Slingshot’s Christian Shepherd had a chance to talk with GrinningGoats’ Haiwei “ADWCTA” Wang, who is one of the two people behind the success of The Lightforge Arena Tier List as well as HearthArena. Having parted ways with HearthArena, ADWCTA talked about the model he has developed for the tier list, his future plans for their contributions to the Hearthstone Arena community, and his long-standing relationship with childhood friend and partner Merps.
Christian Shepherd: Thanks for taking the time ADWCTA. First and foremost, how has the launch been since the expansion came out?
Haiwei “ADWCTA” Wang: The website launch was great, and we have an announcement (Wednesday night). Then, we probably are not going to have anything big come out for a while after that. We just wanted to get this up so there is something to reference other than a Google spreadsheet. We had all (Whispers of the Old Gods cards) up before Blizzard did. We always have everything updated.
Some people think that you can’t predict the value of cards and things like that before you actually play with the cards, and that’s true to the more micro extent of how good the cards are. But in the big picture, it’s the Arena, it’s not constructed, so metas only shift and card values are rarely based on combos, and they’re not based on the kinds of decks you’re facing, for the most part. By ball-parking it, you can get within half a tier without ever playing the cards, as long as you know kind of what cards are in the card pool.
CS: How accurate have you been in the past with your predictions? How did this expansion compare to previous ones?
HW: We have been pretty accurate with it historically. We get more accurate every expansion that we do this because we get more experience. Whispers of the Old Gods, for example, we keep looking at it and thinking about which cards we should move and we are coming up with blanks here. I think we nailed this one at definitely more than 90% accuracy for commons and rares. Obviously, we haven’t played with a lot of the epics and definitely not a lot of the legendaries yet, but at least for commons and rares I am having trouble finding anything that we would move more than four points.
CS: How do the C’Thun synergies affect this iteration of the Arena environment?
HW: Any card with the word “C’Thun” in it is not in the Arena. All of those cards would actually be really good cards if they were, but they’re not. I actually like that decision by Blizzard. I was a little on the fence about it before, but now that you can see and play with the whole set I like the decision. The set is boring enough as it is. It is a lot of vanilla cards and the same mechanics, a lot of stabilizing factors that I think is more geared towards ranked with the Standard format coming out than it was ever geared towards Arena.
The C’Thun text never matters in the Arena right? Then Arena would focus far more on vanilla-style trading and it had already gone far enough in that direction that not having the C’Thun cards in the Arena was not a bad thing. At the end of the day, I think Blizzard came out with the right idea. Generally you want more of those kind of cards to help stabilize the Arena, but this whole set is a stabilizing set so it wasn’t really necessary.
CS: Can you tell us a little bit about how often you will be seeing new cards from the set? I know that Blizzard has done it a little differently this time around. What are your thoughts on that?
HW: Well, there is an offering bonus for new cards in the Arena at 150 percent. Let’s say that you are a Druid and you have a C’Thun common card and two other common cards from the set. If the two other common cards were to get a 50 percent bonus, that would be normal, but now there is no C’Thun card in Arena, so what Ben Brode has said is that the offering rates that would have gone to that third C’Thun card is now going to be split between the other two existing Whispers of the Old Gods Druid common cards. So now, each of those common cards would have double the offering rate, as compared to a regular class common.
So you’ll see something like Feral Rage literally twice as often as something like Swipe. If you do that with a rare card, that rare card becomes offered as often, if not potentially more so, than a common class card. For Warrior, there is Bloodsail Cultist that also buffs your weapon +1/+1 if you have another pirate on the board. That card is seen almost as often as any Warrior common class card even though it is a rare.
CS: That’s a pretty significant change to the Arena environment. I did want to ask you a bit about how the tier list got started and how it became what it is today. Can you tell us a bit about that?
HW: We didn’t set out to do Hearthstone tier list or anything like that. I started writing articles and then some people would ask “OK, but how would you treat this game play decision?” It’s really hard to talk about these things in a vacuum. “Would you trade or would you go face?” I don’t know, it depends on the exact board state. It depends on what your opponent has been doing. It depends on what your hand is. It depends on what your deck is.
I do everything with my best friend Merps. We lived two blocks down from each other back in a Jersey suburb where we grew up, and we have known each other since middle school. We have been using video games as a way to keep in touch since high school ended since we haven’t lived in the same city since then.
We were both really good Arena players and so when we started streaming we just started doing it together as cooperative Arena so to speak so that we could discuss, teach, and debate and that worked out really well. We also got a really good record; all throughout Classic and The Curse of Naxxramas we were averaging nine wins per run, sometimes above ten wins per run and always above eight wins per run. We were just very consistently playing all classes evenly. No one else was doing that. Not only was no one else doing that, but no one else was analyzing the Arena the way we do.
CS: Do you think that your formal education gives you any type of advantage when analyzing cards or developing the model?
HW: We both have doctorate degrees, and most gamers are very smart people, but they aren’t formally trained to think in this sort of analytical way and explain themselves in this way. They are as good of players, if not better than we are, but when it comes to being able to break down exactly what you are doing, much less being able to create an algorithm for it, that is something that I think we have very heavy advantages doing.
Up until then, both Trump and Ant1gravity were making tier lists. We were using them and commenting on them and talking about places where we disagreed, but overall we very much agreed with their analysis, and then they stopped. I knew that Arena wasn’t all that popular, but clearly these Arena lists are helpful to some players. We decided that we weren’t going to just fill the gap, we were going to make it better. We broke it out to specific values, and I created a bit simpler of a mathematical formula back then, but now it’s a whole model that is sitting on my computer that will get you values of cards at certain constants to be put in for opponent behavior that change with the meta and other things that I adjust by hand like offering rates and other cards entering the meta.
75% of the cards go through the model fine, but for the other 25 percent, they are just like “Nope, this is a unique card. It just does its own thing.” No mathematical modeling will get you a very useful number for these cases. It will get you a number, but you would change that number so much anyway that it is kind of pointless to look at it.
CS: Give me an example of a card that just completely rejected the model you had in place and required a personal touch on the number.
HW: Forbidden everything. I think that those cards are some of the best designed cards that they have ever released and are insanely powerful.
Before we would value cards with numbers we would just describe cards. I would always say, “Yeah this card is just ridiculously amazing, ridiculously amazing,” and then we sat down to discuss the numbers. As a general part of my dynamic with Merps when we are evaluating cards, Merps hates on cards that he hasn’t played with or doesn’t know that well compared to me; I get more excited about cards. He kind of keeps me in check and makes sure that I don’t rate cards too high. For the Forbidden cards I was just like OK, I am going to try to get Merps to go as high as possible, so I started at a 150, which there is no card that high, Dr. Boom is the highest card and he is a 130. He was definitely much lower on these cards that he hasn’t played with, he even ended up settling on 120 for Forbidden Ancient and 110 for Forbidden Shaping.
CS: Did those values end up sticking after launch?
HW: Well, I didn’t actually want them to be that high, but I was trolling so hard that we just ended up publishing them at those values until I was on stream and Forbidden Shaping came up. In my opinion, there is no consequence for picking a card over 100, you are always going to pick it and it’s not like there will ever be another card as powerful to choose from, only this time there was – Cabal Shadow Priest, which is better than Forbidden Shaping. Immediately after that run where I was forced to take Forbidden Shaping because of how I had rated it I went back and changed it to where it should have been, around 100 and 90, not 120 and 110.
CS: What is it about these cards that make them so powerful?
HW: These are cards that are never good when you play them, but they prevent you from having a bad turn. It’s definitely a card that is better with the better of a player you are because it provides stability.
CS: Can you tell me a little bit about the factors that go into deciding card value?
HW: Everything, really. That’s the point of our model. It has to cover the whole game of Hearthstone. I actually had to add a part of the model to cover a specific card, a forbidden card, because those cards are weird. It was Forbidden Ritual and I had to change the model to cover it because we never had an issue where we considered potential space on the board to be a scarce resource. Forbidden Ritual creates 1/1s for each mana you use, but you can only generate seven at max, so that caps it compared to the other forbidden cards. The reason that forbidden cards are so good is because you can play them whenever and get something instead of getting nothing. This card doesn’t in many situations. Even this is now built into the model.
Everything is accounted for: reach, taunts, stat distribution, which turn the player is on, the model accounts for every single turn, spell damage, tempo, card advantage, healing as being different from armor…it’s this hugely complicated thing. It isn’t as complicated as an entire algorithm like what we built at HearthArena, but it goes deeper for each individual card. When we left HearthArena, we were able to take this with us because the Tier List always remained separate from the algorithm. They actually don’t have this model at all. Now we get to use it on our Tier List and come up with much more accurate numbers, which is a great advantage. And, now that we have our eyes set on a much more accurate algorithm, we’re developing from the Tier List model as a base and looking to incorporate all this complexity into the new algorithm we hope to release over the winter.
CS: Oh, so there are plans to make a website similar to HearthArena or ArenaDrafts?
HW: Oh yeah, we’re looking to launch with a deeper algorithm, better interface, and more (or at least, some different) features out of the gate. Basically, the idea is to not merely make HearthArena again. That seems pointless to us. Their algorithm will degrade over the course of new cards and new metas without us at the helm and while it would be helpful to the community to remake it and have us pilot it again, it would feel too much like settling for something old, something designed without as much ambition as it’s already realized. HearthArena’s algorithm started out much simpler, so there are legacy systems that form its core and can’t be easily changed. It’s already blocked several improvements we tried to make to it. So, this is our opportunity to start from the ground up and build an algorithm with some better bones.
CS: How are you planning on doing things differently this time around? What are some of your ambitions for the new site?
HW: We get to have a much more ambitious mindset and make the algorithm cover Hearthstone the game, not just the limited “these things are useful in Arena” and that’s it.
Our system will be able to be very flexible in that regard because we are not approaching it from a purely Arena perspective. We are approaching it from a ‘soul of the card’ perspective. What really is reach? How does it really affect everything? How does it play with card advantage? What is a mana cost really? A five-mana card is a card that you can play for five-mana on turn five, but that means it is a dead card before that. How does that affect its value? How does it affect its value in certain deck archetypes? How does having two five mana cards differ from having one five mana cards in terms of piecing cards together for mana efficiency? HearthArena doesn’t do this type of analysis and if it does, it does it on an extremely basic level. The idea is to just blow everything else out of the water when it does come out, because it will be created with such detail and such love and care that anyone who knows what they are doing in Arena will use it and immediately feel its power.
CS: I know that you used to write articles for Hearthstone. Do you still write any content? Can you tell us a bit about the content you wrote in the past?
HW: My first article series was the Mastery of Arena series. It kind of goes back to breaking everything down to the foundational principles of Hearthstone. What is card advantage? What is tempo? How do these things work? How do you add reach to it? When do you trade? When do you hit? What are the consequences of these things? Why are one mana cards like Humility the way they are? When are they useful? Just foundational principles of drafting and game play. I have developed these principles further in my head, but I haven’t published them. They don’t really change all that much though. I still stand by 90% of what I said in those articles. They are still useful, but you will see really outdated references to Yeti and Raptor.
After that I did the Killing series, which is how I look at how you actually play the game. You are always facing an opponent. Your opponent is always one class. All Rogues are one deck to me. I kind of adjust them in my head according to how they are playing and what archetypes they are fitting into. More important, because of the built-in weaknesses and strengths in each class, you play against each class in a very different manner.
CS: Do you remember any particular class that you were effective against just by playing into their intrinsic weakness?
HW: With Paladin, I used to have something like an 82 percent win-rate against even though it was one of the best classes. That was because I focused exactly on what the Paladin could not do and exploited it like crazy. They had no hard removals or nothing that dealt three damage that I cared about, because Hammer of Wrath is so slow. They couldn’t deal two damage to one thing and they couldn’t ping. You can almost always set up your board in a way that the Paladin can’t deal with because that is way too many weaknesses. Many people were out there losing to Paladin because they weren’t playing with a formalized view of how to deal with the class. Now people are much better about doing this. The skill level in the Arena has come a long way since classic.
How to play against Flamestrike was also always a popular topic. That’s one of my triggers, people talk about it a lot less now, but back in Classic people would always say “Flamestrike is overpowered. It breaks the game. Mage is so easy to use.” Mage was never easy to use. As far as how much better you can get with a class based on your skill level, Mage is one of the highest caps. People kept bringing it down because they didn’t know how to play against Flamestrike in most situations. Sometimes the best way to play around Flamestrike is so play directly into Flamestrike, most times it is not, but I tried to explain that as well as I could.
CS: Do you ever take that game knowledge into Constructed? Any Legend grinds or anything like that in between Arena runs?
HW: Well, we grind out to Rank 20 every season. It’s pretty tough sometimes when the meta is so unstable in ranks 23 to 20 and you lose to something like basic Mage. You are kind of just like, what am I doing? How did I just lose to this guy playing a basic Mage deck?
CS: I think everyone has had that experience when the ladder resets. It happens to the best of us.
HW: Yeah, exactly. I did play Constructed back in May of 2014. I created my own deck. It was a Pirate Warrior weapon deck and kept shifting until it fit the meta. Back then the meta was like 40% Miracle Rogue, which was very helpful to me. I had an aggressive deck that destroyed Miracle Rogue before they can start miracling, plus my aggression was not on the board so they could not remove my aggression, namely weapons, then they can’t win. It was such a weird deck. I ended up getting to Legend in less than 200 games, which is a very high win rate. Trump ended up using the deck and giving it a cool name and took it to Legend himself and then streamed with it in a tournament. I was like “Oh cool, Trump is using my deck. That’s awesome.” Especially because it was Trump, one of the top Arena players at the time, and I knew all about him. After I had done it, I was kind of done with Constructed. I had proven to myself that I could create a home brewed deck and also knew that I preferred playing Arena. I accomplished my goals and went back to Arena and never really looked back.
CS: Coming back to Arena, what do you think are the most important factors to consider during an Arena run or draft?
HW: Tempo has always been the most important factor. That was one of the reason we were able to average more than nine wins per run. You can win off of the value game at a very high win rate. You had people like Trump who played a very high value game that had strong arguments. The thing about playing the value game is that value is much more resistant to swings. There weren’t all that many reach cards back in the day. So it was a much more stable game environment. We had always been much more on the side of tempo. One of the first articles I wrote was how Trump had been mis-valuing those one mana cards like Hand of Protection and Blessing of Might that were pure tempo cards. You will never gain value with those kinds of cards. You will never go two for one with those cards, at most it will usually be one for one, which was awful for Trumps play style.
I was advocating that you want one or two of these cards in your deck. It will make your deck infinitely better because it will increase your ability to control the board by so much that you will get that card value back later.
CS: How do you approach synergies in the tier list model?
HW: Well, the tier list is a baseline that you have to adjust yourself throughout your draft. Synergies are the most visible thing that websites like HearthArena look like they are doing, when it really only affects the deck value by 15%. If you are a heavy synergy drafter, you will never be an infinite player. The stats will never work out that way. If they come, you shouldn’t ignore them, especially if they are already in your deck, because at that point they are sometimes just better.
CS: Thanks for taking the time! Anything else you want to say about the new projects you are working on or the tier list?
HW: I hope people start using this website. Our goal really, is for it to be the No. 1 Arena Tier List that people consult when they’re debating values. We try to make it fun for the newer players who are getting into Arena or constructed players getting back into Arena. That’s why we have all these viewing and sorting formulas so you can customize it to the experience you want. We also take care to respond to the community preference very quickly, adding a minimalist mode for people with slower computers and for people who just wanted the info presented as simply as possible. The Tier List is for the community, so their preferences trumps ours every time. We are also going to add a search function that goes well beyond just looking for a specific card, and hopefully that’ll be helpful. The Tier List is available in every language that you can run Hearthstone in, including Thai, which Blizzard has just expanded into. Ultimately, we want our site to be used as the standard for discussion about the game. We want to kind of connect the global Arena community and break down this language barrier. There are so many communities of Arena players in the world, but they are all so separated. This is one of our big goals. When we launch the full website in winter complete with drafting algorithms, we are hoping to have it available to every language that Blizzard supports as well. It’s important to us that we reach out to everyone.