DR#228 Skill & Polarization supplement

Multiplayer Discussion
1

Polarization Calculation Change
In previous Skill & Polarization Supplements, in the case of mirror matches, I was counting them as 50-50 matchups (so 0 polarization) with a weight equal to the likelihood of the mirror match. I’ve changed this to a weight of zero. This means the polarization metric now excludes mirror matches, and very popular decks will no longer show lower polarization simply because they’re very popular.

Additionally, I’m dropping the general Legend category, retaining Top 1000 Legend and Diamond 4-1.

Top 1000 Legend Polarization
Overall polarization averages to 12.16 (56-44) with 67% of the meta having sufficient matchup data. This is down from 15.19 (58-42) prior to Sunken City release, despite the change in polarization calculation.

Burn Shaman 18.51 (59-41)
Aggro DH 16.47 (58-62)
Pirate Warrior 12.74 (56-44)
Naga Mage 12.71 (56-44)
Ramp Druid 11.16 (56-44)
Pirate Rogue 9.58 (55-45)
Mech Mage 9.15 (55-45)

Diamond 4-1 Polarization
Overall polarization is 15.30 (58-62) with 88% of the meta having sufficient matchup data. This is up from 13.30 (57-43) prior to Sunken City release, although there was a change in polarization calculation.

Curse Warlock 38.38 (69-31)
Control Warrior 31.96 (66-34)
Holy Paladin 26.06 (63-37)
Shellfish Priest 23.86 (62-38)
Aggro DH 20.39 (60-40)
Burn Shaman 18.27 (59-41)
Murloc Warlock 17.33 (59-41)
Quest Hunter 17.23 (59-41)
Mech Paladin 16.25 (58-42)
Pirate Warrior 14.50 (57-43)
Ramp Druid 12.99 (56-44)
Naga Mage 12.95 (56-44)
Fel DH 11.60 (56-44)
Face Hunter 11.31 (56-44)
Mech Mage 9.92 (55-45)
Pirate Rogue 7.97 (54-46)

Overall Skill Rating by Archetype

Skill Rating meaning reminder

Skill Rating is not a (direct) measure of the skill of people piloting a certain deck. Skill Rating is a measure of

  1. the opportunities a deck gives its pilot for skillful play
  2. that have a meaningful impact on winrate
  3. given the context of the current meta
  4. relative to the opportunities the opponent receives.

A negative score doesn’t indicate that, as pilots increase in skill, they pilot their own decks worse. It indicates that, as both player and opponent increase in skill, opponents get more opportunities for skillful play than players do.

Archetype, Class, T1KL skill
Naga Mage +4.29%
Holy Paladin +2.36%
Phylactery Warlock +2.26%
Reno Paladin +2.14%
Quest Priest +1.26%
Control Warrior +1.24%
Dragon Priest +1.12%
Shellfish Priest +0.98%
Deathrattle DH +0.93%
Murloc Shaman +0.65%
Thief Rogue +0.58%
Curse Warlock +0.3%
Silence Priest +0.14%
Fel DH -0.14%
Face Hunter -0.2%
Burn Shaman -0.29%
Ramp Druid -0.33%
Pirate Rogue -0.83%
Quest Hunter -1.1%
Big Beast Hunter -1.24%
Aggro DH -1.36%
Murloc Warlock -3.54%
Mech Mage -4.1%
Mech Paladin -4.31%
Quest Warrior -7.62%

Commentary
There’s actually a deck more polarizing than Control Warrior: Curse Warlock. That’s bad. But it’s not very popular and it’s winrate is lousy. That’s good. But the toppings contain sodium benzoate. That’s a joke.

But Curse Warlock isn’t alone. The expansion has seen experimentation, mostly at Diamond and not so much at Legend, with various control decks that are all pretty polarizing, including good ole Control Warrior and Quest Hunter (don’t get upset control purists, I’m only saying it’s the control vs aggro matchups), as well as Holy Paladin and Shellfish Priest. If these decks rise to prominence, the meta will become as paper rock scissors as the last meta, if not worse.

Naga Mage is predictably the highest skill rating deck, and Pirate Warrior is predictably the lowest. What’s impressive and a bit unexpected is the skill rating for Holy Paladin, and just how low in skill rating Mech decks are. I thought there’d be more skill to Mecha-Shark combo turns, but apparently not.

2

yea, holypaladin is not as braindead as i thought i have to admit playing it for a bit. Still not terribly much to think about, which tells a lot about the meta in general.

now we just have to hope there´s some data on the mystified boarpriest next report.

3

Yeah, it’s weird that VS covered Boar Priest in #227 and left it completely out of #228. I was looking forward to it, but I can’t analyze what I don’t have data on

mumbles under his breath

not that that would stop half the people here.

1 Like
4

Well, looking at the Class/Archetype distribution for all ranks in #228, Priest was 10.3% of the overall meta. Shellfish Priest accounted for 3.05% within that; Quest Priest for 2.67%; and Switcheroo Priest of 2.64%. The remaining 1.94% of the meta that Priest decks made up were divided among at least five different archetypes, based on the differently shaded sections of the bar in the graph.

They do include a Low Sample Tier Estimate for Boar Priest at all three of their typical levels in the power rankings, so Boar Priest is likely one of the larger sections, but even then it has to be at sub-1% representation, so it’s not surprising that it disappeared compared to report #227 when Boar Priest was one of four unlabeled Priest decks in the Class/Archetype Distribution chart which together accounted for 4.96 percent of the meta.

So, it looks like Boar Priest just dropped to too low a sample size to report since rotation.

VS can’t really report things they don’t have enough data on any more than you can analyze what you don’t have data on.

[The exact percentages might be slightly off, since VS seems to round class representation to a single decimal point but seems to round archetype representation to two decimal points, but that should affect the overall point.]