% winrate. I expect at least one of Virtus.pro’s games will be a crushing, lasting a very short period of time with them not dying a lot.
I expect Liquid to go far and run a large number of heroes. Both their supports and cores have significant hero pools.
SumaiL, Miracle-, Somnus丶M and fn all average 10+ kills per game, but I expect EG to play longer games as the tournament progresses so I’m going with SumaiL. I am going with Kuku for most kills in a single game, as I think teams mostly will have less respect for TNC Predator and will refuse to give up even when Kuku’s QoP is dominating! Other good picks here, in addition to the ones listed above, are Moogy (a.k.a Moogy) and QO.
For lowest death average it’s between BurNIng, RAMZES666 and Raven all of whom are consistently minimal feeders. Since I expect VP to do very well, it’s likely they get tougher games in both winners and losers bracket, whereas I expect Invictus Gaming will either scrape into top 8 coming out of groups or fall just short of it. I’m going with Burning.
kaka has the highest normalized assists per game (pieliedie in 2nd), so I’ll go with him, but I do expect YapzOr, zai, JerAx or BoBoKa to have one insane Kunka/Rubick/Earth Spirit game and take this category. It could even be a core Zeus in a long game (Miracle perhaps). I’m going to go with Boboka for single game stats, and Kaka for the average.
Fata, fn, Chappie, EternaLEnVy and Sccc average the most last hits per game, but all but Sccc played in their regional qualifiers which were filled with many weaker teams. I’m going with Sccc, although I do expect a Miracle Anti-Mage to set some records.
Lil has played 26 different heroes this patch, the most of any player. That said, part of this includes VP’s shenanigans at The Summit. Instead I’m going with MP here, playing 25 different heroes in a lot more games than Lil.
50.5 is the expected number of games in the main event if every series is played to it’s average length. Kiev saw series go slightly above the average (~ +0.07/game), but we’ll have a slightly different setup given the double elimination bracket with fairer pairings. I’m going to stick with this range: an approximation of 52 games seems good.
Looking at historic stats for picks/bans can always be misleading as teams experiment a lot going into TI (and also have more narrow drafts). Last year we had Wings to try out exotic picks, but without someone like that this year I believe it’ll be a bit more narrow.
Only six games have gone beyond the 80-minute mark this patch of the 834 games; and only three of those went past the 90-minute mark. To me it’s close between 80-89 and 90-99, but I’m going to be conservative here and go with 80-89. Similarly, only two games have ended before the 15-minute mark and 14 more before the 20-minute mark. I’m going with 15-19, although there’s a fair chance that 20-24 is safer.
Only one pro game this patch has broken the 1k GPM mark (SumaiL’s Alchemist), but some performances have come very close (Miracle’s Anti-Mage got to 994). I’m going to be pessimistic here and say 900-999 is the top: I don’t personally think Alchemist will get picked very much and Anti-Mage should peak at just under 1k (only 3 have ever broken 1k GPM).
Most kills / deaths / assists are all based on how long the longest game goes, so I’m going to just take 95th percentile values based on the pro performances this patch.