Guild Wars 2

O_O My eyes! lol

Thanks!

nice, will start working on this asap. thanks mike!

I propose formatting this in multiple ways.
Way one: Ranking based on Overall Score (I think this is how the current rankings are calculated)
Way two: Ranking based on Number of Wins
Way three: Ranking based on what place one came in in a match (Say, three points for first place, two points for second, one point for third)

If we can get all these methods of arranging the info, I think we’ll better be able to judge how well the different servers are doing in multiple ways.

I’ve just finished translating this into readable data.

Technical Babble:
I loaded the EU and NA data into two separate SQL tables, and used my preferred language (PHP) to grab the data and organize it into a per-server per-match array, which contained the server name, the score the server got in the match, and the position (one through three), based on score, that server got in the match.
I iterated over that data to figure out which server had the most/best wins using a simple point system – 1st place = 3 points, 2nd place = 2 points, 3rd place = 1 point, as well as what was the overall score of that server.

And thus we have the data below.
Edit: See my post two posts below; it has updated data.

Um, that isn’t the ranking of the servers though – the only correct one there for Europe is Vizunah Square. Tallying the points between EVERY match up gives no correlation to how well any particular server will perform vs another… Far Shiverpeaks is 22nd there when they’re actually 2nd/3rd with Desolation (who is put as 7th).

The rankings are by how many wins they have in all of the matches, Xonal. I just happened to also put the score of that server there as well.

If you’d like a ranking sorted purely by score rather than wins, then here you go:

Note that the win score is calculated as follows:
1st place in a match = 3 points
2nd place in a match = 2 points
3rd place in a match = 1 point

North America:
#Position Server Name: Overall Score – Win Points

1. Henge of Denravi: 1147004 – 30
2. Eredon Terrace: 1116489 – 23
3. Dragonbrand: 887722 – 20
4. Fort Aspenwood: 880229 – 20
5. Stormbluff Isle: 859446 – 21
6. Sorrow’s Furnace: 854589 – 18
7. Tarnished Coast: 839959 – 22
8. Sea of Sorrows: 824823 – 22
9. Jade Quarry: 819587 – 24
10. Maguuma: 812495 – 21
11. Gate of Madness: 810349 – 23
12. Ehmry Bay: 808505 – 19
13. Crystal Desert: 774556 – 24
14. Darkhaven: 768321 – 19
15. Sanctum of Rall: 710265 – 20
16. Borlis Pass: 683682 – 19
17. Isle of Janthir: 653388 – 20
18. Yak’s Bend: 634147 – 20
19. Ferguson’s Crossing: 604516 – 21
20. Blackgate: 518484 – 18
21. Devona’s Rest: 503058 – 16
22. Northern Shiverpeaks: 500050 – 13
23. Anvil Rock: 469213 – 17
24. Kaineng: 170577 – 10

Europe:
#Position Server Name: Overall Score – Win Points

1. Vizunah Square: 1047117 – 29
2. Riverside: 851117 – 22
3. Arborstone: 840047 – 23
4. Kodash: 826239 – 19
5. Augury Rock: 821430 – 24
6. Fort Ranik: 795677 – 20
7. Aurora Glade: 792951 – 19
8. Gunnar’s Hold: 790290 – 22
9. Abaddon’s Mouth: 772295 – 22
10. Gandara: 768309 – 24
11. Ring of Fire: 760025 – 21
12. Desolation: 756489 – 18
13. Underworld: 750169 – 17
14. Vabbi: 749731 – 19
15. Blacktide: 749717 – 22
16. Jade Sea: 734332 – 18
17. Elona Reach: 727542 – 17
18. Drakkar Lake: 727083 – 18
19. Fissure of Woe: 718217 – 19
20. Seafarer’s Rest: 712685 – 17
21. Piken Square: 701333 – 20
22. Whiteside Ridge: 671981 – 20
23. Baruch Bay: 662832 – 20
24. Ruins of Surmia: 662046 – 18
25. Far Shiverpeaks: 654032 – 22
26. Miller’s Sound: 469223 – 17
27. Dzagonur: 368229 – 13

There was also a minor bug with the win score calculation in my aforementioned initial results, so below is the data sorted by wins correctly, with the win score data displayed as well (same data as above).
Note: It’s important to note that ties of win points are not currently handled by comparing the server score, as they should be.

North America:
#Position Server Name: Overall Score – Win Points

1. Henge of Denravi: 1147004 – 30
2. Crystal Desert: 774556 – 24
3. Jade Quarry: 819587 – 24
4. Gate of Madness: 810349 – 23
5. Eredon Terrace: 1116489 – 23
6. Tarnished Coast: 839959 – 22
7. Sea of Sorrows: 824823 – 22
8. Ferguson’s Crossing: 604516 – 21
9. Stormbluff Isle: 859446 – 21
10. Maguuma: 812495 – 21
11. Yak’s Bend: 634147 – 20
12. Isle of Janthir: 653388 – 20
13. Dragonbrand: 887722 – 20
14. Fort Aspenwood: 880229 – 20
15. Sanctum of Rall: 710265 – 20
16. Ehmry Bay: 808505 – 19
17. Borlis Pass: 683682 – 19
18. Darkhaven: 768321 – 19
19. Blackgate: 518484 – 18
20. Sorrow’s Furnace: 854589 – 18
21. Anvil Rock: 469213 – 17
22. Devona’s Rest: 503058 – 16
23. Northern Shiverpeaks: 500050 – 13
24. Kaineng: 170577 – 10

Europe:
#Position Server Name: Overall Score – Win Points

1. Vizunah Square: 1047117 – 29
2. Gandara: 768309 – 24
3. Augury Rock: 821430 – 24
4. Arborstone: 840047 – 23
5. Far Shiverpeaks: 654032 – 22
6. Gunnar’s Hold: 790290 – 22
7. Blacktide: 749717 – 22
8. Abaddon’s Mouth: 772295 – 22
9. Riverside: 851117 – 22
10. Ring of Fire: 760025 – 21
11. Baruch Bay: 662832 – 20
12. Fort Ranik: 795677 – 20
13. Whiteside Ridge: 671981 – 20
14. Piken Square: 701333 – 20
15. Kodash: 826239 – 19
16. Aurora Glade: 792951 – 19
17. Vabbi: 749731 – 19
18. Fissure of Woe: 718217 – 19
19. Jade Sea: 734332 – 18
20. Desolation: 756489 – 18
21. Drakkar Lake: 727083 – 18
22. Ruins of Surmia: 662046 – 18
23. Elona Reach: 727542 – 17
24. Miller’s Sound: 469223 – 17
25. Seafarer’s Rest: 712685 – 17
26. Underworld: 750169 – 17
27. Dzagonur: 368229 – 13

Nice job man, thanks.

Spreadsheet on google doc:
https://docs.google.com/spreadsheet/ccc?key=0AlZPTfY3loR7dFlrMU9YWnoxQ19QaVlpYWFTc2QyaGc

Easier to read

The one sorted by score is not really comprehensive as my server Far Shiverpeaks is in top 5 according to ANet self based on rankings. However we are constantly matches against Vizunah Square[FR] and Desolation. The french server has a lot of canadians who play alot at night and therefor win almost always as there are always people from them in WvW. FS and Desolation change spot every now and then. So yeah, we do not place high often, but the we play against the best, so we should be somewhere in the top, and not in the bottum of the list.

Your list doesnt weight wins, a win or second place against one of the zergiest 24/7 servers is worth more than a win against a server that sucks.

TimeBomb, your method for ranking those servers is fundamentally flawed. You can’t say 3 points for 1st place in a match up when those match ups range from very top end servers to very bottom end servers… coming 3rd in the absolute top match up should most likely give you more points than coming 1st in the absolute bottom match up. The value of the points can’t be equal across them.

What, Xonal? Are you kidding, right? In league matches a win yields exactly the same number of points no matter who you’re up against. Look at football, but even basket works in the same way, and most of the other sports as well.

Alexious, the WvW system doesn’t work like a football league, you have servers in groups of 3 being laddered against each other… the servers at the bottom of that ladder getting 1st out of their group of 3, bottom end servers, isn’t worth as much as a servers at the top of that ladder getting 1st.

Going by his system servers towards the bottom can end up much higher on his ranking due to their 1sts and 2nds being worth as much, yet servers gettings 2nds and 3rds at the upper end of the ladder, but ending up lower on his ranking would stomp the crap out of them np.

Example:
Gandara he now has as 2nd and Desolation as 20th, Desolation have been matched against them several times now and have owned them sideways and would if we meet again.

The thing is, the servers ARE already ranked in order in the very match ups you’re playing against right now… so if his top 3 servers on his lists aren’t currently against each other then it’s obviously completely wrong.

So until his top 3 EU comes out as:
1. Vizunah Square
2. Far Shiverpeaks
3. Desolation

Then what hes doing is incorrect.

Alexious.6307:

What, Xonal? Are you kidding, right? In league matches a win yields exactly the same number of points no matter who you’re up against. Look at football, but even basket works in the same way, and most of the other sports as well.

This is not a league system. It’s a ranked system where the servers are matched to be about equal in rank. So the servers fighting in the top bracket (even the losing ones) are servers that would crush a server in the bottom bracket, so they are naturally weighted differently. You will never see Henge of Denravi fight one of the bottom servers such as Kaineng, because the matchmaking system won’t allow such an imbalanced fight.

Wow, the complaints came in fast.

The way I see it is this… the “Brackets” weren’t quite FULLY set up by this time. I think the period shown there was still a training period, and the matchups seemed to be decided based on overall score… which is a HORRIBLE way to judge things. I still remember the arguments from one server on how another server shouldn’t be higher than them because that other server never beat them or won a game…

The Point Place system that was instituted here is the best method, in my opinion. It may not be fully accurate at the moment, but after a few matches which would move servers about in the rankings, it would become more accurate to how servers are doing.

The only other way you could rank things would be “You win you go up you lose you go down” but that just wouldn’t work in a 1 on 1 on 1 system like WvW.

Chessrook you are incorrect. That system makes no sense because as mentioned, this is not a league system. You do not get to face every single server in the game. Instead it is based on a tier system. Where the top tier teams face eachother. Stormbluff Isle, for example, has had to face HOD every day for almost a week. Yet we almost beat them yesterday, and are currently catching up to them, and beating Jade Quarry on our first one week match.

In fact, Stormbluff Isle has beaten every single server they have faced that is ranked above them in those rankings except for HOD. Hell, they beat Fergusn’s Crossing, Gate of Madness and Crystal Dessert by 90,000 – 100,000 points a piece when they faced them.

Here is an analogy on why those make no sense. Using Basketball.

Lets say basketball was a tier based system, incorperating the NCAA, but scoring itself using a league based system as those awful rankings are.

The top three teams would be Miami Heat, Oklahoma City Thunder and LA Lakers. The bottom three would be the Bobcats, and two NCAA teams.

If the top tier teams face eachother for a week, and Miami Heat win every single one. Then the Bobcats, who would beat the crap out of the NCAA teams, would be ranked HIGHER then OKC and LA Lakers.

Are we really going to say the Bobcats are better then the Thunder or Lakers? When the Thunder and Lakers could beat any other team in the league except Miami, but have less win points because they don’t get to beat up on the bottom ranked teams, and instead are ALWAYS facing eachother?

SBI has only lost to two teams. HOD, and JQ (and beat JQ two times, including currently). Yet nn the these rankings, they are way below everyone else. Why? because they are ALWAYS facing HOD, always. So now servers they have crushed in the past, are ranked higher b/c they have more “win points”, by beating really bad servers. It makes no sense.

Though his system is completely fair. Even if a higher skilled server gets placed lower from losing. They should win in the lower bracket pushing them back up. This allows servers to face a variety of servers in order to maintain something new. It doesn’t really matter who you put Kaineg vs because they have never won.

Crystal Desert has beaten many servers. Just because it’s not by huge amounts doesn’t make the server bad in any way. That just means we need more higher level competition. Granted Henge is a league above all if the matchmaking was done as they said… “Winners face Winners” Then maybe you’d see more variety and that would make it more fun. In no way should it be weighted by how much more you win by.

Dear Scrimps,

HoD won by more than 12k points. We’re lucky to have that much of a lead because you think you almost beat us.

Sincerely,
Reality

Zyzoxing, I don’t think it’s weighted by how much you win by, it’s weighted by which bracket you are in. The top bracket is weighted more heavily than the bottom bracket, though not that much more than the 2nd bracket as a win in the 2nd bracket seems to count for a bit more than 2nd place in the top bracket (as evidenced by the servers swapping between top/2nd bracket each matchup being placed about equal to the server placing consistently number 2 in the top bracket).

Judging by the actual rankings, which I was unaware of when parsing through the data in this thread, there is definitely some weight – i.e. a specific algorithm – given to brackets and how servers rank up to each other.

There appears to be a specific algorithm that we have not yet been told that goes into how servers are ranked against each other.

I would imagine this algorithm has to do with what places the servers you have been matched up against have gotten in prior matches.

So, once we know what servers your competition has been matched up with prior to the current match, we are able to use this as a moderately solid ranking weight.

Example:
Server A beat Server B and Server C. Assuming everyone has only played one match prior to this, and Server A, B, and C all got first place, then server A would now be higher ranked then B and C.
But Server C would probably still be higher ranked then anyone in the match up of Server D vs E vs F, as those all got third place in the first match.

That’s how I would probably create the ranking algorithm – based on how many wins/loses against how many winners/losers they had prior to their current match.

SBI has only lost to two teams. HOD, and JQ …

According to the results released by ANet, SBI has also lost to Dragonbrand and Eredon Terrace.

You don’t even need that TimeBomb, it’s more straight forward. You can begin by arbitrarily making laddered match ups between all the servers, then 1st place of match up B replaces 3rd place of match up A all the way down through the different match ups, obviously 2nd places remain in their bracket. So even if you were to completely jumble all the match ups right now, within 1-2 weeks of those 24 hour cycles the order of that list would come out the same.

https://docs.google.com/spreadsheet/ccc?key=0AlZPTfY3loR7dFlrMU9YWnoxQ19QaVlpYWFTc2QyaGc#gid=0

Is there one for euro too?

Just to poke a little fun at you Mike, here is a good read!

http://www.troyhunt.com/2012/09/10-lessons-for-uncultured-web-developers.html

Just a little poking fun at the fact that you formatted the dates in a US-only manner, also a pretty decent read if you are interested =)

One thing to take into consideration is server population. For example, IoJ and SoS are ranked very high in ANet’s rankings. However, NA prime these two servers have very little populations and no queues. But during Oceanic and Asian prime time hours they have very good populations.

So naturally, if we go up against other servers NA prime IoJ and SoS don’t do very well but during Oceanic and Asian prime hours IoJ and SoS clear the map. So the scores aren’t the only thing to look at. The time of play of their population should also be looked at.

Example, currently ET is blowing out both IoJ and SoS primarily because they got a 12 hour head start in taking over maps and getting all 3 orbs and then just holding. ET should not go up against IoJ and SoS – they should go up against a similar server that has a strong NA presence and weaker Oceanic and Asian presence. IoJ and SoS should go up against weak NA presence but strong Oceanic and Asian presence.

So its not just about the score but distribution and hour of play for the servers that go up against each other. This data is available by looking at queues and population averages during different parts of the day.

Dethwar.1768:

One thing to take into consideration is server population. For example, IoJ and SoS are ranked very high in ANet’s rankings. However, NA prime these two servers have very little populations and no queues. But during Oceanic and Asian prime time hours they have very good populations.

So naturally, if we go up against other servers NA prime IoJ and SoS don’t do very well but during Oceanic and Asian prime hours IoJ and SoS clear the map. So the scores aren’t the only thing to look at. The time of play of their population should also be looked at.

Example, currently ET is blowing out both IoJ and SoS primarily because they got a 12 hour head start in taking over maps and getting all 3 orbs and then just holding. ET should not go up against IoJ and SoS – they should go up against a similar server that has a strong NA presence and weaker Oceanic and Asian presence. IoJ and SoS should go up against weak NA presence but strong Oceanic and Asian presence.

So its not just about the score but distribution and hour of play for the servers that go up against each other. This data is available by looking at queues and population averages during different parts of the day.

this, i dont think u can just look at scores and wins

need to look at population vs time and take that into consideration as well imo

@Isis Gebnut.4609 : All results are the on the spreadsheet , you just need to select EU Results at the bottom

Just 3 things need to be implemented:

1. Give guilds information. Separate the day into three 8 hour blocks (EU PRIME, NA PRIME, ASIAN/OCEANIC PRIME). Then rank the servers according to queue times during those 3 hour blocks. This gives guilds information on where to move and move away from high populated servers to less populated servers.

2. Don’t penalize guilds for transferring. All of the guilds progress and buff accumulation should go with them – don’t take this away when a guild wants to transfer from a high populated server in their Prime Time playing zone to another less populated server.

3. Match servers that go up against each other according to their prime time 8 hour playtime clock. Pit the Oceanics/Asian strong pop servers that have weak NA Prime pops against each other. Don’t just use point totals to tally up the score – take population balances and server profile into consideration.

I believe with those 3 fixes – we should see a smoother time in WvW. Also, people need to have patience – the smoothing out process takes time so hang in there, but give the guilds information and don’t penalize them when they try to move out of high pop servers.

the best system and balanced games is based on the ranking:
- Part A: ranking 1st against 2nd and 3rd.
- Part B: ranking 4th against 5th and 6th.
and more …

Won match you up on A, you lost macth down on match B.
After 9 weeks everyone is met and grading is finished and will be realistic.

It’s nonsense to rank these servers based on total score or win-loss point score. We have official rankings. Stop complaining about the “ranking” from some random person based on flawed ranking system.

To anyone speculating on how the rankings are currently done, I don’t know how you can look at https://forum-en.gw2archive.eu/forum/pvp/wuvwuv/Current-world-rankings and not realise that its ELO they are using.

Dethwar, I agree with points 1 and 2 but I don’t think 3 is practical.

It is not ELO but a modified version of Glicko

Thorvald.5432:

It is not ELO but a modified version of Glicko

Which is itself a slightly modified ELO in the first place.

Thorvald.5432”

It is not ELO but a modified version of Glicko

So do we know if that is Trueskill, some other published Glicko variant or there own system?

Having thought about this for awhile I have come to the conclusion that population within the battle grounds needs to be accounted for.

There are a couple of ways to do this: Average population in all the borderlands + eternal or average population in each individual borderland and one for eternal.

For example, on Tarnished Coast, we do ‘okay’ in Eternal but we lose out on every borderland due to huge population imbalances. This means that the server who is ‘winning’ the war isn’t better as such they simply have more players.

If population was added as a trackable metric then the statistics would gain greater meaning. I don’t think it is reasonable to rank a server as better than another simply because they consistently have more players in the ‘overflow’ borderlands.

Another option, if this cannot be done, would be to weight the fight in Eternal much higher than in the borderlands or, in terms of ranking, to ignore the borderlands entirely.

What I’ve found, unfortunately this is very much limited in terms of exposure since I’ve only played on the one server, is that people fill up Eternal first and then spill over into the borderlands – usually our own borderland first. I don’t know if this holds true on other servers but if it does then Eternal should have a much greater impact to the overall score (for the purposes of ranking only) than the borderlands.

If you use ELO (or ELO in some modified form like Glicko) then population will be accounted for in a way. The servers will be weighted by how difficult they are to fight against, so the servers around your skill level will have similar day/night shifts to you. People with a large 24/7 population will be harder to beat and thus will have a higher ranking than those with low populations at night.

So desolation the 3rd EU best is down to almost last?
Arenanet logic?

Lol at the last list. Have absolutely nothing to do with reality. Arena posted results are 100x time more accurate in terms of server power than what you sorted.

Dethwar.1768:

One thing to take into consideration is server population. For example, IoJ and SoS are ranked very high in ANet’s rankings. However, NA prime these two servers have very little populations and no queues. But during Oceanic and Asian prime time hours they have very good populations.

So naturally, if we go up against other servers NA prime IoJ and SoS don’t do very well but during Oceanic and Asian prime hours IoJ and SoS clear the map. So the scores aren’t the only thing to look at. The time of play of their population should also be looked at.

Example, currently ET is blowing out both IoJ and SoS primarily because they got a 12 hour head start in taking over maps and getting all 3 orbs and then just holding. ET should not go up against IoJ and SoS – they should go up against a similar server that has a strong NA presence and weaker Oceanic and Asian presence. IoJ and SoS should go up against weak NA presence but strong Oceanic and Asian presence.

So its not just about the score but distribution and hour of play for the servers that go up against each other. This data is available by looking at queues and population averages during different parts of the day.

This. Very well said and something that I believe would help to cure some of the frustration surrounding WvW in it’s current state.

The basic fundamental flaw of WvW game type design is that people get discouraged when their server is losing. Coupled with free server transfer, people will transfer to winning servers, magnifying this flaw of population imbalance. The current outmanned buff is not enough to entice people from losing servers to play against odds. And most losing servers have no real chance of coming back in scoring, so there’s no exciting comeback fights. So after a few months, people will care less about WvW and the match results will mean nothing.

Any cumulative analysis of these numbers have no value. Free and almost instant transfers of accounts from one server to another are allowing large numbers move between servers skewing the intended classification each server score reflects.

Hi Mike, could you please post the rest of the matchups for dates 9/11 to 9/13. I believe the one week matchups started on 9/14. Thank you.