Jump to content
  • Sign Up

Diversity of the Specializations at Raids - in the last 4 years


Recommended Posts


I would like to bring you some data which has been bugging me for a while now. I would like to discuss the diversity of the classes in the instanced, 10-man, PvE content – Raids, in the last 4 years. I am going to use GW2Wingman database with 6 million logs uploaded between 30th of July 2018 and 08th of July 2022 divided per significant balance updates.

Disclaimer: The analyzed data is based solely on ArcDPS logs which were actually sent and uploaded to GW2Wingman. The website removes duplicated entries, so if any squad member of your recent fight has uploaded it, you won't need to. We have double filter: data is based on logs and logs which were uploaded to certain database. We could call it high-end community's database as more casual players usually don't care about databases like that.

We have 21 Balance Updates and 27 Specializations to check.


There are few key points to address:

1) In 2018, less than 4% playable specs were Core (majority of them were Core Warrior) which has decreased to less than 1% in 2020. In 2022, it's less than 0.04% (for every Core Spec in Raids, you will find 12 Vindicators).

2) Since July 2018 until March 2019 the least played class was Revenant. For the next 7 months it was Engineer. I know you were looking for that - since July 2020, the least playable class in Raids was Elementalist and it has lasted for the next two years. But I bring you the happy news - the least playable class right now is... Warrior! You shall encounter Elementalist 21% more likely than Warrior.

3) While having dataset prepared from July 2018, I would like to focus on the piece of the data between November 2021 and June 2022, which is, well, now. Let's see how things have changed in 7 months.

a) Only two old specs are played more: Soulbeast (increased by 165.60%) and Herald (increased by 319.35%),

b) Only one old spec is in the same place as it was before: Firebrand (99.22%). It means that Firebrand is visible in 15% +/- 2% ,constantly since July 2020 till now - it is highly unlikely it would change,

c) Right now, the old specs are more popular than EoD - 55% vs. 45% (24% of which is Mechanist all alone with 10% of Virtuoso),but we must consider that we have 18 old specs vs. 9 new specs.

4) Let's address EoD specs, how they were add the premiere and right now,

a) Harbringer lost 40% of it's players, Willbender lost 65% of it's players,

b) Catalyst, Vindicator and Bladesworn lost some of the players,  but it's less than 10-20%,

c) Specter has gained a little boost in the popularity by 20%,

d) Virtuoso and Mechanist doubled their playerbase,

e) Untamed is played 900% more than at the premiere of EoD but it's still 2nd the least popular spec (poor Vindicator).


I believe, we have ground data filled in, let's analyze it.

The first thing I was starting to wonder, how BIG meta is. I believe we all can agree on the statement below:

If the data shows that the specialization has percentage bigger than 10%, we may assume it was included at least once in every encounter.

I know, it may be false sometimes but we may assume it's minor problem here. Of course, take my statements and observations with grain of salt.

At least 1 spec in Raid Squad on average (2021): https://imgur.com/a/QA08PKp
The popularity of 2021 Specializations: https://imgur.com/a/Q5kXQIE

In 2021, in Raid Squad you would see at least one Druid, Berserker, Scourge, Renegade and Firebrand. At the time, Druid was a standard pick as the healer in almost every encounter - it brought 25x Might due to GotL, it brought Fury, Spirits, utilities, we all know what is what. Great, next. Berserker = Banners, BannerSlave wasn't joke at all. Renegade, our Alacrity delivery for 10man team. Firebrand - solid presence of 15%, so 1.5 per squad, standard Quickness Provider. If we summarize Chronomancer (6.7%) and Scrapper (2.4%), we had Quickness set up.  Scourge - the most popular DPS class, the healer build was rarely used as contrary so we can decide that Scourge was DPS#1 at the time.

7 months later, this is our situation which we deal with.

At least 1 spec in Raid Squad on average (2022) : https://imgur.com/a/3X0LvVd

The popularity of 2021 Specializations (2022): https://imgur.com/a/jgIvfyf

In 7 months, we have 2 less specializations which we would consider meta, as requirements for the squad. Scourge is no more meta, Virtuoso is DPS#1 (probably?), you have 1 in every squad on the average. Firebrand-boy, still strong, still there, at least 1.5 per squad as Quickness Provider, beside Harbringer (3.67%), Scrapper (1.91%), Chronomancer (1.48%) and Herald (0.99%).

And there it is, our birthday boy, The Mechanist. The King of the EoD. The Emperor of Healers, the Destroyer of DPS, all in all, the biggest meta class in the long time. The Mechanist's presence sky-rocketed to 24.36%. It means that almost 2.5 players in a Raid Squad, is a Mechanist. Mechanist is a great healer and provides alacrity, we all have seen the squads with the Double Heal Mechanists, it is also solid DPS specialization, no matter if it's condi or power, there are/were multiple builds for Maces, Pistols, J-Drive, Dynamo.

The Mechanist, 24.36%. Has been there any spec in the last 4 years which dominated the competition that much? The answer is: NO. The highest one was Chronomancer between December 2018 and March 2019 with 21.83% presence. Only once similar situation has encountered - between August 2017 and September 2017, Berserker 24.86% visibility.

The Mechanist dominated the competition, but we have a bigger problem.

Unplayable specializations.

In the last balance patches, we have learnt that we should "bring the tool, not the class" (boons). Let's look at the graph below:


It is the amount of the classes with the presence below 1% in the last 4 years. In the second half of 2021, we had 2-3 specializations (Herald, Spellbreaker, Deadeye*), right now - we have 9 of them (still Herald, Spellbreaker, Deadeye but also Weaver, Catalyst, Untamed, Vindicator, Willbender and Bladesworn). Hold on. Wait. The last 3, we have added Weaver (sorry Ele mains) but also 5 new EoD classes? Yes. 5 of 9 EoD specializations are present in less than 1% of raids. How many of less 1%, 5% and 10% are there? In the graph:


The last point, I would like to raise is the situation we had in 2020 after The Great Balance Patch (fyi, the June was hyped as the successor of February 2020, haha):


We all know it was different time but for it did look better two years ago than it is right now.


TL;DR starts here


I do not want to judge anything (or, at least I try) but we can summarize it.

The Diversity Plan has failed. We have less specializations played frequently than before. Adding Quickness or Alacrity sources to different class has little-to-none impact on play rate of these classes in Raids. The Quickness Provider is Firebrand, the Alacrity Provider is Mechanist (one of them is the healer of course). In the DPS role we either use the mentioned pair or Virtuoso (Scourge is behind a little but we may consider it as playable). We have huge increase in the amount of specializations which noone wants to play, it is even true for all the classes! Only 2.72% players use Elementalist, only 2.17% plays Warrior and only 4.57% players use Revenant. Virtuoso is more popular than all Ranger specializations combined! Firebrand as Quickness Provider is more popular than combined Chronomancer, Scrapper, Herald and Harbringer... twice!

We have the worst diversity among specializations since at least 4 years... while Balance Patches are actively working against it. Interesting, right?

  • Like 14
  • Thanks 8
  • Haha 1
  • Confused 1
Link to comment
Share on other sites

4 hours ago, LKEY.9567 said:

We have the worst diversity among specializations since at least 4 years... while Balance Patches are actively working against it. Interesting, right?


Nice post. I'll add a few things to mention:


Diversity is hard to measure...so I applaud you for going in and getting something for people to wrap their head around in a way that can be easily interpreted... but the problem is much much worse...


Point 1

One issue is that the only kind of measurements taken, are of specs rather than builds, which there can be many of them. Consider that there are 27 (now 36) specializations, each specialization has 27 possible trait options to chose from, meaning that potentially, there are 27 possible builds per specialization (729 total). The actual total number of builds far exceeds this when we consider every permutation of utility options, weapon options, rune and sigil choices...which add up to probably a million or more different possible builds that a player has the option to choose from.


Data that captures diversity of specializations, is but an incredibly tiny drop in the ocean of potential diversity the game could potentially have. For this reason, a more suitable dataset to use, rather than how many specializations are being played at any given time, is the number of builds that are actually in the meta at any given time... for example going onto Meta-battle, and checking out what is currently there everyday and recording how that changes through time. Unfortunately there are no recorded metrics that we can use to see what options people are actually choosing per spec...so measuring diversity in it's truest form that exposes just how bad it is in guild wars 2, is pretty much impossible to do...and the only information we can access is data coming from meta-battle...which is unfortunately controlled by a Non-Darwinian system making it faulty. But...


Humor the following example.


If there are 40 builds listed in metabattle, you can then reliably say that "Out of a million possible combinations only 40 builds are being played in this game"


If there are a million builds considered meta on meta-battle, then the distribution of specializations in your dataset of ARCDPS will all be the same (3.7%). Likewise if there are only 27 builds listed on meta-battle, 1 per each spec, then the distribution of specializations in your dataset of ARCDPS will also all be the same (3.7%), so to annihilate that measurement error, you need to get a much larger dataset. Luckily it doesn't take a genius to just look at meta-battle to instantly conclude there aren't a million builds on it...and so one can conclude rather self-evidently that guild wars 2 lacks diversity...big time.


Point 2

In addition, diversity isn't only a statement about the number of options, but the amount of differences between those different options. This is a lot harder to parametrize in a meaningful way...but it's like asking "If you were to replace skill A on this build with skill B, how different would this build be from each other or are these difference incidental, enough to warrant them as being the same build." 


To measure how incidental the differences are is a bit arbitrary. If you have a Dagger Onslaught Reaper and a Great-sword Onslaught Reaper...both are still "Onslaught Reapers" just using different weapons. We can say that the difference in their builds is incidental at best, as the way that they are played don't warrant enough of a difference to us, worthy of giving it a unique build name. This shows up in the data mentioned above regarding metabattle...so it is quantifiable....just not easy to talk about as those data points are a function of a darwinian system and how it selects.


Anyway Lkey, thanks for taking the time to attempt parametrizing the problem for others, and I want the above to help others further understand just how bad the problem really is that the dataset doesn't capture.



Edited by JusticeRetroHunter.7684
  • Like 2
  • Thanks 3
Link to comment
Share on other sites

First, big words of appreciation to the author. In LKEY.9567's statement, there is not a word of criticism against Arena Net, just raw data and statistics that have been built on tools that are available to everyone, including players. That's right, that's the (l)key word - PLAYERS

We are determined, most of us love this game, we use private resources and skills (often very professional like in this case) to prove that we want the game to get better, more popular and truly balanced. 

Based on the example of other games, the creators often realize how important knowledge can be carried by players. In one of the popular MOBA games, top players are asked for their opinion on major balance changes. It allows you to learn perspective. Because then many people work on one issue. In Guild Wars 2, they could be people who have spent 2,000 hours with Firebrand who use only one set of weapons, but also people who know the exact numbers and demonstrate a professional understanding of the class and are able to show its maximum potential in practice. The opinions of both such people are important in the discussion.

I don't want to bring up the problems of the June 28 balancing patch, and I don't want to go into details that have already been covered multiple times. I will just say this, the presence of the Mechanist and his ability to be the best support, power dps and condi dps is toxic because it kills diversity. And the worst of it all is that - no immediate steps have been taken to eradicate this glaring problem as soon as possible.


Currently in many countries we have holidays, statistically people have more time to play computer games, but we have to wait until September to see the balance patch for the previous balance patch, even though the problem is so huge and visible. 


It's not even an opinion filled with sadness anymore. It is a lament filled with despair. Please, finally pay attention to the opinion of EVERY player who creates this beautiful community and use it in your own but honest way.



From a player who plays every class except the mechanist


  • Like 1
  • Thanks 3
  • Sad 1
Link to comment
Share on other sites

Using ArcDPS logs to provide clear and easy to read numbers which are not in favor of Anet's intentions?

Y'all know what that means... Correct, ArcDPS gotta go.


I was joking, but now I can actually see that happening and don't like it.

  • Like 1
  • Haha 2
  • Confused 1
  • Sad 1
Link to comment
Share on other sites

4 hours ago, Gajex.2386 said:

Using ArcDPS logs to provide clear and easy to read numbers which are not in favor of Anet's intentions?

Y'all know what that means... Correct, ArcDPS gotta go.


I was joking, but now I can actually see that happening and don't like it.

Then they will remove the training area golem, after that the damage numbers that pop up on the enemy will be removed and finally the combat log is gonna go too.

  • Like 1
  • Haha 2
Link to comment
Share on other sites

100% this information is great to have, I've said a few times that the latest patch & the "boons for everyone" approach would make the game less diverse. And while it has felt that way, its nice to have tangible data to refer to.

Here's hoping the focus is on balancing with fun in mind vs. whatever has been happening recently.

  • Like 1
  • Thanks 1
Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
  • Create New...