Snapshots of the Early Un’Goro Arena Meta

Snapshots of the Early Un’Goro Arena Meta

I like to think of the current Hearthstone Un’Goro meta as the “trigger meta,” in that I have become annoyed with a lot of cards. A lot of this has to do with the meta decks in ranked play as Pirate Warrior, Quest (Caverns) Rogue, and Quest (Exodia) Mage are simply frustrating to lose against. While these constructed decks perturb me often, the same can be said about the Arena meta.

We are in the teeth of the new rotation, meaning that this is the lowest card pool in the Arena pool all year. When the second and third expansions come out in 2017, the draftable and playable card pool will increase. Thus, we will see more variety in the draft pool, as well as the Discover and random effect (Burgle, Transform) pool. Combined with the increase in spell rate, we are seeing a lot of cards over and over again. Let’s take an overview of the most common Arena cards, in this early Arena meta.

Hearthstone Screenshot 04-16-17 22.29.30.png
This interaction took a while

Data source

I just pulled the data from hsreplay.net, about 12 pm EST on 4/18/17. The data pulls numbers from the last 14 days. I filtered out cards that are now rotated into wild, but may still be present in “grandfathered” Arena decks. I also filtered out Legendaries, as they are not really important to Arena.

Top neutrals

Neutrals are the glue of the Arena. But with a new card pool, boost to rares and epics, and decreased offering rate to Basic cards, what should we expect?

Frequency

hsreplay_top10neutral.PNG

9 of the 10 top neutral cards are from Un’Goro, with Bog Creeper being the sole holdout. It’s interesting that the two poisonous cards in Stubborn Gastropod and Giant Wasp have the lower played winrates, which goes to show that cards without intiative have lower played winrates. Really no surprises in this list, as they are all very good. It is obvious that Volanosaur is #1 despite not being the best card in this list, since everyone gets to draft a golden one.

 

Deck winrate

hsreplay_neutraltop10win.PNG

Here we have the top 10 neutrals in terms of deck winrate, and hey there’s Primordial Drake again. I think Bright-Eyed Scout may be an underlooked card, as it could be a late game play on Turn 9 for big tempo. Silithid Swarmer and Naga Corsair are on the list because they are good Rogue cards. Charged Devilsaur is also proving it’s worth as a great epic neutral.

Top class cards by frequency

Finally, I will look at all the top class cards in terms of frequency. These are useful in playing matchups against a particular class, to play around certain cards. I am only looking at cards that are in over 30% of decks for each. I picked 30% for no particular reason.

Druid

hsreplay_druid30pct.PNG

Tortollan Forager, despite having a dumb voiceline, will be seen in 50% of Druid decks. Just a very good card. Druids will have a bunch of removal spells from the boosted offering rate, though no hard removal. Moonglade Portal is in 33% of decks, and has the bad RNG aspect to help swing games. As expected, Druid is one of the worst Arena classes right now.

Hunter

hsreplay_hunter30pct.PNG

Hunters are seeing healthy winrates, and their top 3 neutral commons are from the new set. Plus, they all have solid deck winrates. Explosive Shot and Call of the Wild show up over 33% of the time, so come to expect those power cards on Turn 5 and 9. Play around Deadly Shot and Unleash.

Mage

hsreplay_mage30pct.PNG

When Flamestrike saw it’s offering rate cut in half, there was debate as to why Flamestrike and not Firelands Portal. We’ll never get the answer why, but Firelands Portal continues to be the menace of having high deck winrate and played winrate. The initiative of leaving something on board is too good, and the chance to get Leeroy and Doomguard are as high as ever…

Anyhow, you’ve got a lot of powerful stuff appearing for Mage. Important to play around Meteor with good positioning on minion placement. They’ve got some early game now as well, so just a lot of good Mage stuff.

Paladin

hsreplay_paladin30pct.PNG

Paladin dropped to the middle tiers it seems with Gadgetzan, but seems to have cards with higher deck winrates now. Spikeridged Steed is seen a ton, as it has the spell offering buff, and is just a very OP card. Dinosize is a card that I like a lot myself, despite initial impression, and is sporting a solid played winrate, as a finisher. Vinecleaver is also another sneaky card that looked bad to me at first, but has a huge played winrate. Paladin is back, and these cards are quite fun.

Priest

hsreplay_priest30pct.PNG

Priest has the biggest list in terms of diversity, and they are mostly spells. Potion of Madness is still appearing in a maddening 41% of decks, so continue playing around that card. Free From Amber is as I expected, overrated, as the guy you get is a neutral card with likely no battlecry effect. Nothing really new to report, but Priests are doing their thing with reactive spells and just a big variety of choices.

Rogue

hsreplay_rogue30pct

Vilespine Slayer, possibly the strongest card in Un’Goro, predictably has high winrates and is the autopick in the epic slot. Hallucination is amazing, but falls victim to RNG gets now and then. But Rogue just has a ton of hard removal, as almost every card in this list is just that. Still great for Arena.

Shaman

hsreplay_shaman30pct.PNG

With the loss of strong early game tempo minions, Shaman has become a reactive Arena class. Volcano, despite it’s horrible played winrate, will be seen in half of Arena Shaman drafts. A lot of other spells are present, with the Hot Spring Guardian being an okay card seen a bit. Shaman doesn’t seem to be in a great spot, but if you like flashy looking spells, Shaman could be fun.

Warlock

hsreplay_warlock30pct.PNG

Warlock took a step back after being top dog. While Abyssal Enforcer getting reduced is the sexy answer, it is most likely because Imp Gang Boss is gone. Warlocks still have hard removal and AoE options, with things that hurt the player. It is interesting that the power epic cards like DOOM! and Twister Nether aren’t being picked more often. Ravenous Pterrordax is showing up a bit, and could be snowbally like the neutral Pterrordax.

Warrior

hsreplay_warrior30pct.PNG

Bringing up the rear as always is Warrior. There’s a short list here, because Warriors likely have to hedge picks for weapons, which inexplicably don’t get the increased offering rate that spells do. Weapons are spells for weapon classes! Direhorn Hatchling isn’t a great Arena card, since it relies on getting the draw for value, but it is the default leader. Just as a fun exercise, Ornery Direhorn, the class common was played 51,000 times in the last 2 weeks. Meteor, the Mage 6-mana epic, a situational removal play, was played 260,000 times in the last 2 weeks. That can show you what state Warrior is in.

Forecasting Arena Shifts by Class

Forecasting Arena Shifts by Class

As mentioned in the past, I love looking at data, and trying to use it to my advantage when possible. Hearthstone Replay officially made their collected data publicly available yesterday, much to my delight. I didn’t even know that this was planned for a release, but I knew data was collected through Hearthstone Decktracker, a tool I have used for years myself.  There’s lots of data about win rates for constructed decks in Standard and Wild, with great UI to see specific deck changes by deck. There’s also separate Arena data, which is what I was interested in the most, and will explore here.

Arena data

Not surprisingly, the Arena data has information about every single draftable card, as you’ll get a different deck each draft. These are the variables measured for each card:

  • Included in % of decks – What % of drafts have this card.
  • Copies  – Average copies of the card in each draft.
  • Deck winrate – Win% when card is in draft.
  • Times played – Raw times played.
  • Played winrate – Win% when card is played in game.

Assumptions

While having big data is great, it could be worse having data and misinterpreting it. Misinterpreting data leads to faulty reasoning and arguments. Let’s try to make some assumptions of this data before we proceed.

  • Players who use HS Decktracker are better Arena players than the average HS player.
    • HS Decktracker provides a ton of information in the game, which will allow the user to gain an advantage in getting intel. One could assume this helps, and the players who use it are more likely use other methods, like drafting tools.
  • Deck winrate and played winrate are independent by class.
    • As a class struggles, the overall winrates will plummet for the class. Just by looking at rates by class, we can see that Warrior and Druid are struggling a lot in the NA server over the last 14 days. Grimy Gadgeteer and Headcrack have the same deck winrate at 49.8%. One is a lot better than the other in reality, as Rogues win a lot more than Warriors.
  • Played winrate has bias depending on game situation.
    • I noticed that some pretty good cards had lower played winrates than expected. Flamestrike sits at 53.3%, the same winrate as Assassin’s Blade. Think about it this way, if you’re playing Flamestrike, chances are you are behind on the board. Mage players who choose not to, or don’t have to play Flamestrike likely have won already.
    • The same goes with card advantage cards. I often play card draw in a last ditch effort to draw into an out. This would skew card advantage cards to lower played winrates.
    • Pyroblast has a very high played winrate at 73.4%, but a 57% deck winrate. This signifies people playing Pyroblast to achieve lethal.

Methodology

  • I’m going to look at the top 10 class cards for each class in deck winrate. It has been established that deck winrate is likely better at evaluating a card than played winrate.
  • I will only look at commons, rares, and epics. Legendaries are omitted as they show up too infrequently. Arena 7.1 put epics back on the map with higher offering rates.
  • I will omit undraftable cards included from old drafts.
  • I will seek to identify the cards that will be rotating out in the Arena once Journey to Un’Goro releases.
  • These data were collected on 4/3/17, approximately around 4pm EST.

Druid

hsreplay_druiddeckwin10

  • Druid appears to be taking a huge blow to their Arena kit when TGT drops out. 7 cards in the top 10 will be leaving.
  • Without Mulch or any replacements in Un’Goro, Naturalize becomes the only Druid hard removal. While it becomes a better pick, it still isn’t great.
  • Shellshifter and Verdant Longneck are solid cards, but it isn’t enough help given what is dropping out.
  • Prediction: Druid might become one of the worst Arena classes.

Hunter

hsreplay_hunterdeckwin10.PNG

  • Things are looking up for Hunter lately thanks to Arena 7.1. Also, none of the top 10 cards are dropping out.
  • Houndmaster is looking even better in the Beast meta.
  • Hunter gets some amazing early game cards and a seemingly premium early removal with Grievous Bite.
  • Prediction: Hunter gets more tools and isn’t losing much. Beasts will help the class even more.

Mage

hsreplay_magedeckwin10.PNG

  • Faceless Summoner and Forgotten Torch are undraftable cards, so ignore them here.
  • Mage will lose 2 cards in the top 10, Fallen Hero and Ethereal Conjurer. All the powerful spells are still here.
  • Primordial Glyph does the samething as Ethereal Conjurer albeit without the threat on board. A bunch of the other common cards are quite solid as well.
  • Prediction: Mage continues to be an Arena powerhouse. Elementals help bolster neutral minion picks.

Paladin

hsreplay_paladindeckwin10.PNG

  • Paladin loses Keeper of Uldaman, Argent Lance, and Seal of Champtions. These are 3 premium cards, but 3 out of 10 isn’t the worst.
  • Spikeridged Steed and Lost in the Jungle are solid gains.
  • Paladin also gainst other snowbally threats that are more win more.
  • Prediction: Paladin either stays the same or gets a little worse. In any case, it should remain in the middle of the pack, possibly still one of the better picks.

Priest

hsreplay_priestdeckwin10.PNG

  • Priest loses nothing from their top 10 list in the upcoming rotation.
  • The loss of Dragons from rotated set would peg a card like Drakonid Operative a little. 5-mana 5/6 is still great to get though. Dragonfire Potion is still a board wipe to pick.
  • Priests get some Elemental synergy with Radiant Elemental and Crystalline Oracle. Shellraiser and Mirage Caller are pretty good as well.
  • Prediction: Priests remain in their current position as a strong Arena class. Fewer Potions of Madness will be a relief.

Rogue

hsreplay_roguedeckwin10.PNG

  • Undercity Valiant is not a draftable card, so it isn’t in this top 10.
  • Rogue loses a couple tools in Dark Iron Skulker, Buccaneer, and Shady Dealer. The class was propelled to #1 thanks to Arena 7.1, and Dark Iron Skulker was likely the culprit for that.
  • The new Rogue toolkit is very solid. Vilespine Slayer, despite being an epic, looks like an autodraft. Obsidan Shard, Hallucination, Biteweed are all solid.
  • Prediction: Rogue remains a top tier Arena class. SI:7 Agent (top deck winrate card) and tools are all still here to keep the class competitive. Lack of AoE didn’t kill the class before, and it won’t now.

Shaman

hsreplay_shamandeckwin10.PNG

  • Whirling Zap-o-matic isn’t a draftable card, so not included in this top 10.
  • Shaman loses none of it’s top 10 cards.
  • Shaman is going all in with Elemental synergy, and I believe this will benefit greatly with all the neutral Elemental cards. The minions are all unimpressive with their stats, so drafting synergy will be key.
  • Prediction: Shaman makes the leap from middle of the pack to top tier. The ability to curve out in the Arena with Elementals will weigh heavily on how well it will do. Some drafts could possibly stall out, if synergy breaks down.

Warlock

hsreplay_warlockdeckwin10.PNG

  • Darkbomb isn’t a draftable card, so it doesn’t belong in this top 10 list.
  • Warlock loses 4 cards from this top 10 list, including Imp Gang Boss, Dark Peddler, Tiny Knight of Evil, and Wrathguard. Really, IGB and Dark Peddler are really bad to lose, as they are really, really good.
  • The new Warlock cards are definitely decent, with the minions being good stat sticks. Chittering Tunneler could be the new Dark Peddler.
  • Prediction: Warlocks get worse and will need to focus on increased minion-based combat with taunts. Could possibly be forced out of top tier status to the middle.

Warrior

hsreplay_warriordeckwin10.PNG

  • These are some putrid winrates. We are looking at the top 10 here!
  • Warrior loses a couple from this top 10, including Obsidian Destroyer, King’s Defender, and Alexstraza’s Champion.
  • Warrior gets very good minion help from Un’Goro, but nonexistent spell support. The three spells they get are nearly undraftable. Sudden Genesis is draftable but win more.
  • Prediction: Is it possible Warriors become even worse? Going forward, the strategy might just be to taunt up and draft weapons, as the spells are no good. That or just continue not playing Warrior.

Thanks to hsreplay.net for all the data and screenshots!

RNG Series: Un’Goro Edition

EDIT: Thanks to @Old_GuardianHS for reminding me adapt is 3/10 not 3/9. Post fixed.

When I first started this blog in the last quarter of 2015, I wrote a bunch of posts where I tried to quantify RNG in the game, the RNG Series. I guess doing the math became too much work for me, as it appears I have not written one of these in 2016 or 2017. Upon the release of all the Journey to Un’Goro cards yesterday, I felt that this expansion had quite a bit of reduction in randomness. Let’s go in and see look at all the random effect cards, and try to quantify some probabilities.

Hearthstone Screenshot 04-26-16 20.02.29.png
Un’Goro RNG ain’t your Old Gods’ RNG

Adapt

  • Single Adapt (12 cards) = 3/10 = 30%
  • Double Adapt (2 cards) = 1 – (7/10)^2 = 51%
  • Galvadon = 1 – (7/10)^5 = 83%

Single Adapts are easy to calculate, just 3 out of 10 outcomes. For multiple Adapts, the events are independent, meaning each roll will be 3/10. So when Galvadon screws you over by not getting Stealth, that was a 17% chance. It will happen.

Discover

  • Hydrologist = 3/5 = 60%
  • Primalfin Lookout = [3/18, 3/14] = [16.7%, 21.4%]
  • Chittering Tunneler = 3/25 = 12%
  • Tortollan Primalist = [3/32, 3/25] = [9.4%, 12%]
  • Free from Amber = 3/30 = 10%
  • Primordial Glyph = 3/32 = 9.4%
  • Servant of Kalimos = 3/36 = 8.3%
  • Stonehill Defender = [3/56, 3/49] = [5.4%, 6.1%]
  • Curious Glimmerroot = 3/59 = 5.1% + Your Brain
  • Hallucination = 3/59 = 5.1%
  • Explore Un’Goro = 3/366 = 0.8%

Discover cards all have a numerator of 3, since you are selecting 1 of 3 picks. The denominator will vary quite a range. Hydrologist has a fairly low RNG, as there will only be 5 Paladin secrets in Standard. While Curious Glimmerroot has an inherent 5.1% of a card, you will likely know what class card is in the opponent’s deck. Also since Discover picks from the pool of Class + Neutral, some cards will vary like Primalfin Lookout for Murlocs, Tortollan Primalist for Spells and Stonehill Defender for Taunts. Explore Un’Goro draws from a pool of Warrior + Neutral cards, giving a 0.8% chance for each card.

Random Card Advantage + Molten Blade

  • Crystalline Oracle = [1/26, 1/1] = [3.8%, 100%]
  • Megafin = 1/19 = 5.3%
  • Molten Blade = 1/23 = 4.3%
  • Elise the Trailblazer = 5/135 = 3.7%
  • Lyra the Sunshard = 1/31 = 3.2%
  • Shimmering Tempest = 1/32 = 3.1%
  • Stampede / Jeweled Macaw = 1/76 = 1.3%

Crystalline Oracle will vary depending on how many cards are left in the deck, so highly variable, but limited in Constructed with repeated cards. Megafin will give a 5.3% chance for each Murloc, and that probability will increase depending on how many cards you get to draw. Stampede and Jeweled Macaw had the probability for a desired Beast plummet after the heavy insertion of Beasts in the new set.

Draw

636265784240330987

  • Tortollan Forager = Turn 2 – 5 attack minions
  • Arcanologist = Turn 2 – Secret
  • Tol’vir Warden = Turn 5 – 2 1-cost minions
  • Mimic Pod = Turn 3 – Anything

Draw cards always start off with a limited denominator of how many cards are left in a deck. Turn 1 players start with 26 cards, while the Coin player starts with 25 cards in deck. Then you subtract what turn a card can be played on inherent card draw. So, the Tol’vir Warden for example, will likely be drawing cards from the smallest deck pool, on average. Mimic Pod is the most variable of these cards, as there is no limiter on what is drawn, like Thistle Tea. Getting 5-attack guys and Secrets are likely more discriminating than 1-cost minions, but these are all random outcomes of limited probability, given the parameters of 30 card decks.

Damage

636265805068200603

  • Volatile Elemental = [1/7, 1/1] = [14.3%, 100%]
  • Sulfuras = [1/8, 1/1] = [12.5%, 100%]
  • Volcano
    • Chance of getting hit with no minions = 99.9%
    • Chance of not getting hit with 14 minions on board = 38%

Back in GvG, random damage appeared to be the main mechanic. Surprisingly, there are only 3 cards like this in the expansion. Volatile Elemental will range from 1/1 to 1/7, so that can obviously be modified. The Ragnaros hero power from Sulfuras will include the hero, so that bumps an extra character. Volcano, obviously is highly variable, depending on it’s own randomness and how much stuff is on the board. With an empty board, the chance of a hero dodging 15 shots is 0.0031%, so nary impossible. On a full board of 14 minions and 2 heroes, the chance of something not getting hit once is 38%. Of course, there are more complicated calculations depending on how much health everything has, as a minion with 1 health dying off will increase the odds of everyone else getting hit. Too complicated.

Summon

636265786513309245

  • Giant Anaconda = [1/10, 1/1] = [10%, 100%]
  • Cruel Dinomancer = varies

Again, we’re not in GvG, when Piloted Shredder dropped off anything. Giant Anaconda at the least has a 10% chance, and this is the highly unlikely scenario of having a hand of 10 5-attack guys. Cruel Dinomancer can be controlled by how much discard you are running. If Clutchmother Zavas was thrown away a lot, she will likely be the most likely outcome of summon, as a 2/2.

The upshot

It’s safe to say that Team 5 took the feedback that bad RNG is bad for the game and competitive Hearthstone. Most of the RNG in Un’Goro is Discover and Adapt, outcomes which have a skill requirement to it. Cards with really wacky RNG outcomes likely won’t be all that good in this set. A card like Stampede is likely a card advantage engine, where you’ll just be looking to get “A Beast” rather than “OP Beasts.”

I think that state of the RNG discussion will focus on the Discover cards, since there will always be variation in a pool of 59 or so cards. Adapt is pretty safe as a mechanic, with the 30% probability floor (not to mention multiple good Adapts).