Guild Wars 2

Hi!
Actually i write in the german Board, but i like to share these informations i summarized.
(sry for my bad english below)

First of all there is a special note:
The basic of all my Data is founded on 1821 measurements. So this ist just a sample of the entirety and not precise. Large deviations are possible.

ANet never will give us some statistics like this, so i wrote a program witch reads my screen, realizes the informations and puts it into a database. I did this because of interest in changes.

I refer to the attachment image on the right while pointing to some interesting things:

You can see there are 2 areas at lvl 1 to 10. Obviously many people like to start classes of area A. Less from area B.
Through the lvls classifications change.
At lvl 30 and higher I subdivide classes to 3 subparts. Mostly played classes are in area 1 (guardian, mesmer). Seemingly mesmers are underrated.
For area 2 is to say, that necromancers are doing well – especially engineers. On the other hand thiefs and warriors are less favoured in higher lvls.
Noticeable is that ranger is the least seen class in high lvl and falls to area 3

Interpretation is not my strength and all this depends on many factors.
Pls keep disskusions objective so this thread wont be closed and i can give you more statatistics soon. (Excuse my bad english by sending corrections via PM xD)

/e: there’s a little mistake in the right graph on the bottom axis. It should be lvl 36+ and not 30+

Wow, just wow. This is awesome, I love looking at hard data like this. Thank you so much for putting this together and sharing it with the community; you have my eternal gratitude.

Is the vertical axis on the right hand chart the same as on the left hand chart (percentage of total at that level)?

Were you able to remove multiples of the same player/character combination?

If you were able to record this kind of information, how did you handle same player but with different characters?

This is amazing good job man. Now we have data on paper. I’m sure the devs would love to take a look at this.

Allarius.5670:

Is the vertical axis on the right hand chart the same as on the left hand chart (percentage of total at that level)?

No. On the right graph you can see the balance of classes normalized to 100% at each lvl range. I can give an concrete example with numbers if its not understood.

Were you able to remove multiples of the same player/character combination?
If you were able to record this kind of information, how did you handle same player but with different characters?

I assume the probability of that is too low, because of so many people playing tPvP. It seems I never met the same charakter name and i never counted the own random team twice .Same player but with different characters wont have a significant effect on the statatistic.

Why is it always the people that excuse themselves for their bad english who then end up talking it more than fine.

This is really cool. Thank you.

This is really awesome. Good job!

Nice job man! this chart can be used to classify tiers:
A-Tier: Guardian.
B-Tier: Thief, Mesmer.
C-Tier: Warrior, Elementalist, Necromancer, Engineer.
D-Tier: Rangers.

Nice work dude. This seems very helpful and generally accurate

@ Pride N Greed

The Devs don’t need this information because they already have it in spades and with, I can guarantee you, 100% accuracy. In this day and age MMO companies have metrics on their metrics. Hell, TOR’s metrics were so in depth they could tell you what % of people went left in an L/R T junction in any span of time.

Everything is recorded.

Everything.

Enjoyed this data thoroughly. Thanks for putting it together, OP.

great stats. were your numbers pulled from Tournies or Hotjoin or both?

Really nice man, really nice. Thanks for sharing the data.

Furbs.9054:

were your numbers pulled from Tournies or Hotjoin or both?

Only from free tournament pvp matches

Cool idea and good post man! I just have some problems relying on the facts you put down. And what does level have to do with anything really? Matters nothing in PvP.
And also I dont believe in the numbers either. I excludingly play tPvP with my team and we almost Never ever see dual guardians in any tournaments. It is incredibly more normal with dual mesmers, thiefs or necromancers.

Granted I have not collected any data on this, it is just my impression. But still, about half a thousand tournaments should give you a decent idea of what is most oftenly played.
The only reason I can think would be that almost all teams go with a guardian, but then you almost never see one go without a mesmer for the trebuchet or a thief for roaming or elementalist/enginner for support bunker. The only class that is at all less represented than the rest is ranger, and that is not very strange, due to them not bringing anything to the table. They can’t do anything no one else can, and excels at nothing.

Really nice post. The numbers match my impression.

SameHH.1048:

Nice job man! this chart can be used to classify tiers:
A-Tier: Guardian.
B-Tier: Thief, Mesmer.
C-Tier: Warrior, Elementalist, Necromancer, Engineer.
D-Tier: Rangers.

How do you get your Tiers from that data? How did you jump Thief above Engineer and Necro when experienced players drop Thieves so heavily below?

A) Guardians and Mesmers
B) Engis and Necros
C) Thieves, Warriors, Eles
D) Rangers

Unless you’re going by the initial data, which shouldn’t mean anything, I would much moreso go on what a level 30+ player tells me than a level 1.

correction: lvl 36+
there’s a note in the first post about this mistake, sry.

Very nice, thanks for the data, I’m surprised rangers aren’t loved.

Rangers are in a bad spot right now. They have all the same problems as Eles, and unlike Eles they don’t have a single role they greatly succeed in. The only thing they’re decent at is shortbow condition damage, and it isn’t all that impressive compared to other classes.

In response to Rythgar.2896 ( why is quoting such a pita with this forum software ?! ):

The ratio of mid and high pvp levels within a class is a pretty good indicator for whether good players stick with the class. The distribution of levels within a class is actually the useful portion of the data in fact.

Let’s have a look at the Ranger – about 50% of the observed players are PvP level 0-10 and roughly 75% are lvl 15 or below. Of the observed Engineers otoh more than 50% are lvl 20+ and less than 10% are lvl 1-5, by far the smallest bracket.

Interpreting data is always much more difficult than collecting it though. Going on more than 10 years of MMO experience and a job dealing with that kind of stuff here is my take:

Rangers are very popular for style and their excellent PvE performance, so you have a lot of them trying tPvP. Their players either don’t stick with the class or don’t like the game mode though. Combined with the data on classes played released by ANet there are two likely conclusions:

a) the Ranger class is not competitive
b) the Ranger class attracts more casual players than other classes and they just don’t like the very competitive nature of tPvP

Probably both apply, no way to tell how much of the variation either option explains from the data supplied.

Engineers are not very popular with the silent majority ( aka casual and PvE players ) because they require more attention and effort during leveling. People taking them to higher PvE-levels and sticking with them in tPvP are much more likely to be well informed and competitive players – indicating the class performs better at that particular part of the game.

In my experience class distribution varies a lot depending on the origin of the server’s players though. Germans, French and Russians will on average look more for group contribution , while Americans and Asians value 1vs1 performance more. A limited sample like the OP’s is therefore to be taken with a grain ( or two ) of salt.

Time-period for the data? Can’t really make any sense of the data, if I don’t know exactly how they were collected :-S
How many of the players are the same? Could be some very active players playing their favorite class a lot, and the weight of that player need to be taken into consideration.

But it is nice to have some hard numbers, even if they are not as revealing as I would like to.

I get it right if you are especially sceptical about hard numbers.

There is no need to be considerate of active players “atm”( <- implied criticism for other topics) because probability of that is too low and take no significant effect on the statatistic.

time period is given in the headline of the image. pls read my op and take care of my special notes in it before any peek onto my grafics.

for more questions pls first read the posts above. Some questions could already been answered. If not, i’ll continue answering any question :) (can i say so grammatically oô)

blurps.2340:

Combined with the data on classes played released by ANet …

I didnt think about to compare it with that. Thx for your well thought out post.
The image you told about is below.

Obviously there is a correlation to the lower ranks in the stats i gave.

/e: sry. but as op i take the insolence of dbl posting, cause of the incompetence of the board to edit my posts.

This is how you should approach the work on the balancing, and this is kind of data to base your decision on. I hope devs have the tools to gather this kind of information. For the thread starter, applause.

Very interesting, could you post some info on the sample size used to collect this data?

PotatoOverdose.6583:

… could you post some info on the sample size used to collect this data?

As i wrote in the op: stats based on 1821 players in tpvp matches.
In this time duration about a week I collected 209 matches. Yes, 209 × 10 = 2090, but mind that i never counted the own team twice (that does my smart script automatically) an some player left the team before the match ends, so i cutnt record them. If you wanna have hard numbers i can send it to you via pm.
I assumed this as a small sample of the whole so I cut do some approximate exceptions like same players i told about in prior posts.

Therefore I recite blurps.2340:
“A limited sample like the OP’s is therefore to be taken with a grain ( or two ) of salt.”

This is why I love the gaming community. We are the most passionate, smart, and dedicated fans out there. Making a game of this magnitude must be exciting and terrifying at the same time.

This is awesome, please do more!

My main is a Ranger and now i play Guardian and Thieve for PVP.

lol

Arenanet need to fix this class, they have traits/utilitys/elites/weapons that are bad.

The only build that its okey its a shortbow/trap condition build, and necros do better than us, so… why you should bring a ranger to the group?

BTW, awesome work !

This should never have been pushed off page 1. Great work, thank you. Interesting to see high rank players ditching thieves and going for mesmers.

Why did you even need such data to tell you how games are? I could tell you myself
most played classes on tournament finals (not counting r1 and r2 which don’t count – unless its paid) are guard and mesmer, obviously fotm, obviously broken.
Ranger is leeast played not because its bad but because its hard compared to other classes, and it sucks with short or longbow. Warrior has some people playing it because its a fun class, id say right now they are the bottom of the food chain and not ranger. A condition axe/torch ranger kitten so many

Great graph, good job that man.

Shows what I thought though regarding warriors. You get alot of lowbis, who just want to try the HB+charge+frenzy combo, but when they realise that the warrior is actually pretty fragile, and getting that combo off is actually very hard nowadays as most decent players are prepared for it, they just go to the easier mode classes (guardian, mesmer…)

Rangers are a good class, just pointless to have in pvp as they bring nothing to the table.

@Argo: Doh … sorry for making you repeat yourself. And the data is still great even if 1821 should be taken with a grain of salt.

One thing though (which I can’t seem to find, maybe I am blind); it could have been interesting to see at what level in the tournament you met the various professions for obvious reasons.

More than anything, I guess the data reflect peoples perception. But let us not go there … .

Thieves are quite interesting above lvl 36, and the “ele, necro, engi-tier” as well.

^That’s exactly what that means.

as a rank 41 ranger i feel honored to be a rare breed

Could you explain more about this program and how you created it? It is very interesting to see this come around.

I just wanted to point out that your data means nothing. Rank is not tied to the Prof. So you could rank up playing nothing but one Prof and than switch to another. You can not tie Ranks and Prof together. Since the fastest way of ranking up is farming spvp and some Prof are better at farming than others it makes your data meaningless.

to bring your opinion is highly esteemed. thx for that

to some of you i answered in a PM because its not relevant for the topic i think

here is the distribution of classes by tpvp-rounds.

special note about validity is the same as in the op. All the data may in principle not be seen as global condition. Its limited in time period and changing by the shift of opinion and minds of players, even by the wether.

check out my older statistics from the german board, like this:
https://d1ej19d7kwigfi.cloudfront.net/uploads/forum_attachment/file/1408/klassenverteilung_tpvp_20121030.png

btw this time i used another background. As yet im not sure about this :S give your opinion

keep the thread clean and objective in the same way as before, thx :)

Impressive effort to capture data and to present it. Thanks a lot for that.

PS: I prefer black background, I find it easier to read.

Here is my new statistic about the distribution of classes during the last 2 weeks.

First of all the special note:
The basic of all my Data is founded on 1068 measurements. So this ist just a sample of the entirety and not precise. Large deviations are possible. So it has to be taken with a grain of salt.

So what has changed and what is noteworthy:
Principally you can see that classes are counterbalanced more than before.
Particularly higher rank guardians are much less seen at free tournament PvP. As you can see on the chart warriors are on the rise. On the other Hand there are fewer people playing a engineer at the moment, but rangers grow about 2 percent and overtake engineers.
Elementalists and Thiefs enjoy increasing popularity by experienced players. Nearly nothing changed to the class necromancer apart from less players in lower ranks.

Very nice work. I feel hipster maining an engineer again.

will be interested to see if you notice big changes post-patch

“In my experience class distribution varies a lot depending on the origin of the server’s players though. Germans, French and Russians will on average look more for group contribution , while Americans and Asians value 1vs1 performance more. A limited sample like the OP’s is therefore to be taken with a grain ( or two ) of salt.”

I wanted to qoute that, I used to play on normal NA hours but have been working night shift the last month… What I have noticed during Euro hours is not as many Thiefs, Mesmers and Eles but more Gaurdians Necros and Wars. The numbers of Engineers and Rangers seem the same. I play a Necro so really see the boost in necro numbers during Euro hours… never see them in NA hours.

This data makes perfect sense to me and I think its very close to what a total global population might be.

Wow amazing work man

This is pretty helpful to see trends

Sry, but that was the last statistic made by me.
I dont play this game anymore^^ The Topic can be closed.
It seens there is not realy big interest to balance the classes. So there wont be big changes of the distribution.
Maybe i ll make a new chart in 6 month. See ya

<3 thx for reading my thread
any questions? send it via PM

Worth a bump because of all the effort Argo put into it.

Hope he comes back to the game if things get fixed down the road.

Ok, the forum is screwing up on me, and I can’t reload this topic since the first time I read it, so sorry if this has been posted already.

I think this is a fantastic post, and we really need more statistical analysis. However people should be careful about such analysis, as sometimes hard data gives people a sense of certainty, but often thing are much more complicated, and most people are not good at reading data.

On that note, there was one major question about the data that interested me, I only got to read the first graph due to the forum bugging out, but I noticed several sharp turns towards the extreme rank 36+. My suspicion is that this is not in fact a result of a sharp change around rank 30 or so, but rather a smaller sample size. With fewer samples in that data range the difference of only a small number of people could be statistically insignificant, but still cause a notable change in the trend lines. As such I’m curious what the same sizes are at different ranks.

Sorry if this question has already been answered, as I said the forum is bugging out on me, and I am really interested in this topic. Hopefully it will be working when I return to check for replies.

@Argo: Thanks for the effort you have put in this.
Thanks for posting the data I suggested … it was an interesting read as well.

And I am sad to see you go … but I share your pov on that point, so I can’t even try to convince you to do otherwise. I hope we both get a reason to come back … the game certainly has the potential.