Sunday, November 10, 2019

Arlington Qualifier Winning Lineup - Quick Explanation


The most popular and successful lineup at the moment is Evolve Shaman, Tempo Rogue and Combo Priest. There is a very good reason for that: those decks do very powerful things that no other decks currently can match (mostly getting a lot of stats on board extremely quickly and bursting heroes down). Now you might know I am not very fond of aggro decks and the idea of playing mirrors that are mostly determined by starting hands did not appeal to me, so I looked for an edge elsewhere.

The obvious place to start is Warrior, as Rogue and Shaman share a weakness to Control / N'Zoth Warrior variants, so you can swap one of the default decks with Warrior and force a Warrior ban, getting a small edge. After doing extremely poorly with Quest Shaman at first, I got the hang of it and realized that is extremely well rounded and doesn't really have bad matchups out of the common decks, so I locked that deck in as well as since Ienjoy having the opportunity to play Hearthstone every round and not only play matchup roulette. I wanted my third deck to also have favorable matchups against Rogue and Shaman and, well, I couldn't find any obvious one by looking at stats. I talked with Eddie who suggested Resurrect Priest (also known as Big Priest without Barnes). I was a bit skeptical at first but as I practiced I did better and better with it. I started with his Bucharest-winning list then made some small changes for the open cup meta.

One thing I told myself I would do after witnessing Eddie's victory is to stop relying so much on statistics. They are very useful when it comes to quick mulligan advice or to get an idea which decks are the most popular, but the '5 to legend' global winrates will be very different from the winrate a top player can get. The key is to find out which decks have higher ceilings and to navigate towards decks that suit my play style and from which I can extricate the more obscure edges.

So after tuning the lineup for 10 or so open cups this is what I ended up with:



Warrior is very bad against N'Zoth Rogue, Malygos Druid, Highlander Hunter, Highlander Mage, so there will be some trouble or auto-losses against dedicated Warrior target lineups. However, those decks are all lacking in early interaction and can't beat the default lineup so we're just going to have to hope to get through the initial rounds. This is why there is no Hakkar in the Warrior list, as Druid would be the ban and it is not necessary against any other archetype. There is a Zephrys the Great though, which is flexible enough to be useful in any matchup and can break the mirror or other control matchups. I would like to avoid running Brewmaster / Seance but they are a necessary evil to win the Elysiana mirrors. These matchups are pretty fun though as they require a lot of situational awareness to know if it is better to get an extra Elysiana or an extra N'Zoth, for example. Against most lineups you have the luxury of banning Combo Priest, Quest Shaman or even Evolve Shaman, as this lineup farms Tempo Rogue and most Warriors as well. I know the Shaman and Priest matchups against Rogue are not really favored according to statistics but that is not my experience at all over a pretty significant sample size.

I would highly recommend this lineup if the meta does not shift significantly.However, all these decks will require a lot of practice, especially to navigate control mirrors. Just remember that as good as Warrior is against Rogue and Evolve Shaman, it is way way worse against the anti-Warrior decks and requires a lot of mastery to pull off an occasional win. My first round involved a game that took 49 minutes and went to the turn limit double face explosion.... so yeah that can happen. Note that draws count as both a win and a loss on Battlefy and won't be a rematch.

Shout-out to Odemian who inspired my Warrior list and to Eddie for the Resurrect Priest idea! Shaman is super standard but remember to always use the best Shaman, King Rastakhan.


See you in Texas!

Sunday, August 18, 2019

Masters Tour Vegas and Seoul Statistics

I have compiled some statistics from Masters Tour Vegas and Seoul, mostly for my own enjoyment. There is still a lot of interesting stuff here so I'll share some of the data and observations, even though the sample sizes are quite small so no solid conclusions can be reached.

Most interesting to me was that for the first time this year we can compare the win rates of Grandmasters to the rest of the field. High win rates could mean that the format is very skill testing or that Grandmasters are way better than qualified players, and low win rates could mean that the format is not skill testing enough or that there is no meaningful difference in skill level between a player that can spike a qualifier and a player that has been hand picked by Blizzard as one of the best in the game's history.

For the record, I do feel that some Grandmasters are way better than the average player in these events, even if I would not rate all of them highly. I also feel that any Best of 3 format is not going to give a significant enough edge to better players and no matter if Conquest, Last Hero Standing, Specialist or another format is chosen, the most important thing to make Hearthstone feel meaningful and worth practicing again is a return to Best of 5 (or even Best of 7).

Here are the win rates for GMs overall and separated by region for both events:

         Vegas  Seoul
Total  55%    52%
NA     52%    52% 
EU     60%    55%
Asia   51%    49%

I think Hearthstone would be healthy if the best players could achieve around 60% win rate against top competition. Sadly, the data reflects that there is not really much difference between GMs and other players for NA and Asia, and that the EU Grandmasters are doing quite a bit better. It will be interesting to revisit after Bucharest, especially if the format changes so we can evaluate the impact of different formats on the best players' win rates. Also interesting to note that win rates got worse between Vegas and Seoul. Is the new metagame less skill testing or are non-GM players just closing the gap as the year goes on? There are not a lot of meaningful ways for GMs to compete since they aren't allowed to play qualifiers so the less disciplined of them could be getting rusty without a lot of convenient ways to practice.

Other interesting tidbits:

  • 39/48 Grandmasters were in Las Vegas and 40/48 in Seoul.
  • European GMs are the most dedicated when it comes to showing up even though the EU event has not taken place yet. 13 GMs from Europe went to every event, compared to 10 for NA (11 if you count Gallon) and 11 for Asia
  • The bar for qualifying to extra events is placed very high at 9 wins. There were only 7 9+ win performances from GMs in both events, 2 of them coming from Bunnyhoppor.
  • Most wins, adding both events: Bunnyhoppor, Gallon and Kalàxz (18), Narina and Caimiao (17), 9 players tied at 16.
  • Most wins for GMs: Bunnyhoppor and Gallon (18), BoarControl, Hunterace and Seiko (16).
  • Most wins for NA GMs: Eddie (14)
  • Most wins for Asia GMs: Alutemu (15)

NA and Asia GMs have very disappointing results overall but it also seems more difficult than it has ever been to string 8 and 9 win performances, since almost no one is doing it. Variance seems very high and it is funnily enough always possible to talk about consistency when handpicking certain players. It is noticeable that some players have consistent results (though a normal result of chance), and it can be speculated that some players give more consistent efforts than others, but I really hope that the narrative built around so-called consistency dies down.

Most consistent players:
Bunnyhoppor: 9 wins twice
BoarControl, Hunterace, Seiko: 8 wins twice
Eddie: 7 wins twice
Kolento and Orange: 3 wins twice

Least consistent players:
Feno: 9 and 0 wins
Muzzy: 9 and 1 wins
Justsaiyan: 8 and 0 wins
Surrender: 8 and 0 wins

I always go back to the same suggestions, but I think that these events need more rounds and / or more games per round to help flatten variance a bit. I also wish that the bar to requalify for events would be lower than 9 wins (75%) win rate. Only 21 players got 9 wins in Vegas and 23 in Seoul, which is less than 10% of the field. In contrast, professional Magic events have 16 rounds and require 11-5 (68% winrate) to requalify, which seems more reasonable.

That's all I got for now! I'm sure there will be comparisons to be made when more changes are done so I'll be sure to revisit this subject later.

Saturday, July 6, 2019

Seoul Masters Tour Qualifiers Post-Mortem

It's pretty surprising that open cups made a return this year after being almost universally disliked in the past and completely irrelevant to competitive Hearthstone last year. I used to be a solid cup grinder in 2016-2017 and was getting most of my precious HCT points that way. Back then I preferred cups to ladder since it was better preparation for the highest stakes Hearthstone I could play: the Prelims / Champs / Worlds trilogy. I actually had a tough time adapting to grinding ladder last year and that cost me very valuable points in the beginning of the year (I would have made it into GM easily with about 10-20 more points, which is 1-2 top 25 finishes). 

In a weird way it actually makes more sense to use Specialist Bo3 tournaments to qualify for Masters Tour since that is the actual format we compete in afterwards. Unfortunately, it comes with a significant drawback. While I could easily qualify for Playoffs every time through accumulating HCT points, consistency is now a thing of the past. Since I won a lot of open cups back then, I never imagined that I would not manage to qualify for Seoul, but it is indeed what happened (sorry for spoiling it so early). One reason I identified is that the player cap is much higher now: instead of mostly 128 player tournaments, we now have 224 and 256 player events that are always full, and therefore much harder to win. The other main reason is that the better player is a lot less favored in best of 3 than in best of 5. Battling through hoping to dodge bad matchups, bad luck and other good players seemed hopeless.

I started pretty slowly. I initially hoped to get good practice for Vegas so I let some qualifiers go by waiting for buffs to be live. Then, I missed out on some events as I failed to register in advance as they popped up on Battlefy. We seem to be going with a patented first come first serve system when it comes to registration in these qualifiers, and while I completely disagree with this situation, it's a battle I will fight another day. I became an expert Battlefy camper and registered for all events as they became available, whether I planned to play in them or not, since this is what Blizzard implied I should do if I wanted to play. My preferred solution was to complain on twitter and get them to change the system but they were abundantly clear that it was not an option. 

I got 2 top 8's in a row right before Vegas and was feeling good about my preparation. It didn't make a lot of sense to expend tons of energy grinding qualifiers since those efforts would get invalidated if I did well in Vegas, which would qualify me for both Seoul and Bucharest. I therefore prioritized being well rested and working on my lineup. It soon became apparent that after missing out on 2 week-ends of qualifiers due to travel it would be almost impossible to make up for lost time and get those missing 4 top 8's, but I was determined not to miss out and tried my best anyway. I ended up getting 2 more, losing every time in quarters or semis. 

The last week-ends of qualifiers, I was getting desperate and decided to play in all the events I had time for. Unfortunately, there is significant overlap between events so it is difficult to play a lot of them without sacrificing sleep. It also means frequently dropping at 2-1 or 3-1 to start another event was the correct play, since it is more likely to go 5-0 or 7-1 from the start than it is to win the last 4-5 rounds of a swiss event. I started with the 9 PM on Asia, then the Midnight Special, then skipped the 4 AM on Europe so I could get 2-4 hours of sleep and wake up in time for the 8 AM on Europe, the NA Noon and finally the 4 PM on NA. Do that for 3 days every weekend, from Thursday night to Sunday, and I believe if you are a good enough player you can get the 6 top 8's required through sheer force of will. Is that worth it though, when the reward is to pay your own travel to a 300 player event with a 500k prize pool? 

The course of action that makes the most sense for most players is likely to play a qualifier every now and then and hope to highroll one. The system is pretty rewarding to players who happen to qualify on one of their first few tries, and to extreme outliers who manage to get the required 6 top 8's. However, for players in the middle, who play a lot, win a lot, and end up with 3-5 top 8's through a consistent 60-70% winrate, well, it is extremely frustrating to say the least. I have a difficult time imagining that many players will grind, fail to qualify through no fault of their own, and keep investing most of their weekends for 2 months again.

Let's conclude with a few numbers that illustrate just how difficult it is to win these qualifiers (for more statistics and analysis check out MegaManMusic's twitter).

Odds to top 8 and top 2 (threshold for qualifying) a single elimination qualifier:

Win rate    Top 8       Qualify
50%          3.13 %      0.78%
55%          5.03 %      1.52%     
60%          7.78 %      2.80%      
65%          11.6 %      4.90%      
70%          16.8 %      8.23%        

This doesn't take into account that win rate will naturally drop as opponents are tougher on average in the later rounds, so odds to qualify are even lower than shown in reality. Should it be this unlikely to qualify for the very best players with 65-70% winrate? I definitely will carefully consider the opportunity cost when deciding to play in qualifiers for Bucharest and hope for a better system next year...