Meta Polarity and its Impact on Hearthstone

The Boomsday Project has ushered in a meta that is quite similar to the Witchwood meta after the balance changes. This meta is relatively balanced in terms of power levels. No deck seems to be too powerful, and both ladder and tournament formats are filled with different and diverse strategies. The meta is so “healthy” that Team 5 have decided to skip balance changes this expansion, as they were deemed to be unnecessary.

And yet, despite the current state of the Hearthstone meta, many players don’t find Boomsday to be enjoyable. We can see signs of possible loss of interest through different metrics (Streamers Viewership, Google Trends search patterns, and games tracked in Data Reaper), though we cannot claim with 100% certainty that there is a decline in the player population.

So, what’s wrong? Obviously, there are issues close to the community’s heart that have been neglected, the primary one being the lack of new features to freshen up the game. We’re not here to discuss that. We’re here because despite looking healthy on the surface, the meta does have a glaring problem that cannot be fully grasped by deck win rates and play rates alone.

You may have heard common complaints about matchups feeling too polarizing in the current meta, and the sense of hopelessness when queuing into the wrong opponent. The Rock/Paper/Scissors sentiment was brought up numerous times throughout Hearthstone’s history, but does Boomsday truly stand out from other meta’s?

To answer this question, we develop a metric to measure meta polarity. Our goal is to create a metric that accomplished two things:

1. Provided numbers that could be easily compared across different metagames.

2. Was simple enough to understand and immediately apply context to.

Meta Polarity Metric

Let’s say we have a 60-40 matchup between deck A and deck B. The matchup differential between these decks is 20%. To obtain an aggregate measure, we compute the matchup differentials of every matchup in any given meta and then calculate the weighted average of their absolute values based on the popularity of each matchup. This is the Meta Polarity. A meta polarity of 16%, for example, means that the average matchup in the meta is 58-42.

We also calculated an alternative measure of Meta Polarity, by taking the square roots of the matchup differentials before averaging them out. This method leads to nearly identical conclusions compared to the first method. Throughout this article we report results based on the first method, since it’s simpler to understand and apply, even though statistical pedants may prefer the second.

 

We examined every metagame since the Data Reaper launched and calculated the Meta Polarity for each significant period, taking both expansions and balance changes into account.

These are the results for legend ranks, where the meta is closest to being optimized. Differences across rank brackets are marginal.

Whispers of the Old Gods had a Meta Polarity of 10%, meaning the average matchup in that meta was 55-45. As new expansions were released, the Meta Polarity value rose until the balance changes of Mean Streets of Gadgetzan that saw notable nerfs to Small-Time Buccaneer and Spirit Claws.

During Un’Goro, Meta Polarity jumped to an unprecedented level at 15%. Un’Goro was notorious amongst the player base to be polarizing due to the dominance of Quest Rogue and Freeze Mage, but balance changes that rid the meta of Quest Rogue did little to change things, as Jade Druid rose in popularity instead.

KFT saw a reduction in Meta Polarity as Druid, Rogue and Priest established themselves as the top 3 classes after the balance changes. All had well-rounded decks with fairly balanced matchup spreads (especially Tempo Rogue and Raza Priest). The balance changes in K&C saw an end to the reign of Raza Priest and Tempo Rogue, and an escalation in Meta Polarity.

And then came the Year of the Raven with the Witchwood expansion, and Meta Polarity skyrocketed to a new record. Oh… Remember Quest Rogue? That’s when it came back. However, it’s easy to blame Quest Rogue for the Meta Polarity, but it’s clearly not the only deck at fault. Even when looking at the meta after the balance changes, when Quest Rogue disappeared again, Meta Polarity was still higher than in any period before Witchwood.

Fast forward to today, when during Boomsday meta polarity hit yet another record, making Boomsday the most polarizing meta in the history of the game. In a three year period, we’ve gone from an average matchup of 55-45, to the average matchup measuring at almost 60-40.

You haven’t been imagining things: the game has become the most polarizing it’s ever been.

So which decks are “at fault”? To answer this question, we’ve calculated the Deck Polarity values, which are the weighted averages of each deck’s matchups based on their popularity. We will show you three snapshots of the meta at legend towards the end of key meta periods. Remember that a deck’s polarity is influenced by other decks in the meta, so polarizing decks tend to “snowball” polarity for everyone else too. We cannot directly compare a single deck’s polarity to another deck in a different meta as a result. We can only directly compare them to other decks in the same meta.

Let’s look at Whispers of the Old Gods, which was the least polarizing meta we’ve ever measured.

The two most dominant decks in the game, Dragon Warrior and Token Druid, are two of the least polarizing decks with values that are well below 10%. The most polarizing decks are relatively fringe in comparison, with Yogg Hunter topping the chart at 16%.

Let’s fast forward a year, to Un’Goro before the balance changes:

Things significantly escalate with the launch of this expansion as well as the yearly rotation. There are now multiple common meta decks with Polarity values of over 20%. Quest Rogue is one of them, but it’s surprisingly not as polarizing as you’d expect it to be, and there are other decks that are equally polarizing: Taunt Warrior, Jade Druid, Freeze Mage and Aggro-Token Druid. All of them carry high play rates and high Meta Polarity values.

Let’s fast forward another year, to another expansion and yearly rotation which blew up the Meta Polarity to new records: Witchwood.

Quest Rogue made a huge comeback in Witchwood, but this is a different deck compared to its Un’Goro iteration. It has become so absurdly polarizing, unlike anything we’ve ever seen before. Remember that Witchwood Quest Rogue now had Sonya, Zola, Valeera the Hollow and Vicious Scalehide. The potential of infinite value and infinite healing made it even more ridiculous against late game strategies and more powerful overall.

But the list shows other decks that are also quite extreme. So much so that Witchwood was still the most polarizing meta to date despite its most popular decks, Cube Warlock and Even Paladin, being the two least polarizing relative to the field. Remember that even after the balance changes, Witchwood still carried an extremely high Meta Polarity value.

And let’s finish things off with Boomsday:

There’s a reason why we’ve said in the last Data Reaper Report that Quest Rogue and Odd Warrior are a design problem. Odd Warrior is an even more polarizing deck than Quest Rogue in the current meta, and both hover around the absurd value of 30%. A long list of decks carry polarity values of well over 20%, but there are relatively few decks that truly escape these heights and offer a stable experience. Playing Token Druid in the current meta will still feel more polarizing than playing almost any deck during Whispers of the Old Gods.

Rock/Paper/Scissors Hearthstone has never been worse, and we’re left with one final question:

How and why has this happened? We have a few areas to point the finger at, but considering the extreme jump we’ve observed in Witchwood, it’s clear that the biggest fault lies in the sets from the last year and a half, beginning in Journey to Un’Goro. All 4 sets before Boomsday introduced mechanics that enabled and greatly escalated the core issue behind polarization: lack of meaningful counterplay. When you think about it, it all makes sense.

Quests in Un’Goro

There has not been a single quest deck that wasn’t extremely polarizing. Quest Rogue is most famous for it due to its dominance and prevalence in different periods despite being hit with multiple nerfs. However, Warrior/Mage/Druid/Priest all emerged with quest decks at different points in Hearthstone’s history and they were all extremely polarizing as well. The inevitability that some of these quests enabled often offered little counterplay beyond “killing them before they complete the quest”.

Death Knights in KFT

Death Knights were a massive power creep on late game cards, and many of them enabled the potential for infinite value. They simply outclassed and continue to outclass other late game cards to this day. The ability to produce infinite value or infinite damage is often one of the key contributors to polarization. It effectively eliminated resource management as an important avenue for counterplay.

Mana Cheating in K&C

How often have you lost the game to an Oakheart without being able to pretty much do anything meaningful about it? The mana cheating and stat cheating introduced in K&C led to the creation of some of the most busted interactions in the game, leading to many of these cards being nerfed. Some of them are still around and remain very impactful in decks that are very polarizing in nature.

Genn & Baku in Witchwood

We’ve already said in the past that both Genn and Baku introduce a more extreme and polarizing game experience. Genn enables consistent decks, but its hero power upgrade is tempo focused, so it has actually enabled decks that are not very extreme in their matchup spreads.

Baku is a different story. Upgraded hero powers from turn 1 have led to the creation of very polarizing decks, since much like Death Knights, they enable the potential for infinite value, but more importantly, extreme consistency in generating that value. Baku decks often hard counter each other because they’re entirely built around their hero power. They are feast or famine decks, and that leads to a polarizing experience. A good example is Odd Warrior: Its life gain is either oppressively powerful or entirely meaningless in the face of infinite value or infinite damage.

Final Thoughts

We urge the Hearthstone team to learn from this polarizing experience and tone down the elements that enable it. Infinite value, infinite life and infinite damage are dangerous. Mechanics that nullify the importance of card advantage or life as a resource often become degenerate, polarizing and lacking counterplay. It started with Jade Idol and escalated with Death Knights and Baku. Players simply hate it, and when brute force counterplay is introduced in response to it (Skulking Geist), it creates an experience that is just as frustrating and polarizing.

Polarization creates a poor game experience since players do not feel in control of the outcome of the game. Their decisions matter less, and when faced with a game in which you have little control over the outcome, it leads to a frustrating experience. This isn’t just true for a card game, it’s true for every game: Any MOBA or WoW PvP player can attest to the frustration of stun mechanics being too powerful and preventing control of your character. Queuing into 80-20 matchups is a similar experience of lacking that control. We want to be able to actually play the game and feel like most of our decisions matter.

AN IMPORTANT REMINDER

Without the community’s contribution of data through either Track-o-Bot or Hearthstone Deck Tracker, articles such as these would not be possible. Contributing data is very easy and takes a few simple steps, after which no other action is required. If you enjoy our content, and would like to make sure it remains consistent and free – Sign Up!

Thank you, and see you in the next Data Reaper Report.

24 Comments

  1. The most fun in HS for me is building or teching decks, always aiming to counter the meta. (It’s not an effective strategy at all right know).

    But one thing I wished I could do in the past, perhaps could help right know. I have been so often down to the point where I said to myself:

    If I only could ban one class on ladder.

    Please discuss about one class ban on ladder. It could be a good solution for the game experiencie besides balancing and nerfs! Just think about it. Odd warrior loosing relevance, getting banned by aggressive decks. No late game matchups for Quest Rogue anymore. Priest getting a good spot in the meta, banning druids etc. Etc.

    @Blizzard
    1. Get rid of Power Creep like Quests and Baku
    2. Add reliable counterplay (in advance!!!) for prevalent strategies
    3. Allow one class ban on ladder
    4. Players experience an interactive game, where decisions matter, decicions in the deck builder and in the game
    5. Profit

  2. Great content!
    Finally there’s statistical proof about we’re not getting insane. I observed myself in the past year+ getting increasingly more agitated while queueing about “what will I queue into” regardless of the deck I play.

    Token druid is the calmest deck I’ve found since WW, not feeling oppressive or oppressed in the MU. Math proves my feelings to it as it’s the lowest score from the decks I’ve played. While we could argue about how precise the numbers are given the effect of mirror matches on the numbers, it reinforces anecdotal evidence and makes sense.

    Keep up the awesome work guys!
    (ps: if it’s not too cheeky to suggest, if you’re anyway in the business of proving “what’s wrong with HS development”. I’d love to see a statistical analysis about highroll effect in mulligan+first few draws. I think it’s important as it generates a high casino effect in some decks and also skews winrates closer to 50% in any statistic because it either happens or not. In the end the game outcome for such decks (I’m looking at you zoo, taunt druid etc.) is more influenced by RNG than real decisions.)
    Cheers

  3. Very good article!

    In my opinion, the polarity of matchups is near the core of hearthstones problems, but it is not the core itself. However, a too big polarity isn’t good, such as a too small polarity. (Perhapst allready the meta polarity of karazhan was too small, if there was no good counter for shaman.)

    I think the core of the problem is the lack of interaction. The quests and the herocards brought more polarity, but in a bad way of less interaction. You can’t play arround the quest from an opponent, because he plays it only for himself. You can just take a deck to counter it, and so the polarity rises. It’s the same with the death knight cards, you cannot do anything against it in the running game. So the lack of interaction led to the rise of polarity. If you can not counter single cards, you must counter the whole deck.

    Hearthstone needs more interaction. But not only in form of counter-cards like the crabs – they are also important, but they bring more polarity, too. Secrets are a very good example for interaction (maybe the best). Every secret can be very strong, but also very weak – primary dependend on the skill, not the ability to counter the deck. I wish there would be more secrets for more classes. The rush-keyword was the right way from blizzard, but with the actual pool of standard cards it is too weak. Maybe this will change in the future. Project-cards can also bring more interaction, although they can bring more polarity. Another idea are new quest like “the player who plays 5 deathrattle-minion more than the opponent gets …”, so you can activ delay the progress of the opponent.

  4. I’ve always wondered why Blizzard opted for “Once a year we kill three expansions, and slowly introduce 3 new ones over the new year” versus just rotating out the old expansion when the new expansion comes out after they hit the end of the two year window.

    It seems like losing one expansion, and adding a new one every 4 months would shake up the meta more than just having it happen 1 time a year.

    Although this has nothing to do with the mechanics that are causing issues with polarization.

  5. I think you forgot another important point about why the current meta feels boring despite being balanced: BDP is the new TGT. The mechanics it introduced didn’t catch up. Magnetic fell flat on its face, and most legendary spells are shit.

    Sure, BDP brought us Auto-include Inventor, a playable version of Odd Warrior, made Zoolock better, but that’s it. BDP didn’t shake the meta like previous expansions did. Shaman and Paladin suffer from this the most. They’re playable, but players aren’t interested by them because they’re essentially stuck in Witchwood. Not to mention the fact that there won’t be any mid-expansion nerfs to artificially shake up the meta (the first time in 2 years). As of now, BDP is useless (it might change later though, just as how Totem Golem and Alexstrasza’s Champion became broken during the Year of the Kraken).

    If you’re F2P, craft double Giggling Inventor, Zilliax, Floop or the Zoo shell if you like those classes, then don’t bother and start saving dust for the next (hopefully more interesting) expansion.

  6. The funny thing is, they tried to introduce counterplay in the form of rush, which is a great way for players to interact with each other’s boards without enabling the uninteractive go-face strat that sometimes puts aggro over the top. The problem was, most rush cards were weak and decks are no longer worried about board state as they can just play control until their few cards that enable infinite value kick in. This article is great, and I think it really gets to the heart of the issue with Hearthstone today

  7. HS needs more hand disruption, discard, targeted counterspells and more interaction with the opponent. The “I let you play alone” mechanic creates this issues where you can play Druid and win doing nothing but ramping and drawing letting the opponent with no responses than concede. That will force the decks to have more than 1 win condition and limit some decks reliability.

  8. but how much that polarization is tied to the diversification?

    i mean, i suspect that in a low differential meta this is translated in nothing more than only jack of all trades decks are viable. so basically all decks are more or less the same and in this way only the most refinied of them will survive. as an example take keleseth decks, why build other of them when rogue is simply the best of them?

    now, when the differential are so big, this means that each deck could fill relative niches having at the same time respective weakness.

    could be infered all this from the datas?

    • My thoughts is that the diversity of decks don’t matter for polarization, just the kinds of win Condition they have and how interactive they are.

      If druid has 5 different decks but they all need to fight for the board to win, then the game is fair.
      When they can only draw, get armor and kill you anyway?
      Well, then it’s a problem. “kill you really fast” is not counterplay

      Just look at which decks have low and witch have big polarization scores on the boomsday table.
      The most polarized decks have either insane control tools and tons of armor or some really unfair early blowout consistency.
      They rely on legendary weapons, odd decks, quests, etc.

      Most on the lower side of the table are board-control decks. Even zoo is there since the insane turn one plays are not consistent. They don’t rely on last year’s broken legendaries. Only even shaman, and they talk about how Baku is the problem not genn

  9. “Polarization creates a poor game experience since players do not feel in control of the outcome of the game. Their decisions matter less, and when faced with a game in which you have little control over the outcome,”

    This conclusion is totally unjustified. A player who makes better in game decisions than his opponent is going to win their favorable match ups more often than expected AND lose their unfavorable match ups less often then expected. Skill is just as relevant in a 70-30 as it is in a 50-50.

    It is “High roll” decks that lessen the relevance of skillful play. They are also inherently non-polarized since they can “high roll” to win normally unfavorable match ups and “low roll” to lose matches that they should win. Which suggests that we polarization is not even a bad thing.

  10. We also need to take into account how many more cards we have in standard.
    Now synergy becomes more important than ever, because you can pick your 30 cards from twice if not more the card pool we had during WOTOG
    This means that you get the 30 best cards from a much wider pool, so deck’s strengths also get potentialized. We would never dream of seeing some of the classic and common cards that right now are part of the meta. Even cards that got nerfed are back on the game. It’s partially due to the fact that (together with mana cheat) you can make bad cards work together with good ones to make insanely broken combos.
    Let’s not even count the lack of decent defensive tools for 1 and 2 mana, lack of early removal and cards like Vicious Fledgeling that just runs away with the game. There are too many cards that if not answered just win the match on turn 2 or 3.

    Blizzard imo needs to stop printing counters and start personally curating more their game/meta. Hate cards are not a solution, the problem is still there and gets eventually swept into wild.
    I mean, c’mon, why print a build-around card that if it’s ever good get’s destroyed by a hate card? It’s not a fun experience to anyone.
    It’s not exciting to get countered like that, it’s boring to win a matchup because of random tech, and nobody get’s to do the stuff their deck wants to do.

    Just lower the power of the cards and let people play with them. Having something so busted that if you don’t have an imediate counter you lose just shouldn’t be a design choice

  11. Thank you for writing such a great article. Your analysis gives really good insight.

    That being said, I think it should also be noted that the fact that Blizzard intentionally printed a lot of Win Condition (Quest, Death Knight, Baku, Mana Cheating) seems like a marketing decision (along with the decision to have 2 legendary per class).

    We all know the failure of TGT where most card is “balanced” but unexciting.
    WOTG is hyped due to Cthun and Yogg, but then it became normalized again. Deck played is mostly value based or tempo based.

    It is Unqoro that started the trend to create a very powerful card that forced player into creating a deck to support it. And get even worse in KFT due to death knight.

    The metagame shifted away from tempo-based deck or value-based deck into combo-based deck (not necessarily a true-combo deck like Cthun or Mechatun, but a deck revolves around a strong win condition).

    And with the nature of combo-deck, it will be highly polarizing.

    Its ironic though, because in the past, Blizzard is very careful with creating combo-based deck, and they nerf Leeroy and Force of Nature because it gives strong win condition that cannot be countered. And now they print a lot of hard to counter card.

  12. Joe: It’s normal that the most popular decks are the least polarizing.

    1. It’s hard for a polarizing deck to become popular. Polar decks have some very good and some very bad matchups. What this means is that if a polar deck becomes popular and people get tired of queuing into it there will be lots of hard counter decks people can run to push it to the fringes of the metagame.

    2. Popular decks will have lower polarization scores just by virtue of being popular. A deck that is very popular in a stable meta means that the decks it counters are unplayable – no one can play mill rogue in a Jade druid meta for example. Meanwhile it also means that everyone with poor or middling matchups is teched against the popular decks. This will tend to smooth out winrates, the 80-20 matchups won’t happen very often because if you’re playing a deck that is 20% against the most popular decks in the format you’re suicidal.

  13. How do you control for mirror matches of popular decks (like Cube Lock) in the polarization numbers? Are the mirror matches excluded from the calculations? If not, does that account for the relatively low polarization number of some of those decks?

    • If you queue into the mirror, then you are effectively playing a 50-50 matchup. Thus, the times you queue into the mirror should be included in the calculation and count as having a polarity of 0.

      • Right, but wouldn’t that depress the average polarization numbers for a deck? A deck that is very popular because it is strong against all other deck types will frequently face a mirror match. If I’m understanding the methodology here, that will reduce it’s overall polarization numbers because it’s by definition not polarized against its own archetype. If you excluded mirror matches from the calculations, as a primitive way to control for the popularity of the deck, what would the polarization numbers look like?

        (In other words, the numbers on some of the very popular decks seem low to me.)

    • Wonderful analysis and use of data! Nice job VS team!! Thank you for your contributions to the community. I would have liked to also see something about diversity of the field. One of the “benefits” of this current meta is that so many decks are viable–much more than metas in the past. Is that statistically true? And is it possible to BALANCE between meta polarity and meta diversity?

Leave a Reply

Your email address will not be published.