Historical Class Tiers Data

[Tyris Flare]

If people didn't see, subcreation put up a historical list of class tiers for M+ going back to when he started it, BFA S2. So 6 seasons of data, with the current tier pretty much going to map onto the previous season.

https://twitter.com/subcreation/stat...54407534399488

Given the way the rankings work I think it is legit to quibble around the margins about accuracy, but in the broad, macro picture this pretty much maps onto what everyone already knows. Basically, Melee in m+ is completely dominated by the 3 leather classes, with the plate classes and others usually ranging from mediocre to bad every single season.

Ranged sees much more variance, where even consistency good specs like elemental are worse overall than the leather melee train.

Thankfully, it does seem like the DF talents should help out healers and tanks the most, since a lot of those rankings are based on DPS which the developers didn't even seem to bother to try and balance until recently (and it's still not great).

[Mardux]

Its hard to not think the table was created in bad faith seeing as, for example, in BFA S3 they rank spriest 3 points higher than ele sham despite them being the same rank according the the actual tables linked in the subcreation FAQ. Spriest isn't even the first listed in the table on Subcreation but they gave it more points than the actual first listed of the A tier specs.

Same with the melee specs. Sin rogue was given more points than WW, but the SubC table shows WW being the better of the A tier melee that season.

[Tyris Flare]

You can read through the actual analysis or twitter for more specifics, but I think that is all stuff at the margins. This isn't the type of thing to say "hey this was 2.3 and this was 2.4, that's horribly balanced!"

It's useful as a meta macro analysis of broad trends. For example, mistweaver monks suck and the fact this has never been addressed should be pretty embarrassing for the devs.

[Biomega]

I'm questioning the usefulness of linear ranking here. What I mean by that is treating each rank "point" as of equal value, as if someone dropping from #1 to #2 was the same as someone dropping from #7 to #8. The reason for that is simple: you have limited slots in a 5-man group. Being in #1-3 for DPS is *vastly* better than being #8-10 or whatever, because anything that isn't top tier has a disproportionately tougher time getting into groups and being part of meta comps - which means there's less data, more biases, and less stringent analysis, along with the just obvious problem of accessibility for the average player.

To be useful as an overall trend for class representation, there would need to be some kind of weighting going on here that properly represents just how much better being a top class is than being a medium class.

[ceall]

Tbh, this list should have been made on a per-class basis (per dps class). I know for a fact, that shadow priest has never been desired in m+ for example.

[Ashana Darkmoon]

Tbh, this list should have been made on a per-class basis (per dps class). I know for a fact, that shadow priest has never been desired in m+ for example.

I think the ratings are a little inflated in BFA because of the one guy doing high keys as shadow in Consequence lol. He gave up on it in S4 into SL i believe since it was clearly worse as an alternative.

But I also think there's a gradual change to FOTMism where the top couple specs are even farther ahead than previous as the community becomes more efficient.

[enigma77]

all 3 warlocks garbage for several years straight? yep I remember that

season 3 in SL was our fancy vacation to the tropics, slurping martinis

[facefist]

You can read through the actual analysis or twitter for more specifics, but I think that is all stuff at the margins. This isn't the type of thing to say "hey this was 2.3 and this was 2.4, that's horribly balanced!"

It's useful as a meta macro analysis of broad trends. For example, mistweaver monks suck and the fact this has never been addressed should be pretty embarrassing for the devs.

Biased macro analysis is still biased and as incorrect as biased micro analysis. If what the other guy said is true and they use subjective views in determining the final ranking, then this is as valid as youtubers going through world of logs and wowprogess and telling which spec is the best.

A method including subjective view that is being sold as purely statistical evidence just takes away all of the reliability from this stuff. And I mean my personal feel is that those rankings are somewhat right for the most part, but I don't think their data is statistically significant at all. It's just another subjective list.

[Nymrohd]

If I were to do that I'd give a full point to the top spec and give the rest % depending on performance compared to the top at the time on the bracket.