Guild Wars 2

Can we actually get some real information on how this works? All the announcement was vague. :/

They told us the “goal” and “purpose” but they just brushed over all the issues that matchmaking has that isn’t just the pool of available players.

Say {insert known pro player here} plays his/her heart out but is super unlucky with how the matchmaking works and goes, in total Ranked games, 0 Wins 100 Losses. We all know he/she is beyond good. Will they be Legendary/Top 250? Or will the game just see “0 wins? Probably kitten. Bottom 10.”

What I am saying is:
* Is this system just another long grind to the top?
* Will it be a fair, “good players will be ranked high as they should, despite their solo [s]W/L ratio”?
* Will the godly PvP players be 100% subject to the luck of the “unknown forced 50%” matchmaking algorithm?
* Is top 250 simply based on W/L luck streaks?

Evan Lesh and others replied. Continued discussions below.

https://www.guildwars2.com/en/news/upcoming-changes-for-pvp-league-season-5/

Better Matches

First, let’s dive into how league matches are changing. Matchmaking rating (MMR) has always been at the core of how we pair players together. In Seasons 1–4, pips were also used in matchmaking, limiting the pool of eligible players to those within 15 pips of you. In Season 5, pips are taking a backseat and will no longer be used for matchmaking purposes. Removing the pip restriction on matchmaking means more players to match with and subsequently shorter queue times. In addition to this, the results of the recent community poll for Solo/Duo queue restrictions means Ranked Arena will only allow a maximum party size of two, ensuring that you never face off against larger, coordinated premade teams. The goal of these changes is to create the best matches possible at all times, which leads into our next area of focus: measuring skill and prestige.

Measuring Skill

Leagues are where players strive to better their gameplay in a competitive environment with key milestones marking their progress along the way. Part of this system should include a metric for players to accurately gauge and compare their skill level. Starting in Season 5, every player will earn a visible skill rating. Skill rating is a numerical value that places players into more traditional divisions: Bronze, Silver, Gold, Platinum, and Legendary. These new divisions will determine which badge is available on your nameplate, and they will even affect the rate at which you earn league rewards. The top 250 players in each region will also be placed onto a new Skill Rating leaderboard where they’ll fight for exclusive titles.

We want Skill Rating to be as accurate as possible, so we’ve added a few new mechanics to help achieve this goal. First, every season will start with a soft reset of the previous season’s rating. Players will then be required to play 10 placement matches before obtaining their starting skill rating for the season. Additionally, inactive players will experience decay that lowers their skill rating over time up to a maximum amount. Decay can be worked off by playing league matches. Unlike the previous pip-based system, there are no restrictions on backward progress with skill rating; you can move down in divisions by either losing matches or by being inactive. This will help ensure that players are always moving toward their correct rating and are facing players of similar skill levels..

I think it is clear enough.

Matchmaking makes you play with and against players of the most similar MMR. If you win, you gain MMR. If you lose, you lose MMR.

Your league placement and ranking is completely dependent on your MMR.

This is basically how things work in pretty much any competitive PvP game out there.

jonemirant.1653:

so is there any more win streak or come-back after loss streak? I rely on those a lot to get my pips as I’m pretty much a 50% average kitten-suck3r.

Pips are no longer tied to rank thankfully you just have to play to get your rewards.

Exciton.8942:

I think it is clear enough.

Matchmaking makes you play with and against players of the most similar MMR. If you win, you gain MMR. If you lose, you lose MMR.

Your league placement and ranking is completely dependent on your MMR.

This is basically how things work in pretty much any competitive PvP game out there.

True, but any fleshed out competitive PvP game has more of a foundation to calculate how much is lost and gained it’s not only about W/L.

GW2 PvP is just a lucky grind, that’s what every season was before it and they haven’t mentioned if they are changing that. In some instances, players have noticed, you could just lose in unranked and go to ranked and win because the system apparently combined all games played.

Plus, no comment on how the “soft reset” works given previous seasons with, on the outside at least, different systems.

And regards to class stacking…

1) There will be the expected periodic balance updates, but I cannot speak to the extend of them.
2) Class stacking has not been a priority because of the leagues rework, and negative feedback regarding character locking when queuing. This might be something worth polling.

Looks like another season of carrying double threat cheese.

WOOHOO!!!! VISIBLE SKILL RATING!!!!!

LUST.7241:

Say {insert known pro player here} plays his/her heart out but is super unlucky with how the matchmaking works and goes, in total Ranked games, 0 Wins 100 Losses. We all know he/she is beyond good. Will they be Legendary/Top 250? Or will the game just see “0 wins? Probably kitten. Bottom 10.”

What I am saying is:

• Is this system just another long grind to the top?
• Will it be a fair, “good players will be ranked high as they should, despite their solo W/L ratio”?
• Will the godly PvP players be 100% subject to the luck of the “unknown forced 50%” matchmaking algorithm?
• Is top 250 simply based on W/L luck streaks?

It seems like you’re only familiar with the terrible pip system of the first four seasons. Forget all about that.

With the new system, rank is directly tied to rating. Rating is tied to performance (win vs. loss in conjunction with your rating vs. the rating of opponents). You can only move up by winning [down by losing]. The higher [lower] you go, the stronger [weaker] your opponents become. Eventually you’ll reach a point where your rating stabilizes because you’re being matched with equally skilled players. Top players should be able to progress higher.

A big problem with the old system, which led to the “luck” feel was the use of pip range in matchmaking. This restricted the matching pool, which in turn caused the rating difference to expand as you sat in queue longer. By the time a game was started, a massive rating difference had developed and winning or losing became a crap-shoot. With a larger pool, the rating difference should be smaller, leading to closer games and more accurate rating calculations.

Summary:

• Pips are only for the reward system now. They no longer impact rank or matching. Pips can only be gained (not lost). Rewards through pips have a grind element.
• Rank/Tier is not a grind like the last 4 seasons. Matchmaking and ranking is done with rating. If a player is misplaced, they’ll win or lose a lot more and their rating will adjust appropriately.
• Tier is simply a name assigned to a range of numerical ratings. You can go up and down in tier.
• There is some decay in rating so that players can’t sit at the top all season.
• There was NEVER a forced 50% in effect. What happens is that when you’re at the correct rating and matched against players of similar rating, you’re expected to win 50% of your games. If your skill improves, you’ll start winning more and moving up.
• Win streaks are possible, even at the top, but aren’t nearly as prone to luck as they were in previous seasons.
• A top player should never have an awful win rate. If they’re losing more than winning, their rating will drop until they’re where they should be.

My only concern is that the solo/duo queue restriction will lead to “solo queue hero” behavior rather than true teamplay – the strong solo player will focus on winning fights rather than controlling nodes and helping the team.

Evan Lesh.3295:

LUST.7241:

What I am saying is:

• Is this system just another long grind to the top?
• Will it be a fair, “good players will be ranked high as they should, despite their solo W/L ratio”?
• Will the godly PvP players be 100% subject to the luck of the “unknown forced 50%” matchmaking algorithm?
• Is top 250 simply based on W/L luck streaks?

• You will absolutely not be able to grind to the top of the Skill Rating leaderboard.
• Good players will be ranked highly. They already are. We’ve had MMR for years and track it constantly.
• No one is being forced into an unknown system. See below and read this: https://wiki.guildwars2.com/wiki/PvP_Matchmaking_Algorithm
• Sometimes players can overshoot their ‘true’ rating by chance, but it will not be by far and not last very long. They certainly won’t reach the top 250.

I don’t mean to pick on you specifically as the “dreaded 50% matchmaking” conspiracy has a lot of the community in its grips. Ratings obviously can’t be known unless matches are played to generate data. Do you want to play random matches and have a chaotic experience with an inaccurate rating? Or have balanced, competitive matches with an accurate rating?

Sure, we could have zero matchmaking, but two things will happen: First, your teammates will have random ratings and your opponents will have random ratings. Did you lose because you are less skilled than your opponents, or because your teammates are less skilled than your opponents? There is no easy way to tell without a lot of complicated math. “Make a system to figure it out, then!” you say. Well, Glicko does this for us already. Matchmaking assumes its data is correct by pairing opponents that appear to be evenly matched, then Glicko does its job to adjust ratings in case the outcome wasn’t what was expected. Second, random matches make your rating update verrry slowly. Every match won against a much less skilled opponent won’t make you go up because of course you were suppose to beat them, your rating must already be accurate. Every match lost against much higher rated opponents won’t make you go down because of course you were suppose to lose, your rating must already be accurate.

I honestly do not know why people think 50/50 matchmaking is a bad thing. My best guess is that people just want want something to blame when they lose.

it was bad in season 1 and 4 because the pip system made very very bad players get to legend and then queuing in ranked was basically worse than queueing in unranked this offseason (which is really bad)

Evan Lesh.3295:

…
https://forum-en.gw2archive.eu/forum/game/pvp/Season-5-MEASURING-SKILL-MATCHMAKING/first#post6422623
…

Thank you for your detailed response.

I’m aware of the Glicko system, but given its previous season usage: You might understand why I’m hesitant to trust its dependability. I’m aware that several other games use it as well, and in those games (take CS:GO) there are modifications to it that make it valuable. Obviously me being Global Elite on CS doesn’t mean the same in GW2.

In the wiki, the modifications are somewhat presented but have they even been adjusted? Has the justification been explained? In a game like GW2, there is way more factors in play than GW2 that determines skill.

If you have played PvP enough you can easily catch onto how the system is working. Players can very accurately determine whether or not their next game is a W and the chance of winning. Now admittedly this was with the Pip system (and Stronghold abuse), but even after hitting Legendary, the “matchmaking” never actually seemed to be any different as players were playing just as good/bad as people who just learned what a keyboard is.

It got to a point where, in non-premade matches, you got to know EVERYONE in your division. I assume, since this isn’t a 1v1 matchmaking system, it’s based on the “team’s average MMR”. Say you know you are evenly matched (again we got to know literally everyone in our division), but the system has us a point off of each other. A close game is had (499-498, clock ran out). The underdog (W) gets a boatload of rating and the L gets slammed? Are close match ups a factor? Is individual player skill a factor?

I know there is A LOT of math behind the system, but like CS:GO this is modified. In CS, I can play great, lose, and still go up because of my performance. I can go on a win streak against better players, play poorly, and drop/go nowhere. The system knows me, the team, and my history.

GW2 hasn’t felt like that in the past seasons. Here all it seems like is: Your individual contribution doesn’t matter. You could have won the game 1v5, and you’ll still get the same gain as you would if you were 5 and the same predictable match up next game.

50/50 matchmaking is a bad thing because of the weight it has on the system. It needs to be adjusted because it’s so painfully obvious that this is the biggest factor in matchmaking right now. Instead of matchmaking based on SKILL, it’s matchmaking on what the outcome alone (was it a W or an L). Its way harder to carry solo/duo in this game than other MULTIPLAYER games that use Glicko. Plus, other games that effectively use Glicko put less priority in W or L.

I accept an evenly matched loss. It’s a super rare treat to have CLOSE games in GW2 Ranked PvP (which is why people talk about those times so highly). The current set up ensure that doesn’t happen frequently. Most competitive players like a CLOSE game (W or L) as EASY games aren’t a challenge and don’t make you a better player.

Now, with the removal of pips and opening up the pool of players to choose from? That, on paper, sounds fantastic…but every season something like that has been said and it fell real short because it was pretty much vaguely presented like it was ‘No Man’s Sky’ or something. I think it’s better not to be so vague and have each Season just be a “test” (not a stab at GW2 PvP only…because, for example, there is a company that named itself after as “a severe snowstorm” and haven’t learned from their competitive changes in their other games).

i see some people have already started to work on excuses why they wont get the rank they think they deserve

Lexander.4579:

i see some people have already started to work on excuses why they wont get the rank they think they deserve

Haha, it is always like this in any competitive PvP games.

please lock class (not build) on Q and use class MMR :-).

So how does the matchmaking work? Are teams formed of overall mmr, so that a team can have high mmr player and very low mmr player and then it produces average and pair it with similar average? Or the teams are formed with only players of similar mmr, so that actual average is quite similar to all players mmr?

Evan Lesh.3295:

I honestly do not know why people think 50/50 matchmaking is a bad thing. My best guess is that people just want want something to blame when they lose.

Are you serious?

How about maybe because you(Anet) created a grind based pip ladder system where a 50/50 mmr system is totally counter productive to how you are supposed to achieve progression.

And when the losses that prevent progression and ultimately win streaks, are kittened up by preamades v soloers, or afkers, or AN MMR SYSTEM PURPOSEFULLY TRYING TO PRVENT YOU FROM WINNING MORE THAN ONCE AT A TIME, it’s very frustrating.

It’s very easy to say, “well skill will prevail, if you are better than your mmr you will still win games” but it should not be a case if me, premade queueing causally for 3 seasons, getting legendary, then doing solo for 1 season and not getting above ruby IN THE SAME TIME FRAME (we all have irl and cannot grind 50/50), because its a constant win, loss, win, loss, win, win, loss, win, loss etc etc caused by other factors that are not skill based.

Sorry if i sound condescending or passive aggressive but i am amazing that Anet staff haven’t realised this already from the countless threads from people stating how your mmr system is totally counter to having a 3- win streak to get any meaningful progression (win streak extra pip ).

Finally, having a binary progression system where you either, get a pip, or you lose a pip is a really poor reflection of how pvp games are played out. No incentive for players who came close v a premade, or had a ‘gg’ (480-500), or a team who won comfortably v a team they should not have had a chance.

A proper mmr ladder where only win % and player/team rank matters is what is needed. We had it for the first 2 years, it meant something, then you got rid of it and replaced it with this pip grinding Well done for bringing it back, sort of.

Liam McColgan.7689:

Evan Lesh.3295:

I honestly do not know why people think 50/50 matchmaking is a bad thing. My best guess is that people just want want something to blame when they lose.

Are you serious?

How about maybe because you(Anet) created a grind based pip ladder system where a 50/50 mmr system is totally counter productive to how you are supposed to achieve progression.

And when the losses that prevent progression and ultimately win streaks, are kittened up by preamades v soloers, or afkers, or AN MMR SYSTEM PURPOSEFULLY TRYING TO PRVENT YOU FROM WINNING MORE THAN ONCE AT A TIME, it’s very frustrating.

It’s very easy to say, “well skill will prevail, if you are better than your mmr you will still win games” but it should not be a case if me, premade queueing causally for 3 seasons, getting legendary, then doing solo for 1 season and not getting above ruby IN THE SAME TIME FRAME (we all have irl and cannot grind 50/50), because its a constant win, loss, win, loss, win, win, loss, win, loss etc etc caused by other factors that are not skill based.

Sorry if i sound condescending or passive aggressive but i am amazing that Anet staff haven’t realised this already from the countless threads from people stating how your mmr system is totally counter to having a 3- win streak to get any meaningful progression (win streak extra pip ).

Finally, having a binary progression system where you either, get a pip, or you lose a pip is a really poor reflection of how pvp games are played out. No incentive for players who came close v a premade, or had a ‘gg’ (480-500), or a team who won comfortably v a team they should not have had a chance.

A proper mmr ladder where only win % and player/team rank matters is what is needed. We had it for the first 2 years, it meant something, then you got rid of it and replaced it with this pip grinding Well done for bringing it back, sort of.

Community: “Anet why don’t y’all communicate more? We’re left in the dark here!”

Anet: “Okay we’ll communicate a bit, see how that goes. Who knows, maybe they’ll appreciate it.”

Community (AKA Liam here): “Oh look an anet employee showed up in the forums. Lets criticize him and feel good ‘cause he can’t fight back because that would be bad for an employee to do to his customers. Who knows? Maybe they’ll even stop communicating!”

Anet: *goes silent

@Evan Lesh The changes you’ve made to S5 look really good, and I’m super-hyped about getting into the action. It’s a step in the right direction. Don’t let general negativity in the forum get to you. (I’ve almost stopped posting to the pvp forum because of it)

I wanted to ask you about class-based MMR. It’s stated on the wiki page that Glicko makes adjustments to MMR based on the class you are playing but in practice I haven’t really seen this being a thing. So is it really? As a multiclasser this is something I’m interested in.

Look dude, i’m always appreciative of things that deserve praise. I’ve constantly stated how amazing this game is and its combat mechanics. It’s the best out there by far.

But when you still get devs saying “i don’t understand why people hate 50/50 mmr IN A PIP GRIND SYSTEM THAT RELIES ON 3 GAME WIN STREAKS”, despite many people telling otherwise, that is quit worrying.

Of course people appreciate devs telling their customers/the community updates or answering specific questions.

I’l even take this opportunity to say to Evan, i think the up coming changes are very positive and i fully approve of what you guys are trying to make better in S5 despite me not really caring, or enjoying pvp in this game as much anymore.

But when countless people are telling Anet why the 50/50 mmr was not making pip grinding fun, after all fun is the ultimate goal, and the fact they are even accepting this to an extent with the S5 changes, and it’s taken to Season 5 ( kitten happens in real life, sadly not everyone can stay in this bubble of being committed to gw2 all the time) then, you know….criticism is sometimes warranted. Deal with it.

Liam McColgan.7689:

But when you still get devs saying “i don’t understand why people hate 50/50 mmr IN A PIP GRIND SYSTEM THAT RELIES ON 3 GAME WIN STREAKS”

The point is that he never reffered to the pip system, you just read it in that context (and thus had to edit it into his quote). The 50/50 is not bad in itself, only when combined with the pip system and progress only possible through streaks. Now that progress is independent of 50/50 it is no longer an issue.

cephiroth.6182:

Liam McColgan.7689:

But when you still get devs saying “i don’t understand why people hate 50/50 mmr IN A PIP GRIND SYSTEM THAT RELIES ON 3 GAME WIN STREAKS”

The point is that he never reffered to the pip system, you just read it in that context (and thus had to edit it into his quote). The 50/50 is not bad in itself, only when combined with the pip system and progress only possible through streaks. Now that progress is independent of 50/50 it is no longer an issue.

Absolutely. I clearly miss placed my quotation marks :P.

But it took 4 seasons, nearly a whole year of people moaning and quitting the pvp scene for them to realise it was just not fun for people out side of tryharding with full premades.

The Glicko is indeed a good system and realistically thats all people ever want; competitive match making where they feel, with enough trying, the can come out of any match with a win, which is why a 50/50 mmr system is good. But to use it for 4 season in conjunction with a pip grinding system and relying on multiple win streaks for any progression, regardless of tier, and to STILL say, i dont get why people hated it……i guess it triggered me.

Evan Lesh.3295:

I don’t mean to pick on you specifically as the “dreaded 50% matchmaking” conspiracy has a lot of the community in its grips. Ratings obviously can’t be known unless matches are played to generate data. Do you want to play random matches and have a chaotic experience with an inaccurate rating? Or have balanced, competitive matches with an accurate rating?

Sure, we could have zero matchmaking, but two things will happen: First, your teammates will have random ratings and your opponents will have random ratings. Did you lose because you are less skilled than your opponents, or because your teammates are less skilled than your opponents? There is no easy way to tell without a lot of complicated math. “Make a system to figure it out, then!” you say. Well, Glicko does this for us already. Matchmaking assumes its data is correct by pairing opponents that appear to be evenly matched, then Glicko does its job to adjust ratings in case the outcome wasn’t what was expected. Second, random matches make your rating update verrry slowly. Every match won against a much less skilled opponent won’t make you go up because of course you were suppose to beat them, your rating must already be accurate. Every match lost against much higher rated opponents won’t make you go down because of course you were suppose to lose, your rating must already be accurate.

I honestly do not know why people think 50/50 matchmaking is a bad thing. My best guess is that people just want want something to blame when they lose.

My experience last season was about a 50/50 win/loss rate. What made me stop playing last season was:

• I could not progress unless I grinded and got lucky with win streaks. This is being addressed.
• Almost every game ended as a blow out with one team winning by 200-300 points. This happened regardless of whether or not I was on the winning or losing side.

So how is this blow out mechanic being addressed? It sucks the fun out of the game entirely. They’re totally not balanced or competitive,and it’s my experience that nearly every match is this way. We’re not talking the occasional blowout that will happen, it felt like it was designed to be that way.

Evan Lesh.3295:

Or have balanced, competitive matches with an accurate rating?

To be honest, I haven’t seen those very rarely since the start of the League-System. :-(

Evan Lesh.3295:

I honestly do not know why people think 50/50 matchmaking is a bad thing.

50/50 matchmaking with many close matches is good matchmaking. Many blowouts either if you lose or if you win is bad matchmaking.

Evan Lesh.3295:

My best guess is that people just want want something to blame when they lose.

That’s the number one reason for all the forum whining. People can’t admit that they aren’t the best players in the world so if they lose it can’t be because of a more skilled opponent, it must be something different, bad matchmaking, bad balance …

Evan Lesh.3295:

1) There will be the expected periodic balance updates, but I cannot speak to the extend of them.
2) Class stacking has not been a priority because of the leagues rework, and negative feedback regarding character locking when queuing. This might be something worth polling.

I am honestly shocked that class stacking is not a priority. Literally. No words.

Evan Lesh.3295:

I don’t mean to pick on you specifically as the “dreaded 50% matchmaking” conspiracy has a lot of the community in its grips. Ratings obviously can’t be known unless matches are played to generate data. Do you want to play random matches and have a chaotic experience with an inaccurate rating? Or have balanced, competitive matches with an accurate rating?

Sure, we could have zero matchmaking, but two things will happen: First, your teammates will have random ratings and your opponents will have random ratings. Did you lose because you are less skilled than your opponents, or because your teammates are less skilled than your opponents? There is no easy way to tell without a lot of complicated math. “Make a system to figure it out, then!” you say. Well, Glicko does this for us already. Matchmaking assumes its data is correct by pairing opponents that appear to be evenly matched, then Glicko does its job to adjust ratings in case the outcome wasn’t what was expected. Second, random matches make your rating update verrry slowly. Every match won against a much less skilled opponent won’t make you go up because of course you were suppose to beat them, your rating must already be accurate. Every match lost against much higher rated opponents won’t make you go down because of course you were suppose to lose, your rating must already be accurate.

I honestly do not know why people think 50/50 matchmaking is a bad thing. My best guess is that people just want want something to blame when they lose.

Also wanted to say thanks to Evan and the rest of the PvP dev team. These changes are amending many long-standing issues. Looking forward to a more positive experience in S5, S4 was depressing for me.

Hesacon.8735:

My experience last season was about a 50/50 win/loss rate. What made me stop playing last season was:

• I could not progress unless I grinded and got lucky with win streaks. This is being addressed.
• Almost every game ended as a blow out with one team winning by 200-300 points. This happened regardless of whether or not I was on the winning or losing side.

So how is this blow out mechanic being addressed? It sucks the fun out of the game entirely. They’re totally not balanced or competitive,and it’s my experience that nearly every match is this way. We’re not talking the occasional blowout that will happen, it felt like it was designed to be that way.

It’s being addressed by expanding the pool of people with whom you can be matched. Previously the pool was restricted by your current pip tier (within your current division) and expanded from there. Now your tier isn’t being taken into consideration, thus you’re more likely to matched with and against people with comparable MMR even though they’re in a distant tier. Doesn’t make it perfect, but simply by pulling from a larger pool of people and prioritizing skill level over tier, you should be getting more even matches. The code hasn’t been released (to my knowledge) so none of us know the exact metrics that they’re using to determine skill level.

Spartacus.3192:

Evan Lesh.3295:

1) There will be the expected periodic balance updates, but I cannot speak to the extend of them.
2) Class stacking has not been a priority because of the leagues rework, and negative feedback regarding character locking when queuing. This might be something worth polling.

I am honestly shocked that class stacking is not a priority. Literally. No words.

Maybe cause u can reroll?

Exedore.6320:

LUST.7241:

Say {insert known pro player here} plays his/her heart out but is super unlucky with how the matchmaking works and goes, in total Ranked games, 0 Wins 100 Losses. We all know he/she is beyond good. Will they be Legendary/Top 250? Or will the game just see “0 wins? Probably kitten. Bottom 10.”

What I am saying is:

• Is this system just another long grind to the top?
• Will it be a fair, “good players will be ranked high as they should, despite their solo W/L ratio”?
• Will the godly PvP players be 100% subject to the luck of the “unknown forced 50%” matchmaking algorithm?
• Is top 250 simply based on W/L luck streaks?

It seems like you’re only familiar with the terrible pip system of the first four seasons. Forget all about that.

With the new system, rank is directly tied to rating. Rating is tied to performance (win vs. loss in conjunction with your rating vs. the rating of opponents). You can only move up by winning [down by losing]. The higher [lower] you go, the stronger [weaker] your opponents become. Eventually you’ll reach a point where your rating stabilizes because you’re being matched with equally skilled players. Top players should be able to progress higher.

A big problem with the old system, which led to the “luck” feel was the use of pip range in matchmaking. This restricted the matching pool, which in turn caused the rating difference to expand as you sat in queue longer. By the time a game was started, a massive rating difference had developed and winning or losing became a crap-shoot. With a larger pool, the rating difference should be smaller, leading to closer games and more accurate rating calculations.

Summary:

• Pips are only for the reward system now. They no longer impact rank or matching. Pips can only be gained (not lost). Rewards through pips have a grind element.
• Rank/Tier is not a grind like the last 4 seasons. Matchmaking and ranking is done with rating. If a player is misplaced, they’ll win or lose a lot more and their rating will adjust appropriately.
• Tier is simply a name assigned to a range of numerical ratings. You can go up and down in tier.
• There is some decay in rating so that players can’t sit at the top all season.
• There was NEVER a forced 50% in effect. What happens is that when you’re at the correct rating and matched against players of similar rating, you’re expected to win 50% of your games. If your skill improves, you’ll start winning more and moving up.
• Win streaks are possible, even at the top, but aren’t nearly as prone to luck as they were in previous seasons.
• A top player should never have an awful win rate. If they’re losing more than winning, their rating will drop until they’re where they should be.

My only concern is that the solo/duo queue restriction will lead to “solo queue hero” behavior rather than true teamplay – the strong solo player will focus on winning fights rather than controlling nodes and helping the team.

@ Exedore / Just wanted to thank you for a proper response that made people (myself for sure) know for sure what is going on instead of a non-informative response which insinuated people just whine about the 50/50 mmr when they’re actually just ignorant of the limiting matchmaking issues with the pip system of previous seasons. Looking forward to this season for sure!

I have played thousands of games, forced 50% win rate does exist, no matter how devs attempt to deny it. It is painfully obvious if you played so many games and just have blowout after blowout (100-500, 500-100) matches over and over again. Not to mention, my team completely destroys enemy team – next match i teamed with players i just completely destroyed – THIS IS FORCED 50% win rate. If matchmaking worked properly after win you would be facing better enemy and teammates with better teammates, but in GW2 you face pro league players with potatoes from diamond that die to Svanir just because you won a match before.

Remember when we would get multiple pips for winning games? Yeah, those were great tell of how desperately matchmaking attempts to make you lose. I had so many games (after an easy win) where i would get 3 pips for the game we barely won somehow (by huge luck and some insane hardcarry) and it was clear as day we were supposed to lose it.

Krysard.1364:

Spartacus.3192:

Evan Lesh.3295:

1) There will be the expected periodic balance updates, but I cannot speak to the extend of them.
2) Class stacking has not been a priority because of the leagues rework, and negative feedback regarding character locking when queuing. This might be something worth polling.

I am honestly shocked that class stacking is not a priority. Literally. No words.

Maybe cause u can reroll?

Well obviously any fix to class stacking requires the prevention of rerolling. This has been discussed to death and a ton of bad matchmaking would be fixed if class stacking was not allowed.

Every time I see 50/50 matchmaking pop up in discussion I always wonder why people hate it. When you get placed in the match you and the 9 others are sorted based on which combination gives both teams as close to a 50% chance to win.
Why would it sort you on a team that has an 80% chance to win if there is a more balanced solution? Sounds like this is what people want.

That’s how I have understood it to mean, if I am incorrect in my assessment please correct me.

Is class stacking really that bad ?

Role stacking can be a problem but class stacking ? In a normal situation class = role. But in a class stacking situation you just need to change build or re-roll.

Warcry.1596:

Every time I see 50/50 matchmaking pop up in discussion I always wonder why people hate it. When you get placed in the match you and the 9 others are sorted based on which combination gives both teams as close to a 50% chance to win.
Why would it sort you on a team that has an 80% chance to win if there is a more balanced solution? Sounds like this is what people want.

That’s how I have understood it to mean, if I am incorrect in my assessment please correct me.

No, what people mean by 50/50 matchmaking is that they believe that 50% of the time they’re being placed on teams that are predisposed to win, and 50% of the time they’re being placed on teams that are predisposed to lose. So they don’t believe that they have anywhere near a 50% chance to win/lose each match, they believe that pretty much every match they play is unfairly imbalanced in terms of each team’s skill level. Unfortunately 50/50 matchmaking also sounds like your interpretation, we need 2 different terms to differentiate them. Evan says that a system that intentionally creates grossly imbalanced matches does not exist.

Wargameur.6950:

Is class stacking really that bad ?

Role stacking can be a problem but class stacking ? In a normal situation class = role. But in a class stacking situation you just need to change build or re-roll.

Because in this wonderful state of game we are in where build diversity is practically zero, anytime you see a class you are 90% sure of the exact build that person is using. Sure there are some niche builds out there and surprise factor can help but lets not kid ourselves, most build have MUST HAVE traits and traitlines. Most people will just tweak a sigil or rune.

Class stacking is a major problem which is why it was removed from ESL.

Evan Lesh.3295:

I honestly do not know why people think 50/50 matchmaking is a bad thing. My best guess is that people just want want something to blame when they lose.

Because I am awesome and should be winning every match. Even the ones I am not playing in!!!!!

Evan Lesh.3295:

I honestly do not know why people think 50/50 matchmaking is a bad thing. My best guess is that people just want want something to blame when they lose.
EDIT: I agree 50/50 matchmaking sucked for the pip-based league. My comment is more general, regarding desired match quality and rating accuracy.

People don’t like it because statistically most people think they are above average. When they end up with a 50/50 win rate, they are made to feel average. Also, I think a very large part of it is how horribly it meshed with the pip system where it basically worked against you going up in divisions, which just left an extremely sour taste in people’s mouths for the system.

Honestly I don’t know why 50/50 matchmaking gets so much hate. Maybe because of the pip system and how it was implemented in past. But in an ideal matchmaking system, you would have 2 teams of equally skilled players matched up against each other. By simple statistics you will average a 50% win rate in that scenario.

Too many people feel like they are entitled to long win streaks and if they don’t win every game they play in, then the system is broken.

I’d certainly like to see a tournament system that takes into account the difficulty if your opponent. If you go up against tougher and tougher competition, you should advance or earn better reward if you win.

As for blow outs, they happen for a variety of reasons. In my experience, most of the blow out losses I’ve been a part haven’t occurred because my team was worse than the other team. It was more about miscommunication on rotations or pushing too many points. The actual skill of the players was fairly matched.

Evan Lesh.3295:

Visible skill rating should help shift focus from win rate to an actual performance metric.

Are we only going to see our own skill rating and the ones on the leaderboard(s)? Or will we know the ratings of other players in our matches?

Evan Lesh.3295:

Just your own and leaderboard.

Ok. I wish we got more, but I understand that it could lead to some toxicity. Could we perhaps get some anonymized or generalized ratings at some point to get an idea of where we stand within the entire community?

While I might know that my rating is 1500, the top player is 2200 and everyone who made the leaderboard is >2000, that doesn’t really mean a lot in a large community. I don’t really know if 1500 is average, above or below, in the 80 percentile, or 40 percentile.

Evan Lesh.3295:

LUST.7241:

What I am saying is:

• Is this system just another long grind to the top?
• Will it be a fair, “good players will be ranked high as they should, despite their solo W/L ratio”?
• Will the godly PvP players be 100% subject to the luck of the “unknown forced 50%” matchmaking algorithm?
• Is top 250 simply based on W/L luck streaks?

• You will absolutely not be able to grind to the top of the Skill Rating leaderboard.
• Good players will be ranked highly. They already are. We’ve had MMR for years and track it constantly.
• No one is being forced into an unknown system. See below and read this: https://wiki.guildwars2.com/wiki/PvP_Matchmaking_Algorithm
• Sometimes players can overshoot their ‘true’ rating by chance, but it will not be by far and not last very long. They certainly won’t reach the top 250.

I don’t mean to pick on you specifically as the “dreaded 50% matchmaking” conspiracy has a lot of the community in its grips. Ratings obviously can’t be known unless matches are played to generate data. Do you want to play random matches and have a chaotic experience with an inaccurate rating? Or have balanced, competitive matches with an accurate rating?

Sure, we could have zero matchmaking, but two things will happen: First, your teammates will have random ratings and your opponents will have random ratings. Did you lose because you are less skilled than your opponents, or because your teammates are less skilled than your opponents? There is no easy way to tell without a lot of complicated math. “Make a system to figure it out, then!” you say. Well, Glicko does this for us already. Matchmaking assumes its data is correct by pairing opponents that appear to be evenly matched, then Glicko does its job to adjust ratings in case the outcome wasn’t what was expected. Second, random matches make your rating update verrry slowly. Every match won against a much less skilled opponent won’t make you go up because of course you were suppose to beat them, your rating must already be accurate. Every match lost against much higher rated opponents won’t make you go down because of course you were suppose to lose, your rating must already be accurate.

I honestly do not know why people think 50/50 matchmaking is a bad thing. My best guess is that people just want want something to blame when they lose.
EDIT: I agree 50/50 matchmaking sucked for the pip-based league. My comment is more general, regarding desired match quality and rating accuracy.

its sounds good… but… to adjust to get fine matches it should have a way of traking classes, a player can be good with one, two… classes but suck with others and they are things (dayli class achiev, ascencion class achiev, pvp class achiev) that makes players go on matches whith non her optimal classes and getting a suposed your level player whith a class that they dont know imbalances the match.
If making a class rank is to heavy it can be considered taking in some way the deviation of win loss ratio on hiss “main class” versus the one he is queuing of course it needs a class block on q that forces to play whith the class you q

Evan Lesh.3295:

Well now that we will have visible ratings, we will start talking to the community about averages, ranges, deltas, everything MMR. Average/starting rating is 1200, and has a bell curve distribution.

Excellent. Thank you.

For all TIME, I applaud the PVP team for the revamp.

Their new league seems to kinda have listen to a skill based system.

I had put extensive effort for this to happen.

Here we go:
https://forum-en.gw2archive.eu/forum/game/pvp/Skill-based-or-Grind-based-rank-system/first#post6285847

Then, I want to Cheer UP Evan Leesh that tried so many things to make us HAPPY. He work hard and know that if what he do is not good enough, he don’t fear to look at it and make it better.

So Tribute to Evan Leesh:
https://forum-en.gw2archive.eu/forum/game/pvp/Tribute-to-Evan-Lesh/first#post6265263

Those two link are my best rate by the community.

I want to go out a bit further and ask Evan to take great care about the PVP only pack proposal many players did here:
https://forum-en.gw2archive.eu/forum/game/pvp/sPvP-needs-its-own-business-model/first

Evan:
Thank you very much. I really expected to have a great time in the League Season 5.

But… please, make something for TEAMs, so new team form up with goals in mind. The auto-tourney at released was a strong vector of team creations (who doesn’t want to try their best to earn GEMS???).

Even, a copy cat of the season 5… where SOLOQ Season 6 and TEamQ (for duo or more) Season 6 would be awesome to help people have goals to make teams.

We definately need more goals for TEAM.

Jourdelune.7456:

For all TIME, I applaud the PVP team for the revamp.

Their new league seems to kinda have listen to a skill based system.

I had put extensive effort for this to happen.

Here we go:
https://forum-en.gw2archive.eu/forum/game/pvp/Skill-based-or-Grind-based-rank-system/first#post6285847

Then, I want to Cheer UP Evan Leesh that tried so many things to make us HAPPY. He work hard and know that if what he do is not good enough, he don’t fear to look at it and make it better.

So Tribute to Evan Leesh:
https://forum-en.gw2archive.eu/forum/game/pvp/Tribute-to-Evan-Lesh/first#post6265263

Those two link are my best rate by the community.

I want to go out a bit further and ask Evan to take great care about the PVP only pack proposal many players did here:
https://forum-en.gw2archive.eu/forum/game/pvp/sPvP-needs-its-own-business-model/first

Evan:
Thank you very much. I really expected to have a great time in the League Season 5.

But… please, make something for TEAMs, so new team form up with goals in mind. The auto-tourney at released was a strong vector of team creations (who doesn’t want to try their best to earn GEMS???).

Even, a copy cat of the season 5… where SOLOQ Season 6 and TEamQ (for duo or more) Season 6 would be awesome to help people have goals to make teams.

We definately need more goals for TEAM.

yeah, nobody is saying that had not to be a team mode, the problem is that in anterior leagues forming a team, a real team(not geting a bunch of pugs and q as a team), is a way for farming soloers, not for playing “competitive” againts other teams

Digging the detailed responses, thank you again.

Evan Lesh.3295:

Close team score is not a factor, but close rating is. If two teams all have the exact same rating, winning or losing will barely effect your rating because everyone’s skill level is probably the exact same. I would really like to factor in team scores, but it’s very tricky because conquest scoring is not very indicative of skill. A totally outclassed team can often get 200 points just because of how 3 point conquest works.

Could there be some way we can change the way conquest scoring works? Take PvE (yeah…) “capture” events for example, if enemy mobs are on on the capture…the capture doesn’t tick. If they out number the player forces, it does. Could something like that open the doors to better conquest scoring and factoring in team scores?

Evan Lesh.3295:

There is no good way of tracking individual player skill in gw2 conquest to help get more accurate ratings. There are way too many variables. This has been a case of not making kittenumptions that actually end up making ratings LESS accurate.

Well, this brings me back on how heavily modified CSGO has become. There are a lot of variables (different than this game of course) that, over the course of their competitive reworks, have more accurately placed people closely into their true rankings.

I agree, there are a lot of variables…but I trust that some variables are more important than others. These should be prioritized. Are there not statistics on your end that can show this?

Evan Lesh.3295:

Matchmaking has always been based on skill (your Glicko rating), not your win/loss record. Glicko rating updates obviously requires the outcome of the match (W/L) to adjust your rating. I think my answer above addresses this as well. There seems to be some confusion between two concepts: Using rating to pick 10 players for a match and using the outcome of a match to update an individual player’s rating.

So we have the use of rating to get a match, then the use of the outcome to update individual rating. You mentioned you’ll barely move in a close match. So, ideally, it should be giving you close matches VERY frequently (almost always) this season (unlike previous seasons), correct?

In that case, is the luck of the draw heavily based on how you placed in the placement matches? How much weight does the initial outcome (W/L) hold? And, since these season apparently changes the system dramatically, why isn’t everyone fully reset to 0 (instead of a soft reset)?

What would happen in this event: A pro player goes 0-10 placements (assume the reason is because of the team). The system will try to match them closely (assume this always happens), like you said, with their given MMR. Doesn’t this just equal the same thing as the last system? A grind to the top?

Evan Lesh.3295:

We tried multiple flavors of matchmaking specifically to address the conflict of goals between balanced matches and the pip-based league. None of them worked, which is why we’ve scrapped it. Also, none of the changes we tried with matchmaking expanded the pool, it just changed how teams were formed.

Since Unranked (apparently) didn’t follow how Ranked worked. Was there any testing done in Unranked? Or was each Ranked season tests like everyone assumes thus far?

Shadowstep.6049:

I have played thousands of games, forced 50% win rate does exist, no matter how devs attempt to deny it. It is painfully obvious if you played so many games and just have blowout after blowout (100-500, 500-100) matches over and over again. Not to mention, my team completely destroys enemy team – next match i teamed with players i just completely destroyed – THIS IS FORCED 50% win rate.

There is not a forced 50% win rate. You example does not support your claim and is actually an example of why pip-range matching failed miserably.

What was actually happening was that your pip range had relatively few people queuing. With a limited pool, the rating differential expands and you get teams both with a high deviation in rating. That makes the outcome a bit of a crap-shoot. The losing team loses rating and the winning team gains it. Next time, because the pool is still small, you get most of the same people. Since you won and your rating is higher, you’re more likely to get the lower people on your team – the ones who lost last time.

When you remove the pip restriction, you increase the pool, which reduces rating differential between teams and within teams, and reduces the chance of seeing the same players over and over. It also means that the system does less swapping of high and low players in order to try and balance teams.

LUST.7241:

Well, this brings me back on how heavily modified CSGO has become. There are a lot of variables (different than this game of course) that, over the course of their competitive reworks, have more accurately placed people closely into their true rankings.

The CS:GO MVP system seems like a good way to boost better players out of the middle of the bell curve.

So we have the use of rating to get a match, then the use of the outcome to update individual rating. You mentioned you’ll barely move in a close match. So, ideally, it should be giving you close matches VERY frequently (almost always) this season (unlike previous seasons), correct?

Yes. Remember that the player population isn’t infinite, so you can’t find a perfect rating match immediately. The initial look is within a fixed range and if you’ve been in queue for a few minutes, the differential expands. However, having one pool of players to check for similar ratings as opposed to 20% or less of that pool will allow for closer and quicker matches.

In that case, is the luck of the draw heavily based on how you placed in the placement matches?

Placement matches give you an initial rating. If it’s grossly inaccurate, you’ll either see win or loss streaks as you reach a better rating. In addition, Glicko2 tracks rating deviation and volatility. If you’re misplaced, your volatility is higher, so you’ll move further in rating with each game until you settle.

Evan Lesh.3295:

Liam McColgan.7689:

Evan Lesh.3295:

I honestly do not know why people think 50/50 matchmaking is a bad thing. My best guess is that people just want want something to blame when they lose.

Are you serious?

How about maybe because you(Anet) created a grind based pip ladder system where a 50/50 mmr system is totally counter productive to how you are supposed to achieve progression.

I agree 50/50 matchmaking was bad for the pip system. That is why we tried 3 different styles of matchmaking for seasons 1, 2, and 3, and now revamped the system entirely. 50/50 will once again be best.

Glad you guys finally figured this out. It made no sense for the pip system.

Ironically, 50/50 matchmaking is always going to have about 50% haters. It helps those who are below average experience a better chance to win while making things more difficult for better than average players.

Should a good player only win 50% of their games? Should a bad player win 50%? Most people would say “no”, but it’s a Catch-22. Without 50/50 matchmaking, people would complain about having less competitive games.

My only concern is that even with forced 50/50 matchmaking most games were not competitive in the prior seasons. Thus, you have all the negatives of 50/50 matchmaking, but haven’t really seen the positives yet.

IMO, excluding pips and premades from matchmaking will help, but I suspect you guys are still creating mismatched teams from the very start by not factoring in class stacking/builds.

For example, your matchmaking can get lucky and find the EXACT SAME SKILLED players on both teams. However, if you put a couple necros on one team with no healer/rezbot then everybody knows what’s going to happen 90% of the time. The necros will get zerged in the initial team fight and the rest of the game is just a lopsided 4v5.

There is almost no coordination/synced builds with premades banned now, so it’s more important than ever to address obvious issues with class balance and stacking. Classes now need to be balanced for non-coordinated play or else you will only be solving one problem to create an even worse problem.