Choose Your Champion: HCT Winter Playoffs
Hello and well met! Today I would like to share with you all my statistical analysis of deck archetypes for the HCT winter playoffs. Last championship in the fall, I used this method to choose Bloodyface, who ended up taking second place. I figured I would give this another try and improve on what I did last time. I know it is a bit late and many people have already chosen champions, so I apologize for not posting sooner.
After first seeing the deck lists, I used HSreplay to create a custom scenario in the matchups tab of the meta overview where each deck is as prevalent as it will be in Winter Playoffs. I then sorted the decks based on expected win rate using legend data from the last seven days. Here were my findings:
The first immediately obvious trend in the chart to the left is that aggro, if matches are played at the level of the average legend player, is likely to do very well in playoffs. The most common deck brought to the tournament was midrange hunter, and this turned out to be a very smart move as the deck is good against the vast majority of this field. In fact, every deck with a win rate above 50% against this field is an aggro deck, or in the case of dragon inner fire priest, at least a proactive one very capable of stealing wins before tools like Psychic Scream come out. Following those are the more well rounded combo decks with cubes and quest rogue, which create a big problem for control lineups while still having a shot against midrange. The slower combo decks look really bad based on these numbers, but it isn't fair to rule them out. Last time I did this analysis for HCT Fall Playoffs, Shudderwock Shaman looked like it was heading into an extremely bad field, but ended up performing much better than expected, so it is possible that decks like Cloning Gallery do better than this chart predicts.
So these are the base win rates of each deck in a simulated ladder where they each play all of the other opponent decks once. Obviously, this is different from the tournament, where bans can get rid of the most polarizing decks. While I cannot possibly know the inner thoughts of the 16 players in HCT Winter Playoffs, I did try to identify the most problematic archetypes for players to face, then remove those decks from the field and determine the average win rate of the lineups without the decks that will be banned. Since this method doesn't ban any player decks as a counter ban, win rates of all line ups will be inflated above 50%
Based on the chart above, Jing and Bunnyhoppor look like the strongest contenders against this field based on lineups alone. Tansoku and Bloodyface also look set to do well, and Yueying, Viper, Tyler, and Ike also have lineups that should give them a solid chance. The rest of the field would have a hard time if the format was a giant swiss round. Luckily for them, the groups system could protect them from some bad lineups, but I suspect even that can't save Definition or Faeli from their terrible odds according to this data. However, it is worth pointing out that HSreplay combines classic UI mill Druid with Azalina with Hakkar Gadgetzan Druid as a single "Mill Druid" deck. I had no choice but to use the combined data for match up calculations. One of these players may be in better shape than the data looks, because Faeli brought the UI version of mill while Definition brought Gadgetzan/Hakkar.
While all players must be equipped with lineups capable of contesting the entire field in order to have a chance at winning playoffs, it is more important for players have strong lineups against the rest of their groups. After all, only half of players will survive the group stage, and it is possible to play the same person twice in groups. So lets try to predict who will make it to the top 8:
Group Analysis:
Wow! How do you know that Tansoku will beat Yueying 48.05% of the time? Well, I don't. But I do know that if Tansoku and Yueying have their best decks against each other banned away, an average legend player piloting one of Tansoku's other 3 decks will beat another average legend player piloting one of Yueying's other 3 decks 48.96% of the time based on HSreplay stats. Over the course of a match, this difference in chance for each game means that Yueying has a 51.95% chance of taking a match victory. Obviously, the highest tier of play may have slightly different match ups, but I hope that the trends of the match ups would still be similar.
To determine the average match win chance, I had to account for the fact that each player's first match is predetermined. Therefore, they can play the same person twice and the chance of beating the player they play first is more important than the chance of beating other players. The importance of each win rate can broken down by thinking about it like this:
-The first match is against one person
-The second match is against one of two people, who are close to equally likely to win
-The third match can be against any of the other three people in the group, but it is only played half of the time
Knowing these three statements to be true, I assigned a value of 7 to the match against the first opponent, and values of 4 to the matches against the other two opponents. I then divided the sum of all these values by 15 to find the average match win chance. At this point, the chance of leaving a group was determined by finding the probability of getting WW or WLW or LWW and dividing the sum of these probabilities by the sum of the probabilities of WW, LL, WLL, WLW, LWW, and LWL (this sum should be 1 if I did everything right).
That's enough math for now. Let's talk about what the data says! According to the group A predictions, Viper is the most likely to move out of the group, while Tansoku is also likely to do so. Viper went with an anti-priest strategy for his lineup, and it paid off because everyone in his group brought the class. This makes him a relatively safe pick, and it helps that he brought a solid lineup based on his win rate against the entire field shown above. Noblord, however, also tried a hard counter lineup, but his lineup is designed to beat control priest. Since Viper didn't even bring priest and he is Noblord's first opponent, Noblord is in for a rough time. I think out of all lineups in this field, Noblord's is my least favorite. It feels like he is letting his opponent pick what they want to play against. Aggro lineups will ban the odd warrior and have a field day against the AFK until I OTK decks he brought, while other OTK decks can just ban an OTK or the hunter and farm the odd warrior for free. Noblord is a big name in the Hearthstone community though, so maybe he can show us some of his skill and pull off a miracle.
Here in group B is one of the players identified earlier as having one of the strongest lineups against the HCT Winter Playoffs field: Jing. Jing's aggressive lineup featuring Zoo is well equipped to tear through the greedy field, and he is likely to dominate Faeli, who has one of the weakest lineups in the tournament. The cubes and Mill Druid will get run over by the aggro, while Jing can just ban away the control priest. However, this control priest is a pain for Faeli if he has to face a lineup like Bloodyface's with Quest Rogue.
One of the first games in the group is between Bloodyface and Tyler. This will absolutely be a match to watch, as it could so easily go either way. But the first time they play won't be the important one, losers finals will likely feature these two playing again, and only one of these top players can move on to the top 8. What I predict happening is that Tyler and Bloodyface will play, and the loser beats Faeli while the winner fails to stop Jing's aggro. These two top players will then have to battle a second time for the loser's spot in top 8.
Here in group C we have a great example of how the meta in a group can be very different from the meta of the entire tournament. Caimiao has a greedy anti-control lineup that won't get far against Jing's or Bunnyhoppor's aggro, but his group doesn't have much aggro in it, allowing him to have a good chance of getting out. He needs to ban away Quest Rogues and Midrange Hunters, but there is only one midrange hunter in group C and Caimiao doesn't have to play him first. That midrange hunter is Ike, a fan favorite many were hoping to see do well. Unlike Viper and Noblord in group A, Ike went with a very safe lineup that has a chance to beat pretty much any strategy brought to playoffs. This means that more than other matches, matches featuring Ike will rely on plays more than the cards brought. The only lineup that really spells doom for Ike is Bunnyhoppors aggro one with pirate rogue. Jing's Zoo is actually pretty bad against what Ike chose to bring, and its presence gives Ike a much better chance if he has to face him. It is therefore ideal for Ike to get first in his bracket, because Bunnyhoppor is likely to do very well in his, and getting first decreases the chance that Ike will have to face Bunnyhoppor earlier.
ThunderUP made the unusual decision to bring secret hunter instead of midrange, and I don't think it was a very good one. Secret hunter is much worse against all priest types, so bringing it really limits ban flexibility. This tournament is light on aggro and heavy on combo, so having secret is a weakness against most opponents. Definition was the only player to go for wall priest, which has recently proven itself to be pretty strong on ladder. His lineup actually seems really good against token-style aggro decks, but he unfortunately has to play Caimiao first, which is a terrible lineup for him to face. If he falls to Caimiao however, he could make it out of groups with the other two favorables as long as he doesn't have to play Caimiao again.
In the final group, D, there are 3 similar lineups and one very different one. Roger and GeoLionKing brought the exact same lineups, and Bobbyex has a similar one with a more defensive cloning gallery priest and malygos druid instead of evenlock. Because these lineups are so similar, all of the matches between these players should be very close and very dependent on player decisions. Bobbyex's choice of Malygos Druid and a more defensive cloning gallery build may help him against aggro if he gets out of groups, but they make him unfavored against the similar lineups from Roger and GoeLionKing. I'm counting on Bobbyex though to beat Roger, I really don't want to see him get out of groups due to his history of cheating. Roger doesn't even deserve to be at playoffs...
While the other 3 players in his group brought anti-control lineups, Bunnyhoppor has full aggro. He has a similar lineup to Jing, but Bunnyhoppor's group is far more favorable for it and it therefore almost guaranteed to get out of groups. Bunnyhoppor predicted the right meta, and will be rewarded for it with the only chance above 65% of leaving the group stage.
Final Thoughts:
With the tie for the best lineup against the field and the only chance greater than 65% of leaving the group stage, I am confident in my pick of Bunnyhoppor for Choose Your Champion. Ultimately however, Hearthstone is as much about skill as the decks you play, and Bunnyhoppor has already proven himself to be championship material once by qualifying for Blizzcon and winning the HCT Summer Championship. I think he is capable of a repeat with the great lineup he brought for this weekend.