I recorded the statistics of all my games for Season 14. Here are the results

[Rumpduree.4360]

One of the things I have wanted to do for the past few seasons is to record the professions of each player in my matches along with the information (damage, healing, % game, % team, etc) displayed post-match. The introduction of specialisation icons to the PvP scoreboard seemed like a good opportunity to start and also enabled more accurate information to be collected on what classes/builds are the most prominent. The numbers have been summarized below and there are also some basic graphs available at https://imgur.com/a/hvBRLnu.

The total number of matches I'd played this season is 166. Unfortunately, I only remembered to start recording the information after I had played the 10 placements and one extra match, so the following data is for the 155 matches from number 11 to 166. All of the matches were played on a power shatter mirage (Sw/T - GS, Domination/Dueling/Mirage, Demo Amulet) on NA servers. The time I was queuing was generally between 8pm - 12pm AEST (UTC +11).

 

General/Interesting Information:

* Wins: 91

* Losses: 75

* Placement Rating: 1512

* Lowest Rating: 1377 (Match 152)

* Highest Rating: 1600 (Match 108)

* Finishing Rating: 1527

* Average game duration: 10m4s

* Average Winning game K:D ratio: 11.9

* Average Losing game K:D ratio: 2.0

* Matches finishing with <50 points difference: 16

* Matches finishing with >50, <100 points difference: 24

* Matches finishing with >100, <200 points difference: 38

* Matches finishing with >200, <300 points difference: 46

* Matches finishing with >300, <400 points difference: 29

* Matches finishing with >400 points difference: 12

 

Specialisations:

(Please note that this list is not counting the specialisation i was queuing on, nor does it differentiate between teams. If a player swapped in the pre-match i would record what they had swapped to. Think - "Across 155 games, excluding myself, there are 1,395 players, what were their specialisations?")

* Elementalist: 6 (0.4%)

* Tempest: 9 (0.6%)

* Weaver: 33 (2.4%)

* Engineer: 3 (0.2%)

* Scrapper: 16 (1.1%)

* Holosmith: 95 (6.8%)

* Guardian: 126 (9.0%)

* Dragonhunter: 23 (1.6%)

* Firebrand: 61 (4.4%)

* Mesmer: 7 (0.5%)

* Chronomancer: 30 (2.1%)

* Mirage: 156 (11.2%)

* Necromancer: 8 (0.6%)

* Reaper: 133 (9.5%)

* Scourge: 131 (9.4%)

* Ranger: 21 (1.5%)

* Druid: 29 (2.1%)

* Soulbeast: 142 (10.2%)

* Revenant: 1 (0.1%)

* Herald: 61 (4.4%)

* Renegade: 6 (0.4%)

* Thief: 59 (4.2%)

* Daredevil: 93 (6.6%)

* Deadeye: 34 (2.4%)

* Warrior: 16 (1.1%)

* Berserker: 8 (0.6%)

* Spellbreaker: 92 (6.6%)

 

Maps:

* Battle of Kylo: 22 (13%)

* Eternal Coliseum: 42 (26%)

* Forest of Niflhel: 38 (23%)

* Legacy of the Foefire: 36 (22%)

* Revenge of the Capricorn: 15 (9%)

* Skyhammer: 3 (2%)

* Temple of the Silent Storm: 7 (4%)

 

It would be cool to see other people collect similar information for Season 15 to see how it compares. If anyone is interested in seeing the spreadsheet let me know and I can post a link.

[mortrialus.3062]

!!!!

 

I love stats!

[rank eleven monk.9502]

Pretty interesting statistics. It's also worth noting that mesmer does not stand out in the number of players, even though you queued as a mesmer, which would generally mean higher numbers thanks to the class stacking rules.

[otto.5684]

I think what is telling in these statistics is the game scores. Out of 166 games played, 87 had a score deferential of more than 200 points (52.4%). If we include the games between 100 and 200 point deferential, the number of games goes up to 125 (75.3%).

 

Granteed, this is the experience of one player, but my overall experience and is similar and most players I have talked to as well. The match making system rarely works. Or more specifically rarely results in competitive games. You could urge that competitive and match making are not the same, however, undoubtedly they are very strongly correlated. Also, you games with all the players are above 1500 should be fairly competitive, since the odds of under experienced players throwing the game is not likely.

 

@"Ben Phongluangtham.1065" not basing on you guys, but trying to draw conclusion from input (like the system only matches players within 25 rating together), is not a good way to conclude. You must use the results.

 

I know you guys mentioned before there are other factors impacting game outcomes. Beside match making the only 2 large factors is: class balance and lack of individual specific class rating.

 

I do not deny I (and honestly any other player), lack sufficient data to be able to precisely analyze and accurately conclude. However, the overwhelmingly player experience is that sPvP games are rarely fun and/or rarely competitive. I can not determine a specific culprit. All I can tell you, it is hard to play fun games. And to be specific, fun does not equal win.

[AliamRationem.5172]

> @"otto.5684" said:

> I think what is telling in these statistics is the game scores. Out of 166 games played, 87 had a score deferential of more than 200 points (52.4%). If we include the games between 100 and 200 point deferential, the number of games goes up to 125 (75.3%).

>

> Granteed, this is the experience of one player, but my overall experience and is similar and most players I have talked to as well. The match making system rarely works. Or more specifically rarely results in competitive games. You could urge that competitive and match making are not the same, however, undoubtedly they are very strongly correlated. Also, you games with all the players are above 1500 should be fairly competitive, since the odds of under experienced players throwing the game is not likely.

>

> @"Ben Phongluangtham.1065" not basing on you guys, but trying to draw conclusion from input (like the system only matches players within 25 rating together), is not a good way to conclude. You must use the results.

>

> I know you guys mentioned before there are other factors impacting game outcomes. Beside match making the only 2 large factors is: class balance and lack of individual specific class rating.

>

> I do not deny I (and honestly any other player), lack sufficient data to be able to precisely analyze and accurately conclude. However, the overwhelmingly player experience is that sPvP games are rarely fun and/or rarely competitive. I can not determine a specific culprit. All I can tell you, it is hard to play fun games. And to be specific, fun does not equal win.

 

"Should be" competitive. But matchmaking is extremely broken at the upper levels due to the small player pool. I haven't check in this season, but if it's the same as it's been then top 250 is basically plat 2, plat 3, and legendary tier. If every player in the top 250 played for an hour every day and the distribution of play times were evenly distributed (of course, they aren't!), you'd have just enough players in plat 2, plat 3, and legendary tier to fill a single match with 5 players on each team at any given time.

 

That would still represent a horrible mismatch as players would have no choice but to be matched with players spread across several tiers (not to mention being matched with the same group of players match after match!). Based on my experience with the sensitivity of rating, I don't think it's ideal to have players matched more than +/- 50 points from their current rating. It's obviously impossible to come remotely close to meeting that criteria in matchmaking at 1500+. We simply need a lot more people to make that happen.

 

What I'd be curious to know is if matchmaking does any better in terms of outcomes at the middle ratings like gold and silver? Do the majority of matches come at 200+ point differentials? Or does the matchmaking algorithm essentially settle for whatever allows a player to achieve rating equilibrium that accurately represents their performance level over time in this game mode? It can obviously do that without ensuring that all or even a majority of matches are close. Essentially, blowout matches don't count - you're as likely to have a match you can't win as one the opposing team can't win. Thus whether you climb or fall in rating is determined by the minority of matches in which you have a chance to win and make an impact for your team.

[Doto.6357]

> @"rank eleven monk.9502" said:

> Pretty interesting statistics. It's also worth noting that mesmer does not stand out in the number of players, even though you queued as a mesmer, which would generally mean higher numbers thanks to the class stacking rules.

Are you sure about that? It looks like mirage was the most common type of player he encountered out of all elite specs, with 156

 

 

[Arheundel.6451]

Mesmer

Necromancer

Guardian

 

The most spammed classes in PvP with ele being the lowest....no surprise here

[Duncanmix.5238]

I don't know if its EU vs NA thing, but this season I had plenty of close matches. What I noticed is overall sad state of HoT specs. With exception of Necro where its split with Reaper and Scourge, and maybe with thief where u will see both Daredevils and occasional Deadeye, all other classes are dominated by PoF specs. But I guess balancing 18 specs is really difficult.

[Imperadordf.2687]

> @"Doto.6357" said:

> > @"rank eleven monk.9502" said:

> > Pretty interesting statistics. It's also worth noting that mesmer does not stand out in the number of players, even though you queued as a mesmer, which would generally mean higher numbers thanks to the class stacking rules.

> Are you sure about that? It looks like mirage was the most common type of player he encountered out of all elite specs, with 156

>

>

 

I think he meant that the OP queued as Mesmer so the Mesmer statistics would be higher in reality (like, Mirage would be 200 or so) due to class stacking rules.

[Illconceived Was Na.9781]

I count 1399 specializations in the list presented by the OP. The top 5 were:

* Mirage 156 appearances

* Soulbeast 142

* Reaper 133

* Scourge 131

* Guardian 126

 

If we combine specs to look at each class, the top 5 are:

* Necro 272 appearances

* Guardian 210

* Mesmer 193

* Ranger 192

* Thief 186

 

For the bottom four: Revenant (class) & elementalist (class) combine for 116, which is about the same as Warrior (116) & Engineer (114)

 

****

PS as noted by @"Imperadordf.2687", the OP queued as a mesmer, which would skew the class stats due to stacking rules for matchmaking. In other words: this isn't close to enough data to show patterns relevant to everyone.

[thijsken.5031]

Only 3 games of skyhammer, dam

[Mbelch.9028]

Neat numbers -- thanks for recording all of them.

[MyPuppy.8970]

I like numbers but it's too much of a pain to record everything. But from my feeling and experiences, It's quite like that. I saw more Druids though.

[Rumpduree.4360]

> @"Illconceived Was Na.9781" said:

> I count 1399 specializations in the list presented by the OP. The top 5 were:

> * Mirage 156 appearances

> * Soulbeast 142

> * Reaper 133

> * Scourge 131

> * Guardian 126

>

> If we combine specs to look at each class, the top 5 are:

> * Necro 272 appearances

> * Guardian 210

> * Mesmer 193

> * Ranger 192

> * Thief 186

>

> For the bottom four: Revenant (class) & elementalist (class) combine for 116, which is about the same as Warrior (116) & Engineer (114)

>

> ****

> PS as noted by @"Imperadordf.2687", the OP queued as a mesmer, which would skew the class stats due to stacking rules for matchmaking. In other words: this isn't close to enough data to show patterns relevant to everyone.

 

I must have done something wrong when counting up all the classes, it shouldn't be 1399 in total. I'll have a look through the data again and edit the OP.

 

As has been mentioned, the sample size here is too small to be representative of the general population and the specialisations would be skewed because I queue on mesmer. However it does show how prominent mirage, reaper, scourge, soulbeast and core guard are at the moment.

 

It would be interesting to see what influence a players rating and the class they queue on has.

[Illconceived Was Na.9781]

> @"Rumpduree.4360" said:

> I must have done something wrong when counting up all the classes, it shouldn't be 1399 in total. I'll have a look through the data again and edit the OP.

For this limited sample, it's "close enough" to give people the right idea (well, as long as it was just a minor typo or two, not something fundamental, which seems unlikely).

Either way, if you do update, let me know & I'll update my summary accordingly.

 

> As has been mentioned, the sample size here is too small to be representative of the general population and the specialisations would be skewed because I queue on mesmer. However it does show how prominent mirage, reaper, scourge, soulbeast and core guard are at the moment.

Yes.

 

> It would be interesting to see what influence a players rating and the class they queue on has.

I don't think we'd be able to see that from a half dozen samples or even a dozen. ANet's metrics come from all matches, not just the few any of us would happen to participate in. What we can see is which specs or classes predominate versus not in the current balance.

 

What would be interesting is if we had the same data from the same player over the various balance patches. There's always some build that people think is OP. Does the predominate build/spec have roughly the same fraction of participation all the time? Or has there been more even distribution in some seasons?

 

****

I forgot to say above: thanks for taking the time to collect & publish the data. It's a breath of fresh air compared to people basing their reactions on their impressions/memory of what they've experienced.

 

[yanniell.1236]

Well, the matchmaking sucks. Who would know...

[DanAlcedo.3281]

Holy Moly.

Average K/D if winning of 12.

 

But he plays Mesmer so no wonder.

[Rumpduree.4360]

> @"DanAlcedo.3281" said:

> Holy Moly.

> Average K/D if winning of 12.

>

> But he plays Mesmer so no wonder.

 

I dont think this is a mirage thing necessarily, rather just playing the role of a roamer.

 

Generally in these games the rest of the team playing well enough to allow me to pick up easy +1's, swing teamfights and gain further momentum.

 

The opposite sort of thing happens for the losses. A k/d of 2 means I'm not really filling the role properly, for whatever reason, and then it just piles the pressure on to the rest of the team.

[Spartacus.3192]

> @"otto.5684" said:

> I think what is telling in these statistics is the game scores. Out of 166 games played, 87 had a score deferential of more than 200 points (52.4%). If we include the games between 100 and 200 point deferential, the number of games goes up to 125 (75.3%).

>

This is the more interesting stat.

 

Thanks OP for creating this thread.

[Spartacus.3192]

Another interesting fact which i am sure the mesmer mains will jump on.

If you combine all the elite specs into their respective core class names, so sum core guard, dragonhunter and firebrand as one stat, you get the following results,

 

Class Sum of Pct

Necro 19.5

Guardian 15.0

Mesmer 13.8

Ranger 13.8

Thief 13.2

Warrior 8.3

Engineer 8.1

Revenant 4.9

Elementalist 3.4

 

Necro (core, reaper, scourge) in all combined version is the most played class. Combined Mirage, Chrono and core Mesmer is only 3rd place.

 

ELE and REV needs some kind or buffs/reworks.

[RedShark.9548]

> @"Spartacus.3192" said:

> Another interesting fact which i am sure the mesmer mains will jump on.

> If you combine all the elite specs into their respective core class names, so sum core guard, dragonhunter and firebrand as one stat, you get the following results,

>

> Class Sum of Pct

> Necro 19.5

> Guardian 15.0

> Mesmer 13.8

> Ranger 13.8

> Thief 13.2

> Warrior 8.3

> Engineer 8.1

> Revenant 4.9

> Elementalist 3.4

>

> Necro (core, reaper, scourge) in all combined version is the most played class. Combined Mirage, Chrono and core Mesmer is only 3rd place.

>

> ELE and REV needs some kind or buffs/reworks.

 

How much a class is played shouldnt effect your balance choices honestly, ppl just might not like playing them.

 

What you should look at when trying to balance are ingame stats like as an easy example winrate. Could also look at kill participation, average dmg output etc.

 

[Spartacus.3192]

> @"RedShark.9548" said:

> > @"Spartacus.3192" said:

> > Another interesting fact which i am sure the mesmer mains will jump on.

> > If you combine all the elite specs into their respective core class names, so sum core guard, dragonhunter and firebrand as one stat, you get the following results,

> >

> > Class Sum of Pct

> > Necro 19.5

> > Guardian 15.0

> > Mesmer 13.8

> > Ranger 13.8

> > Thief 13.2

> > Warrior 8.3

> > Engineer 8.1

> > Revenant 4.9

> > Elementalist 3.4

> >

> > Necro (core, reaper, scourge) in all combined version is the most played class. Combined Mirage, Chrono and core Mesmer is only 3rd place.

> >

> > ELE and REV needs some kind or buffs/reworks.

>

> How much a class is played shouldnt effect your balance choices honestly, ppl just might not like playing them.

>

> What you should look at when trying to balance are ingame stats like as an easy example winrate. Could also look at kill participation, average dmg output etc.

>

 

Agreed i was making a general observation. However if you believe the player base of any given MMO is spread evenly then if a class is disproportiantely under represented then a rework (not necessarily a buff) is probably required to help increase popularity.

[RedShark.9548]

> @"Spartacus.3192" said:

> > @"RedShark.9548" said:

> > > @"Spartacus.3192" said:

> > > Another interesting fact which i am sure the mesmer mains will jump on.

> > > If you combine all the elite specs into their respective core class names, so sum core guard, dragonhunter and firebrand as one stat, you get the following results,

> > >

> > > Class Sum of Pct

> > > Necro 19.5

> > > Guardian 15.0

> > > Mesmer 13.8

> > > Ranger 13.8

> > > Thief 13.2

> > > Warrior 8.3

> > > Engineer 8.1

> > > Revenant 4.9

> > > Elementalist 3.4

> > >

> > > Necro (core, reaper, scourge) in all combined version is the most played class. Combined Mirage, Chrono and core Mesmer is only 3rd place.

> > >

> > > ELE and REV needs some kind or buffs/reworks.

> >

> > How much a class is played shouldnt effect your balance choices honestly, ppl just might not like playing them.

> >

> > What you should look at when trying to balance are ingame stats like as an easy example winrate. Could also look at kill participation, average dmg output etc.

> >

>

> Agreed i was making a general observation. However if you believe the player base of any given MMO is spread evenly then if a class is disproportiantely under represented then a rework (not necessarily a buff) is probably required to help increase popularity.

 

Well if we just look at ele, wether or not it needs a buff doesnt matter rn, its mechanics are rather complicated compared to other classes (all those attunements and even mixes with weaver are rly scaring alot of player off, and also increase the difficulty of mastering the class)

This will always lead to less ppl playing ele than other classes imo.

But would i change the design of the class to make it less complicated? No, because there has to be a class thats more complicated than others.

 

Would you change the design of ele because of that?

[Spartacus.3192]

> @"RedShark.9548" said:

> Well if we just look at ele, wether or not it needs a buff doesnt matter rn, its mechanics are rather complicated compared to other classes (all those attunements and even mixes with weaver are rly scaring alot of player off, and also increase the difficulty of mastering the class)

> This will always lead to less ppl playing ele than other classes imo.

> But would i change the design of the class to make it less complicated? No, because there has to be a class thats more complicated than others.

>

> Would you change the design of ele because of that?

 

If something is difficult to master by a majority of players doesnt that make it high skill floor? In which case should those combos be buffed so that the person who can execute those combos/rotations/whatever is rewarded for it? Not make them easier but make it higher reward for the higher skill needed? Just thinking aloud here.

[RedShark.9548]

> @"Spartacus.3192" said:

> > @"RedShark.9548" said:

> > Well if we just look at ele, wether or not it needs a buff doesnt matter rn, its mechanics are rather complicated compared to other classes (all those attunements and even mixes with weaver are rly scaring alot of player off, and also increase the difficulty of mastering the class)

> > This will always lead to less ppl playing ele than other classes imo.

> > But would i change the design of the class to make it less complicated? No, because there has to be a class thats more complicated than others.

> >

> > Would you change the design of ele because of that?

>

> If something is difficult to master by a majority of players doesnt that make it high skill floor? In which case should those combos be buffed so that the person who can execute those combos/rotations/whatever is rewarded for it? Not make them easier but make it higher reward for the higher skill needed? Just thinking aloud here.

 

Like i said, my example was about the complexity in general and how it affects playtime of a class, not looking at the state of it right now.

 

I agree that ele is rather weak right now, and that player that master complex classes should be rewarded for it, but that wasnt rly what i was talking about.

 

I was just talking abouut why you shouldnt balance around playtime of a class.