Guild Wars 2

Feel free to correct me cause this is just based on what I have noticed and read from wiki.

I’ve been wondering why glicko over elo. This is the answer according to wiki:

• Glicko was chosen over its main alternative, Elo. Like Elo, Glicko tracks MMR for each player and updates that rating over time as you play the game. Glicko’s main improvement over its predecessor is the inclusion of a ratings deviation (RD), which measures the reliability of the rating. By using RD, the matchmaking algorithm can compensate for players it has little or incomplete information about.

So basically this system is fair for new players. This is how matches are currently calculated in anets system:

• Ladder weight=“5“ – If present, each team’s average ladder is used for the calculation.
• Rank weight=“1“ – If present, each team’s average rank is used for the calculation.
• Rating weight=“0“ – If present, each team’s average effective rating (i.e. rating – deviation) is used for the calculation.

• Roster weight=“2“ - If present, each team’s maximum roster size is used for the calculation.

As you can see the rating weight is 0 so elo rating is disabled.

In elo you start at certain point for example rating 1500 and based on win you gain rating and based on lose you lose rating and it eventually balances itself to the correct mmr of a player when players play more games.

Now what anets glicko code does is trying to predict every match and trying to make it always balanced. This results the system being broken due the possibility to exploit the system with division ranks. At the moment anets glicko code doesn’t take the players individual skill level and match them against each other. The thing it tries to do is to balance the games between players in a way that teams would always be balanced not players. For example 1 amber, 1 emerald, 1 legend and 2 ruby players would face a team between ruby and sapphire.

Then glicko has it’s own mmr system that predicts the odds of winning. In this kind of team it would be 50/50 cause the teams are balanced according the system. When mmr comes in it wants to balance players ranks as close as possible or to players that are available at time. Some that have low mmr in legend queue with diamonds sometimes. When the system tries to balance players according to their mmr also it makes division system to match teams that have 5 diamond players against 2 legends + 3 diamond players. The odds of winning are calculated from divisions so in this case if 5 diamond player team would win it would mean that they can get more than 1 pips from winning the match. This feature was just forgotten and it’s a good indication of how anet thinks for their players. Basically anet saying kitten you to players. I guess they lost their creativity somewhere between the game and gemstore

So the main problem here seems to be that the game is not balanced around players individual skills. The solution could be to have the players individual rating calculated and match players with similar ratings against each other instead of team balancing.

(yeah it can be confusing to see players from different divisions at first when you queue your first matches and people are already in different divisions when your rating hasn’t been set but trust me, a blog post about this system would explain and the system would be improved much. The rating will balance games and divisions by itself after a while)

The thing that should also be remembered is to reset that rating when the next season begins so that in the next season the players can climb to their own ratings and divisions and not get stuck in low division because they already have high rating and they get legend matches in amber.

It seems to be broken. Exhibit A:

Evan Lesh.3295:

The Ace.9105:

Snip

Snip Snip.

This would explain that team match making is nothing short of being a total nightmare.

The question is, why bother with team queuing? At this point it causes more problems than it helps.

One time I queued with a player that was low rank (he wanted advice on how he can improve his game) and we won every match with flying colors, one match we even got two pips. Yet, if I soloQ matches are significantly different. I can completely understand why players take advantage of this, you can go from 50-60% win rate to 100% just by having the right ally.

Ironically I can tell the match making is trying to work, pitting an emerald-rank warrior and a diamond tempest against rubies. I’d have to be carrying the warrior through 2-tiers, but the thing is, the skill difference between some tiers really isn’t that large. In the end it doesn’t matter, as long as my team pulls some weight we were able to win matches with much ease. Even won matches that “favored” against us and gave 2 pips. I stopped playing rank after that, I cannot take PvP with any amount of seriousness anymore.

It also makes me not appreciate anyone who has legendary rank.

jessiestiles.9437:

It seems to be broken. Exhibit A:

Ooo, is it exhibit time?

B:

Ross Biddle.2367:

jessiestiles.9437:

It seems to be broken. Exhibit A:

Ooo, is it exhibit time?

B:

That’s working as intended. It’s called abusing match making.

Nova Stiker.8396:

Ross Biddle.2367:

jessiestiles.9437:

It seems to be broken. Exhibit A:

Ooo, is it exhibit time?

B:

That’s working as intended. It’s called abusing match making.

lol

jessiestiles.9437:

It seems to be broken. Exhibit A:

there is nothing wrong with this.

Did the system forgot to take into account the weight of a premade? It is balanced, division wise but the rest, doesn’t seem so. Lost 2 pips if memory doesn’t fail me.

Exhibit C:

Ross Biddle.2367:

jessiestiles.9437:

It seems to be broken. Exhibit A:

Ooo, is it exhibit time?

B:

I would’ve had kittens on the floor (yes I spelled kittens). My language would’ve broken Google translate

Deimos Tel Arin.7391:

jessiestiles.9437:

It seems to be broken. Exhibit A:

there is nothing wrong with this.

Look again. It’s a ruby team against a saphire team.

Evan Lesh.3295:

The Ace.9105:

Snip

I’m not entirely following some of what you said, but I will say some things back to see if it helps:

Solo players each have a Glicko rating and a ladder position. Both are updated at the end of a match based on the results and values of the other team. The Glicko algorithm does not care about fair matches, it merely takes results of matches and updates ratings according the the outcome. As for Glicko vs Elo, they perform the exact same function for our purposes, but Glicko is more accurate.

Is the ladder position your divisions and number of pips?

Nova Stiker.8396:

Ironically I can tell the match making is trying to work, pitting an emerald-rank warrior and a diamond tempest against rubies.

I honestly think the fact that it does emerald+diamond vs rubies is actually what makes running with a low rank lead to more wins.

From what I can tell it handles parties by simply averaging the ladder position and probably their MMR rating too, and then pitting them against a team that has that average.
The problem here is that if the two partied players have a huge difference in rating and ladder:
I’ll give a example: a Diamond and Amber duo queue.
The Diamond and Amber average to tier 3 or sapphire. The matchmaker dutifully pits them against a team of sapphires. However there is a large skill gap between diamonds and sapphires which means that the Diamond player will probably be able to 1v1 anyone on the sapphire team. Thus the sapphires need to +1 the diamond to counter him, but in doing so leave themselves +1’ed elsewhere, and even that Amber player can win a +1.

Based on anecdotal evidence I think it averages MMR too, judging by the matchups I’ve gotten in unranked when queuing with some no-so-skilled friends.

Evan Lesh.3295:

The Ace.9105:

Snip

we have our matchmaker. attempts to make fair matches.

Attempts

Evan Lesh.3295:

As for what changes would most improve leagues, you’ll have to wait for more info on that.

Ugh. terrific.

Pinkunicorn of Dethecus.3217:

Evan Lesh.3295:

As for what changes would most improve leagues, you’ll have to wait for more info on that.

Ugh. terrific.

even they don’t know yet how to fix the screwed matchmaking XDDDDD

Aidenwolf.5964:

Ross Biddle.2367:

jessiestiles.9437:

It seems to be broken. Exhibit A:

Ooo, is it exhibit time?

B:

I would’ve had kittens on the floor (yes I spelled kittens). My language would’ve broken Google translate

I had a few words through PM with a buddy XD

Two of my team didnt even score. I lead in score and was the only one to get downs/kills, saving home at the start, and forcing the bunker chrono off mid etc etc. Though props to our ele who kept mid decaped forever. Still, the rest of the team were rolled over, out-bunked, and 1/2’d for the duration of the match. They had no hope.

Evan Lesh.3295:

As for what changes would most improve leagues, you’ll have to wait for more info on that.

Can we at least get someone (you or someone else) to explicitly state that something is in the works to address situations such as these? All I read from your post was ’we’ve got systems in place and matches are made from that. Working as intended. At the end there’s just this little bit of non-information suggesting that changes will happen but nothing even slightly specific about the changes. They have something to do with leagues. We know that. That’s all we know.

Please realize that we have no faith that such changes will change anything. It would be an incredibly good idea for the PvP team at Anet to give at least an idea of what will be changed. At the moment, when I hear ’you’ll have to wait to hear’ I’m thinking ‘there is nothing currently in development but we’re gonna get to it soon.’

Evan Lesh.3295:

As for what changes would most improve leagues, you’ll have to wait for more info on that.

I hope that you and your team are able to workout the kinks. In the past, not just in pvp but in all parts of the game, Anet’s had this tendency to take a chance and push something new or bold and then not really followup to polish or workout kinks. The reason I say this is, that despite all of the rage fueled feedback on the leagues, its not the concept but rather some issues with the implementation of the pvp leagues and it’d be unfortunate if you weren’t to iterate another season or two until things clicked. Granted there are other items outside the realm of the league sysetm and matchmaking that are having a negative impact as well and hopefully a compromise can be reached to address at least some of these issues (some egs: pay-to-win level power creep in elite vs core specs, pvp/pve skill coupling which now has to interact with tighter tuned content in raids, soloq vs teamq, etc). This season felt more like a beta than a finished product and that’s not inherently a bad thing; instead whether its a bad thing or not will ultimately be determined by the follow-up and polishing (or lack of) in the upcoming season(s).

imo, the fundamental component that leagues got right was implementing a way to communicate player progression without going against the Anet’s philosophy of refusing to show player MMR. Players have complained about the grind of the progression system, but I think if we look at the broader picture that grind coupled with the progression meter (i.e. pips) and a bit of a reward (back skin, new stuff to craft) all combined to get new players (or vet players new to pvp) into pvp and stick long enough to perhaps become regulars in this content. More players in the pvp pool that will naturally distribute along the skill spectrum can only help make your job of match making easier and my time as a player more enjoyable with more frequently landing in those close matches. It’s a win-win, and the right direction to steer the design toward.

TLDR: the concept of leagues is solid; implementation needs polish; please don’t neglect/abandon/discard this system.

jessiestiles.9437:

It seems to be broken. snip:

This could actually be working properly. All you know is that there are Ruby players and a sapphire player, but that doesn’t tell you where in ruby those players are. If they had just crossed from sapphire into ruby than its entirely possible they are still being placed against sapphire players. Like a high Amber going against low Emerald player.

Evan, a couple of suggestions:

• Please adjust the system so that matches like the one in my pic below are not allowed. If for some reason you decide that matches such as this are ok, then please make it so that the losing team does not lose pips.

• Please revamp the league so that ladder divisions more closely represent player MMR. Currently the ladder is only about grind and anyone can reach legend as long as they spam enough games, even if they are utterly mediocre players. In the current system, reaching legend division only represents a player’s ability to invest time in the game, whereas I would like it to reflect player skill.

• Please add a visible match making rating that can be viewed in-game.

Thank you.

Kycoo Grim.9463:

jessiestiles.9437:

It seems to be broken. snip:

This could actually be working properly. All you know is that there are Ruby players and a sapphire player, but that doesn’t tell you where in ruby those players are. If they had just crossed from sapphire into ruby than its entirely possible they are still being placed against sapphire players. Like a high Amber going against low Emerald player.

At mid saphire I started playing more rubys. One game I was the only saphire in an all ruby game. While your idea has a certain rationality to it I find it highly unlikely.

I probably don’t know enough about matchmaking systems to comment, but I do feel that greater weight should be put on player skill rather than divisions when forming teams. I’m currently at the upper end of Sapphire division, and I can guarantee you that were the divisions based on skill, I totally should not be there. I’m an Emerald player, at best.

The thing is, because you can ascend divisions purely through luck or teaming up with a much better player, divisions aren’t a true indication of skill. There are thus people in Sapphire, or maybe even Ruby and Diamond, which shouldn’t really be there.

Evan Lesh.3295:

Solo players each have a Glicko rating and a ladder position. Both are updated at the end of a match based on the results and values of the other team. .

This explains so much. I always wondered why a team would absolutly crush us, then let us come back a lot and then beat us. Because it gets updated based on results you simply let it be a close win, so your MMR doesn’t skyrocket and you can keep beating lower MMR players easy mode. Is this true?

Evan Lesh.3295:

The Ace.9105:

Snip

I’m not entirely following some of what you said, but I will say some things back to see if it helps:

Solo players each have a Glicko rating and a ladder position. Both are updated at the end of a match based on the results and values of the other team. The Glicko algorithm does not care about fair matches, it merely takes results of matches and updates ratings according the the outcome. As for Glicko vs Elo, they perform the exact same function for our purposes, but Glicko is more accurate.

Aside from this, we have our matchmaker. This is not part of the actual Glicko algorithm, but instead uses Glicko ratings along with ladder position and other metrics in a way that attempts to make fair matches. The way it does this can be seen on the wiki. While each team average is attempted to be balanced, we still don’t want the spread on each team to be large either.

The last part is the prediction. This is not actually part of matchmaking, but is part of our ladder formula. It is a very simplified approximation, meant for players to understand, that we use after the matchmaker has created a match with the most similar players it could find.

As for what changes would most improve leagues, you’ll have to wait for more info on that.

Okay I think I understand now. This league system is a visual indication of the prediction and ladder. Also you said you don’t want the spread to be too large but it’s kinda needed in a way. What I think is the true problem here is the scoring part of matchmaking. From wiki:

• During this phase each player is scored against every other player being considered for matchmaking. The metrics used during this phase include: rating, rank, party size, profession, ladder position, and dishonor. With each metric the system is looking for players that are as close as possible to the average of those already selected. The system also attempts to keep the number of duplicate professions to a minimum.

I have a feeling that there are too many factors in this. 6 different factors are making the match including ladder and mmr also so a total of 8 measurements makes our every match.

This system seems to be designed for soloq players but it malfunctions heavily in teamq when the system still calculates matches based on each individual player and their division. In leagues now we get matched against and with players in a party (2-5) more than we should.

What comes to the match manipulation and carrying lower tier players to higher tiers for in-game gold or just for fun, it could be prevented by placing the players in a party to a same level than the highest player in a party. For example if legendary would que with amber player the system would calculate the amber as legend also.

Now I understand why they don’t want to show us our personal MMR. Because then we will have evidence every game of how kitten the matchmaking is. classic.

Evan Lesh.3295:

The Ace.9105:

Snip

As for what changes would most improve leagues, you’ll have to wait for more info on that.

Given that you are unable to go into details – what might be extremely helpful at this stage, would be an explanation of what issues you have identified that need ‘tweaking’ and what ‘exploits’ (eg Teaming with lower rank/AFK/Tanking) you have concerns about and will be addressing.

Ross Biddle.2367:

jessiestiles.9437:

It seems to be broken. Exhibit A:

Ooo, is it exhibit time?

B:

You queued as a warrior, you were already kittened.

Evan Lesh.3295:

The last part is the prediction. This is not actually part of matchmaking, but is part of our ladder formula. It is a very simplified approximation, meant for players to understand, that we use after the matchmaker has created a match with the most similar players it could find.

Evan,

Does the match prediction algorithm use league division as a factor? It seems to me that it does, and that’s where 80% of the problem is.

If it doesn’t how could I have possibly lost 2 pips on a tied match that was genuinely fair and even skill wise?

Aereniel.7356:

Evan, a couple of suggestions:

… snip …

• Please add a visible match making rating that can be viewed in-game.

Thank you.

Please consider that MMR is subjective to with whom you play. In case of people mentoring other players, their MMR would be “underated”. Many great players have bad MMR because they are doing the “pig” work. Showing Noob how to play the game.

I think that MMR should be visible but only for the carrier of it (private). So he can know if he progress in the “social” ladder and adjust accordingly. Putting that number for all to show, would worsen the gap between hardcore and casual way worse than it is.

I can name you many TOP players right now, that have good MMR because some top players “mentored” them for a while. With a visibile public MMR, many would never had done so.

It’s a Myth or a Urban Legend to think that MMR equal skills. MMR equal your social positions in the game. Do you play with good players, winning match versus good players? (or not) ( I know many people who had bad old leaderboard rating that can carry game versus top players )

There is people that can carry a team for a week or two (myself included) to the 100-200 leaderboard positions (old skill based system 2013-2014) without being rated 100-200 consistently (because they try new people out for their guild, mentor some new promising pvp prospect, want to really give a player a chance to prove his worth).

-=-=-=-=-=-
But YES, skill based leaderboard are way better than grind based casual jokes of a competitive game. GRINDING LEAGUE is bad. Actually, it’s easier for low MMR to get to Legendary than high MMR player.

This is the most ironic joke of LEAGUE, right now.
-=-=-=-=-=-

Dal

I’m enjoying this pvp so much that I have to post something here. Was 1 pip away from diamond, I get teamed against classic legendary/amber duo, Blondiine and Wilhelmina Kane, and because i had few games long winstreak and my teammates just got worse and worse, it was basically no match at all. 2 pips lost. Then the next game, same legendary/amber duo against, this time even worse teammates. 1 pip lost. Then third game against this grandmother of abusers guild team, bunker+abuse meta was good as usual and yeah. There it goes, 1 pip lost.

I’m so happy that what once was PvP is now just Player versus Abuse, and Anet doesn’t even care. GG!

Fatallion.5389:

I’m enjoying this pvp so much that I have to post something here. Was 1 pip away from diamond, I get teamed against classic legendary/amber duo, Blondiine and Wilhelmina Kane, and because i had few games long winstreak and my teammates just got worse and worse, it was basically no match at all. 2 pips lost. Then the next game, same legendary/amber duo against, this time even worse teammates. 1 pip lost. Then third game against this grandmother of abusers guild team, bunker+abuse meta was good as usual and yeah. There it goes, 1 pip lost.

I’m so happy that what once was PvP is now just Player versus Abuse, and Anet doesn’t even care. GG!

This is a good example of how the prediction system tries to balance division pips. There it calculates the average from both teams and a team with 3 rubies and 2 diamonds have higher average than 2 legend, diamond, sapphire and amber. Therefor the system calculates your loss as 2 pips because your team average is higher than the enemy teams average.

Evan Lesh.3295:

While each team average is attempted to be balanced, we still don’t want the spread on each team to be large either.

This creates a problem with the prediction system when it comes to teamq. As I said before this system seems to be made for soloq and it doesn’t work as well in teamq environment.

Well at least you guys are getting a mix, i have multiple pics of facing ESL players while im solo queing and i have plenty of facing full premades and 3-4 of them are in legendary division and my team being all solo ques in diamond.

I have no problem with me being matched up against them but i have a problem with losing PiPs when the team i was on couldnt have possibly been picked to win.

If you took a class on “slapping yourself in the face” a forth year class would be difficult for a first year face slapper and their classmates regardless of skill would show an overall higher quality of face slapping. Games become more difficult as the divisions progress and if nothing else they are an indication of experience in the current league.

jessiestiles.9437
“If you took a class on “slapping yourself in the face” a forth year class would be difficult for a first year face slapper and their classmates regardless of skill would show an overall higher quality of face slapping. Games become more difficult as the divisions progress and if nothing else they are an indication of experience in the current league.”

So if I have 25 pips in ruby, solo queue, and 3 times I get the same pre-made team, it means my place is at ruby, not diamond?
Or some solo players don’t want to play, they don’t care about loosing/winning.. or don’t understand the language, they are germans/french, my place is just in ruby, and will be there for ever in this season?!… here is an example: Solo queue, 1 min till game, we all agree in a tactic/strategy, and the game starts, 2 player plays how he wants, like in solo mode… then I yell at them, my temp. gets up high, I say bad stuff in team chat, and then I realize that they are germans… don’t even understand the tactics we all agreed on, but they understand the bad words what I’m writing. So I write bad words, then they say “deine vater ist isis”, “kittenst” and they will personally send e-mail so I get banned. Yes I got 72 hour suspension, and I know they didn’t get any…

The real problem is that the matchmaking is putting us solo queue players in teams who are from german/french servers, and I can’t communicate with them, and I don’t want to, they insult me that they understand the tactics, and they don’t even understand a word… I understand german, can speak german, love german people, but I hate those 2 guys, they are simply low minded, that’s it!

That’s not at all what I said. “(D)ivision(s).. are an indication of experience.” So that once you reach diamond you’ll be playing tougher games and will have more experience than a ruby. It was in response to people suggesting that division doesn’t indicate skill and is therefore irrelevant which it isn’t.

People insult each other regularly in anger. It’s probably your stereotyped nationalistic comments (which is tantamount to racism) that is getting you in trouble. It’s not Germany’s fault you lost.

1) Why am among diamonds?
2) Why again another premade against solo queuers?
3) Why considering the odds the system decided that even scoring 150 wasn’t good enough and we lost a pip?

It seems system decided to punish me due to two victories in a row pushing to the 50% w/l creating such a bad lopsided match.

Hopefully the incoming changes will put more weight and reward accordingly when solo queuers faces premades besides, of course, reduce the amount of the disparity between matches.

Now I remember why I decided to stop with pvp for a while. Played 3 games, faced 2 premades, the first one was the reverse of the pic on the attachament, not fun at all either to be the stomper or the stomped, the second was balanced and the third was the punishment. One good match out of 3 so far.

Jourdelune.7456:

It’s a Myth or a Urban Legend to think that MMR equal skills. MMR equal your social positions in the game. Do you play with good players, winning match versus good players? (or not) ( I know many people who had bad old leaderboard rating that can carry game versus top players )

There is people that can carry a team for a week or two (myself included) to the 100-200 leaderboard positions (old skill based system 2013-2014) without being rated 100-200 consistently (because they try new people out for their guild, mentor some new promising pvp prospect, want to really give a player a chance to prove his worth).

Sample size is the key here, because given enough games, every player’s rating level will balance itself out. Even if a skilled player likes to charity-carry less-skilled guildies from time to time and takes a temporary rating hit because of it, it will only have a negligible impact on his overall MMR in the long run, as long as that player is active and plays the game consistently. It’s a marginal concern with a marginal impact. As for social position, bad players get dropped from good teams as competition tightens up. There are a few notable examples of this from the EU scene.

I’m a casual solo player and currently 12 pips away from legend playing 100% solo and only conquest. Hardly hardcore material. I have zero problems with advocating a visible MMR system, and neither should any competitively-minded PvPer. Only poor players who try to pretend they are better than they truly are would have anything to gain from hiding their rating.

As for your argument that a visible rating would segregate the population and lead to an “upstairs-downstairs” divide, please note that skill differentiation is exactly what a PvP ladder system is all about. Sure, Anet mucked the current iteration up completely and made the ladder about grind (as opposed to skill, as it should have been), but they seem to have come to their senses and will tie the ladder more closely to MMR in season two. Having a visible MMR is a healthy continuation of that development path that will spice up the competition nicely. This game desperately needs some PVP bragging rights, and there is no better motivator than visible rating. No, grindable PvE wings do not count.

I think the lower the total pool of available players, the less effective the glicko system is.

MMR correlates with percentile and ladder points right?

Stand The Wall.6987:

I think the lower the total pool of available players, the less effective the glicko system is.

That’s true for any system really. With game design you can pretty much fix anything by throwing more players at it. Unfortunately you can’t just whisk players out of thin air.

WoW is a perfect example this. In terms of managing players, most of WoW’s systems are horrendously designed, however WoW gets away with this because it has such a huge playerbase that the problems don’t matter.

Stand The Wall.6987:

I think the lower the total pool of available players, the less effective the glicko system is.

The low population is a big one for sure, but GW2’s pvp matchmaking also gets beat up by factors like:
- requiring a ‘functional team comp’ to be competitive
- no one person can carry, but one much weaker or bad player can completely ruin the match
- at low level play matchmaking likely over estimates the value of comms and at mid to high level play it grossly underestimates the advantage players in comms over those that aren’t
- the system can’t determine other build related factors

A larger player population will certainly help, and if they’d get away from letting premades stomp out soloq’ers it’d go a long way to help things, but matchmaking is always going to be pretty shaky due to the nature of the game.

Bringing back more memories from the past !

Evan Lesh.3295:

As for what changes would most improve leagues, you’ll have to wait for more info on that.

We’re 4 seasons in and we still get stuck 4 DH vs 4 druids instead of placing 2 of each on each teams. I mean these things are really basic it shouldn’t take 1 year to fix. Anet said in season 1 that class stacking was something you are looking into. I mean we are having games now where there are 5 dragon hunters on a single team, and none on the other, and everyone was solo queue.

Xillllix.3485:

Evan Lesh.3295:

As for what changes would most improve leagues, you’ll have to wait for more info on that.

We’re 4 seasons in and we still get stuck 4 DH vs 4 druids instead of placing 2 of each on each teams. I mean these things are really basic it shouldn’t take 1 year to fix. Anet said in season 1 that class stacking was something you are looking into. I mean we are having games now where there are 5 dragon hunters on a single team, and none on the other, and everyone was solo queue.

What a ridiculous statement to make, there is nothing “basic” about your expectations. How could Arenanet balance for class stacking and allow players to swap classes at the same time? You can only have one or the other and I’d rather be able to play multiple classes and swap when I feel it is necessary than be confined to one class because of some arbitrary lock. Anyone on either side of the match you described could have swapped classes…doesn’t sound like they did. That doesn’t mean Arenanet needs to change the way a match is setup because you lost or whatever.

I don’t know what but for sure the game doesn’t try to balance the match to be 50/50 at least what I could see so far. The game rather forces you to 50% win ratio which as a result you got: win lose win lose win win lose lose etc end most matches are already decided before they even start.

Evan Lesh.3295:

The Ace.9105:

Snip

I’m not entirely following some of what you said, but I will say some things back to see if it helps:

Solo players each have a Glicko rating and a ladder position. Both are updated at the end of a match based on the results and values of the other team. The Glicko algorithm does not care about fair matches, it merely takes results of matches and updates ratings according the the outcome. As for Glicko vs Elo, they perform the exact same function for our purposes, but Glicko is more accurate.

Aside from this, we have our matchmaker. This is not part of the actual Glicko algorithm, but instead uses Glicko ratings along with ladder position and other metrics in a way that attempts to make fair matches. The way it does this can be seen on the wiki. While each team average is attempted to be balanced, we still don’t want the spread on each team to be large either.

The last part is the prediction. This is not actually part of matchmaking, but is part of our ladder formula. It is a very simplified approximation, meant for players to understand, that we use after the matchmaker has created a match with the most similar players it could find.

As for what changes would most improve leagues, you’ll have to wait for more info on that.

i wonder if you and the pvp team know the difference between:
1) a game and a beta
2) players and betatester

you know we pay for a solid game and we are stuck in a beta for 4 years…
ty

If we could just prevent class stacking that would solve a lot of the imbalance issues we face.

The sad truth is you will never please everyone with the matchmaking because one team always loses and one always wins. I preferred season 2 and 3 because at least the people on your team were at least considerably close to your skill level but I realize it lead to some really horrible stomp matches. This season I am actually managing to hover above the 50% win rate but have seen pretty slow progress because of the poorly implemented pip system.

The bigger issue is people dc’ing/afking and the like. People wouldn’t be as upset if there was a reason to try your best even if you lose but when you lose a pip for losing a match 499-500 or 0-500 what is the point. You are better off losing faster and getting into a new match where you can stomp some fools instead.