Let's talk about the MMR...

Thogard

TL:DR
MMR doesn't segregate premades from solo queuers, it just inflates the MMRS of the people in the premade because they're in a premade. MMR is based on the average of the group. The longer you're in queue for, the greater the gap can be between opponents' MMRs for a given match. This sucks for high MMR solo queuers because they get matched up against premades who have usually been waiting so long that the MMR system has essentially subtracted their premade bonus to find them a match due to how long they've sat in queue. The MMR system keeps allowing lower and lower MMR solo queuers to fight them, but it's tiered.. so it'll start the match with three, and won't add a fourth solo queuer to fight the premades until more time passes to subtract even more from the MMR of the premades.

So Let's talk about the MMR...

I'm going to separate the points I'd like to make into either "Facts" or "Working Hypotheses" based on whether I'm certain about a given piece of information, or am just hypothesizing based on the data I've collected.

FACTS

Is the MMR actually a thing?
Yes. And it always has been. Anyone that's every tried to synchronize queues to get into specific matches can tell you that it wasn't simply a matter of "who queued first." But now it's different...

What is MMR?
MMR stands for Match Making Rating (or Ranking depending on who you ask). It's a way to rate opponents so that players of a similar skill level can get matched together for a better, more even match. Think of it kind of like the weight classes for UFC or Boxing... it wouldn't be very fair to match a 140lbs guy against a 220lbs guy, so make sure the fights are fair, the fight organizers tend to make sure boxers fight people with a similar weight (I.E. the same "weight class".) But every once in a while, a fighter will fight someone in a weight class different than their own...

So if the MMR already existed, what did they change with Summerset?
To quote the patch notes: "Battleground match making now adds an additional multiplier to the value of players who are grouped. This value scales differently between 2, 3, and 4 player groups.... Pre-formed groups will generally be pitted against pre-formed groups, and solo players will generally be pitted against other solo players."

But wait... Are they or aren't they separating the premades from the solo queues?
Technically, NO, they are not separating the premades from the solo queues. What they ARE doing is increasing the MMR of the premades solely because they're premades. The more people in the "premade" group, the higher the increase to MMR for matchmaking purposes. So to use the boxing "weight class" metaphor... if you weigh 170 normally, but you queue as a duo, the system might consider you as actually weighing 180. If you queue as a trio, it might consider you as someone who weights 190, and if you queue as a four man, it might consider you as someone who weighs 200. In other words, it adds +10 each for a duo, +20 each for a trio, and +30 each for a four man

Tested Hypotheses:

So if you're in a four man group with four different individual MMRs, which MMR is used for the group?
The MMR for the group is the average of the individual characters, after adjusting them for the size of the premade. To explain using the boxing "weight class" metaphor again, if you have a 150 lbs, a 160 lbs, a 170 lbs, and a 190 lbs fighter queuing together, we'd add 30 lbs to each since they're queueing as a premade four man... that gives us 180, 190, 200, and 210. Then we would take the average of that, which is 195. That 195 is the MMR.. or the "weight class" that the premade would have to fight in. I am 99.999% sure that it's the average of the group, and I am 99.999% sure that it's character based rather than account based.

What is MMR based off of?
This is the thing we're the least certain of at this moment. I strongly suspect that it's based off of an aggregate of your medal scores.... in essence, the sum of all your leaderboard scores. I believe this because players with both a high win rate and a low win rate seem to have high MMRs if they've played a lot of games...Similarly, players with high K/D ratios but also players with low k/d ratios tend to have high MMRs if they've played a lot of games... but my MAIN reason for thinking this is the case is because we already know that this is the system ZOS uses for the "Leaderboards" so we already know that they're tracking it.. and it would make sense for the "leaderboard" metrics (cumulative medal score in all games) to also determine MMR.... the difference between the "Leaderboards" and the MMR system is that the Leaderboards reset each week but the MMR doesn't... we know that players who used to play a ton of games but haven't played one in a month or two will still have really high MMRs. This theory also means that your MMR can't go down.... since you can never "lose" leaderboard points by performing poorly... so if someone does have their MMR go down, this theory will be disproven.

So what happens if there's nobody at your MMR level to fight?
Here's where things get interesting. Let's use the example above of a four man with an MMR of 195... they queue up, but there is no other premade out there with an MMR that high.. so they wait. It appears that the battleground group finder will be more flexible with your MMR the longer you're in queue. For instance, if this premade with MMR of 195 queues up immediately, the system will only look for groups or players with an MMR of between 190-200. Then, after five minutes, if it can't find anyone in that "weight class" it widens the range by looking for matches with an MMR of between 180 and 210. Then after another five minutes, it widens it again, looking for opponents with an MMR of between 170 and 220. Then another five minutes, it widens again for opponents between 160 and 230, etc.... Do I know how often it widens the range? no, but i do strongly suspect it's on either 5 minute or 10 minute intervals.

So if I solo queue, how come I keep fighting premades?
Sounds like you've got a high MMR! The longer you spend queued, the more likely you are to fight "outside" of your weight class (MMR level).

How come when three premades all queue up at the same time, they have to wait 20 minutes to fight each other, even if all have been waiting for twenty minutes?
This is something that drives me insane but it is a distinct phenomenon of the new matchmaking system that didn't exist before. Let's say that two of the premades are full of people playing their BG mains... but one premade has a few alts in it, or maybe some veteran PvPers that just didn't do BGs until now. While all three premades are going to get the "premade bonus" to MMR, the MMR is character based... so the group finder would consider it an uneven match due to the low MMR of the team that's playing alts and new characters. Let's say that the average MMR of team A is 200, team B is 210, and team C is 160. Then let's add the premade bonus of 30 onto that and we get an MMR of 230 for A, 240 for B, and 190 for C... A and B will get matched together and considered "acceptable" after approx five minutes in the queue... but they'll have to stay queued for a while before the group finder will consider team C as eligible to fight in their weight class... so they wait...

How do high MMR solo queuers fit into all this?
Let's use the example in the previous question but add a twist... So:

• Team A has an MMR of 200(base)+30(premade)=230,
• Team B has an MMR of 210(base)+30(premade)=240
• Team C has an MMR of 160(base)+30(premade)=190

But then let's say that there are a few solo queuers with a really high MMR. Let's say that

• Solo queuer 1 has an MMR of 210(base)+0(solo)=210
• Solo queuer 2 has an MMR of 210(base)+0(solo)=200
• Solo queuer 3 has an MMR of 200(base)+0(solo)=200
• Solo queuer 4 has an MMR of 180(base)+0(solo)=180

So what happens? Team A and B get "matched up" almost immediately because their MMR is so similar... but then there's nobody else in that weight class for them to fight. After 5 minutes, they're eligible to fight teams with an MMR of 220. After 10 minutes they're eligible to fight teams with an MMR of 210, after 15 minutes they're eligible to fight teams with an MMR of 200, and after twenty minutes they're eli...

..hey wait a second?! Queue pop after just 15 minutes! Teams A and B just entered into a match against... a pug group comprised of solo queuers 1, 2, and 3?!? How is this possible? It's because the BG system only needs 3 players on a team for the match to start.

So solo queuer 4 with an MMR of 180 gets added in immediately, right? NOPE! The BG queue finder keeps doing what it's doing. Where were we? Oh yes..

After twenty minutes (or five minutes into the match), the match is eligible to have MMRs of 190 join. Team C was a premade with an MMR of 190... but there are four of them and we can't fit four players into just one slot... So team C is out of luck, and the match maker continues lowering the MMR requirements for eligible players. After another five minutes (ten minutes into the match), MMRs of 180 become eligible, and solo queuer 4 (who has an MMR of 180) gets added onto the 3 man team consisting of solo queuers 1, 2, and 3... who, by the way, have been trying to fight as a pug 3 man against two full premades for the last ten minutes. Super fun for them.

And what happens to Team C, that never got into that match? Well they wait another 5 minutes or so and then fight two full groups of pug Solo queuers with MMR averages of around 130. Because their minimum MMR threshold for suitable opponents has been going down this whole time too, and they started off at 190...

Well, that kind of makes sense.. but why is it a problem?
It's a problem because if you're a high MMR solo queuer, you're almost always going to be the pug in a fight against premades. Solo queueing with a high MMR is, based on the current population trends, almost a guaranteed way to get stuck against premades. Why? Because the lower MMR solo queuers are matching up against each other so fast that their BG queue finder doesn't need to look for players in a higher weight class. They are getting into matches so quickly that the "allowed MMR range" of that match never gets high enough to grab the high MMR solo queuer. But meanwhile, on the other side of the spectrum, the high MMR premades have been waiting for 20 minutes. After a 20 minute wait, the "bonus" to their MMR that they get for being a premade is essentially nullified by the BG queue finder's willingness to allow them to fight in different, lower weight classes. So the high MMR solo queuer, being the closest thing to the time-lowered MMR of the premades who have been waiting, gets gobbled up to fight the premade before that solo queuer can wait long enough to allow their own MMR to get low enough to fight in the medium/low MMR solo queue matches.

And this is really frustrating because it is making BGs miserable for every solo queuer who has played a ton of BGs... In other words, it takes the people who love the BGs the most and makes them wait or gives them the worst matches.

Is there a silver lining?
Yes. High MMR premades tend to fight against High MMR premades when they are in queues together. This is a good thing if there are three High MMR premades all queueing around the same time. And if there are going to be solo players fighting the high MMR premades, it makes sense to take the "best" (let's not discuss the correlation or lack thereof between being High MMR and being good) solo players and put them in the match. It just really sucks to be a solo player who is always getting put in those matches...

@ZOS_BrianWheeler @ZOS_GinaBruno @ZOS_JessicaFolsom

(Kudos to anyone who loves BGs enough to read this whole thing... it turned out a lot longer than I planned lol)

ecru

Kinda been over this a bunch of times on the forums. How do you implement MMR with three teams? Balanced matchmaking with three teams means that you lose two thirds of your games. This would be better for people right now who are screwed by the matchmaking system and end up going 1-5, but if they managed to implement it properly, would you personally be okay with losing two thirds of your games, even with a stacked premade?

The three team system is broken from the start and good matchmaking only averages out the misery caused by it.

BTW, I don't actually believe the current matchmaking system is working as it should be, because I know that players are not losing two thirds of their games. Some are winning 90%, while others are losing 90%.

Thogard

ecru wrote: »

Kinda been over this a bunch of times on the forums. How do you implement MMR with three teams? Balanced matchmaking with three teams means that you lose two thirds of your games. This would be better for people right now who are screwed by the matchmaking system and end up going 1-5, but if they managed to implement it properly, would you personally be okay with losing two thirds of your games, even with a stacked premade?

The three team system is broken from the start and good matchmaking only averages out the misery caused by it.

BTW, I don't actually believe the current matchmaking system is working as it should be, because I know that players are not losing two thirds of their games. Some are winning 90%, while others are losing 90%.

The MMR system doesn't really have anything to do with whether it's two teams or three teams. There are plenty of games that have multiple teams... hell even a 12 person free for all mode can use MMR effectively.

I know it's super long, but I strongly suggest you read the whole thing.. particularly:

What is MMR based off of?
This is the thing we're the least certain of at this moment. I strongly suspect that it's based off of a lifetime cumulative total of your medal scores.... in essence, the sum of all your weekly leaderboard scores since you started the character. I believe this because players with both a high win rate and a low win rate seem to have high MMRs if they've played a lot of games...Similarly, players with high K/D ratios but also players with low k/d ratios tend to have high MMRs if they've played a lot of games... but my MAIN reason for thinking this is the case is because we already know that this is the system ZOS uses for the "Leaderboards" so we already know that they're tracking it.. and it would make sense for the "leaderboard" metrics (cumulative medal score in all games) to also determine MMR.... the difference between the "Leaderboards" and the MMR system is that the Leaderboards reset each week but the MMR doesn't... we know that players who used to play a ton of games but haven't played one in a month or two will still have really high MMRs. This theory also means that your MMR can't go down.... since you can never "lose" leaderboard points by performing poorly... so if someone does have their MMR go down, this theory will be disproven.

Thogard

Oh and I forgot the most important bullet point...

What would you recommend we do to fix it?
The biggest problem right now is the WAITING. We have 12 people with high MMRs waiting to fight each other...we know they'll fight each other as long as no new players queue... it's just frustrating that the BG queue finder makes them wait long enough for the difference in their MMRs to be "acceptable"

They need to be more flexible with how quickly players can be matched up against other players with a lower (or higher) MMR, and I would strongly suggest that they make that a function of how many people are queueing for BGs.

For example, if there are 60 people "in queue" then the system will incrementally increase its tolerance for MMR mismatches every five minutes. If there are 30 people in queue, the system will incrementally increase its tolerance for MMR mismatches every three minutes. If there are 20 people in queue, the system will incrementally increase its tolerance for MMR mismatches every two minutes... etc.

There are actually better ways to do it (make it logarithmic for the higher MMR players...) but I've already gotten so nerdy in this thread that I don't think going deeper on this concept would be appreciated lol. I also suspect that going down that route could very well end up with me describing the old system (if it's fully logarithmic it might be indistinguishable from the old system)

ecru

Thogard wrote: »

ecru wrote: »

Kinda been over this a bunch of times on the forums. How do you implement MMR with three teams? Balanced matchmaking with three teams means that you lose two thirds of your games. This would be better for people right now who are screwed by the matchmaking system and end up going 1-5, but if they managed to implement it properly, would you personally be okay with losing two thirds of your games, even with a stacked premade?

The three team system is broken from the start and good matchmaking only averages out the misery caused by it.

BTW, I don't actually believe the current matchmaking system is working as it should be, because I know that players are not losing two thirds of their games. Some are winning 90%, while others are losing 90%.

The MMR system doesn't really have anything to do with whether it's two teams or three teams. There are plenty of games that have multiple teams... hell even a 12 person free for all mode can use MMR effectively.

I know it's super long, but I strongly suggest you read the whole thing.. particularly:

What is MMR based off of?
This is the thing we're the least certain of at this moment. I strongly suspect that it's based off of a lifetime cumulative total of your medal scores.... in essence, the sum of all your weekly leaderboard scores since you started the character. I believe this because players with both a high win rate and a low win rate seem to have high MMRs if they've played a lot of games...Similarly, players with high K/D ratios but also players with low k/d ratios tend to have high MMRs if they've played a lot of games... but my MAIN reason for thinking this is the case is because we already know that this is the system ZOS uses for the "Leaderboards" so we already know that they're tracking it.. and it would make sense for the "leaderboard" metrics (cumulative medal score in all games) to also determine MMR.... the difference between the "Leaderboards" and the MMR system is that the Leaderboards reset each week but the MMR doesn't... we know that players who used to play a ton of games but haven't played one in a month or two will still have really high MMRs. This theory also means that your MMR can't go down.... since you can never "lose" leaderboard points by performing poorly... so if someone does have their MMR go down, this theory will be disproven.

I read it. I have no idea what your MMR in ESO is based off of, but I doubt they're weighting medals, damage, or healing, and if they are, that might explain why the system isn't really working. These things don't have as much to do with whether your team wins or loses as you'd expect.

ELO in Rift was effective enough to pin most players at a 50% win rate and it was purely weighted by win/loss. Any player close to 60% was in the top 1% of players. Previously they allowed damage and healing to influence ELO but the system created too many outliers who were not where they should be and ended up with very poor matches due to that. I don't believe ZOS could implement it any better and a system based purely on win/loss would likely work the best.

ELO is supposed to decrease if you lose, but a good matchmaking system will use an adjusting k-factor to determine how much is gained or lost, with the best players having a lower k-factor so they can't sandbag out of very high ELO (losing very little ELO when they lose).

edit: Your issue with high ELO players facing long queue times is completely unavoidable btw. If there aren't enough players to create a good match, then a match isn't created. This is how all matchmaking systems work. If you're an extreme outlier, you will wait longer, you will get bad matches, you will get matched against premades, etc. I experienced this in Rift often, having one of the highest ELOs in the game. It was me with potatoes vs experienced premades or generally good players. It sucks, but I don't know if there is any real solution to this problem.

Thogard

ecru wrote: »

Thogard wrote: »

ecru wrote: »

Kinda been over this a bunch of times on the forums. How do you implement MMR with three teams? Balanced matchmaking with three teams means that you lose two thirds of your games. This would be better for people right now who are screwed by the matchmaking system and end up going 1-5, but if they managed to implement it properly, would you personally be okay with losing two thirds of your games, even with a stacked premade?

The three team system is broken from the start and good matchmaking only averages out the misery caused by it.

BTW, I don't actually believe the current matchmaking system is working as it should be, because I know that players are not losing two thirds of their games. Some are winning 90%, while others are losing 90%.

The MMR system doesn't really have anything to do with whether it's two teams or three teams. There are plenty of games that have multiple teams... hell even a 12 person free for all mode can use MMR effectively.

I know it's super long, but I strongly suggest you read the whole thing.. particularly:

What is MMR based off of?
This is the thing we're the least certain of at this moment. I strongly suspect that it's based off of a lifetime cumulative total of your medal scores.... in essence, the sum of all your weekly leaderboard scores since you started the character. I believe this because players with both a high win rate and a low win rate seem to have high MMRs if they've played a lot of games...Similarly, players with high K/D ratios but also players with low k/d ratios tend to have high MMRs if they've played a lot of games... but my MAIN reason for thinking this is the case is because we already know that this is the system ZOS uses for the "Leaderboards" so we already know that they're tracking it.. and it would make sense for the "leaderboard" metrics (cumulative medal score in all games) to also determine MMR.... the difference between the "Leaderboards" and the MMR system is that the Leaderboards reset each week but the MMR doesn't... we know that players who used to play a ton of games but haven't played one in a month or two will still have really high MMRs. This theory also means that your MMR can't go down.... since you can never "lose" leaderboard points by performing poorly... so if someone does have their MMR go down, this theory will be disproven.

I read it. I have no idea what your MMR in ESO is based off of, but I doubt they're weighting medals, damage, or healing, and if they are, that might explain why the system isn't really working. These things don't have as much to do with whether your team wins or loses as you'd expect.

ELO in Rift was effective enough to pin most players at a 50% win rate and it was purely weighted by win/loss. Any player close to 60% was in the top 1% of players. Previously they allowed damage and healing to influence ELO but the system created too many outliers who were not where they should be and ended up with very poor matches due to that. I don't believe ZOS could implement it any better and a system based purely on win/loss would likely work the best.

ELO is supposed to decrease if you lose, but a good matchmaking system will use an adjusting k-factor to determine how much is gained or lost, with the best players having a lower k-factor so they can't sandbag out of very high ELO (losing very little ELO when they lose).

Yeah the system you're describing sounds way too sophisticated to be what we're going through right now. The current BG matchmaker is very predictable for me now, and I am 100% certain that win rate is not a factor at all. We've swapped out newer, high win rate players for older, low win rate characters.. and the older, low win rate characters have a SIGNIFICANTLY higher MMR.

It could also be based off of "# of BG games played" but would that really be distinguishable from lifetime medal score?

ecru

Thogard wrote: »

ecru wrote: »

Thogard wrote: »

ecru wrote: »

Kinda been over this a bunch of times on the forums. How do you implement MMR with three teams? Balanced matchmaking with three teams means that you lose two thirds of your games. This would be better for people right now who are screwed by the matchmaking system and end up going 1-5, but if they managed to implement it properly, would you personally be okay with losing two thirds of your games, even with a stacked premade?

The three team system is broken from the start and good matchmaking only averages out the misery caused by it.

BTW, I don't actually believe the current matchmaking system is working as it should be, because I know that players are not losing two thirds of their games. Some are winning 90%, while others are losing 90%.

The MMR system doesn't really have anything to do with whether it's two teams or three teams. There are plenty of games that have multiple teams... hell even a 12 person free for all mode can use MMR effectively.

I know it's super long, but I strongly suggest you read the whole thing.. particularly:

What is MMR based off of?
This is the thing we're the least certain of at this moment. I strongly suspect that it's based off of a lifetime cumulative total of your medal scores.... in essence, the sum of all your weekly leaderboard scores since you started the character. I believe this because players with both a high win rate and a low win rate seem to have high MMRs if they've played a lot of games...Similarly, players with high K/D ratios but also players with low k/d ratios tend to have high MMRs if they've played a lot of games... but my MAIN reason for thinking this is the case is because we already know that this is the system ZOS uses for the "Leaderboards" so we already know that they're tracking it.. and it would make sense for the "leaderboard" metrics (cumulative medal score in all games) to also determine MMR.... the difference between the "Leaderboards" and the MMR system is that the Leaderboards reset each week but the MMR doesn't... we know that players who used to play a ton of games but haven't played one in a month or two will still have really high MMRs. This theory also means that your MMR can't go down.... since you can never "lose" leaderboard points by performing poorly... so if someone does have their MMR go down, this theory will be disproven.

I read it. I have no idea what your MMR in ESO is based off of, but I doubt they're weighting medals, damage, or healing, and if they are, that might explain why the system isn't really working. These things don't have as much to do with whether your team wins or loses as you'd expect.

ELO in Rift was effective enough to pin most players at a 50% win rate and it was purely weighted by win/loss. Any player close to 60% was in the top 1% of players. Previously they allowed damage and healing to influence ELO but the system created too many outliers who were not where they should be and ended up with very poor matches due to that. I don't believe ZOS could implement it any better and a system based purely on win/loss would likely work the best.

ELO is supposed to decrease if you lose, but a good matchmaking system will use an adjusting k-factor to determine how much is gained or lost, with the best players having a lower k-factor so they can't sandbag out of very high ELO (losing very little ELO when they lose).

Yeah the system you're describing sounds way too sophisticated to be what we're going through right now. The current BG matchmaker is very predictable for me now, and I am 100% certain that win rate is not a factor at all. We've swapped out newer, high win rate players for older, low win rate characters.. and the older, low win rate characters have a SIGNIFICANTLY higher MMR.

It could also be based off of "# of BG games played" but would that really be distinguishable from lifetime medal score?

I edited my post with a bit more, but it would be nice if ZOS would make ELO visible so players could figure out exactly what is going on.

ELO systems aren't difficult to integrate. The algorithms are out there for ZOS to use. A system that weights damage, healing, medals, and other factors is probably much more difficult to implement than a basic ELO system. I would be very very surprised if they did not weight wins and losses.

Thogard

Regarding your edit - see my post about "What can we do to fix it"

If there are only 12 people queueing, even if those 12 people have different MMRs it's a bit ridiculous to make them all wait 30 min to fight...

The interval at which the system gradually increases its tolerance of mismatches needs to be shorter when there are fewer people queueing. It needs to be a function of the BG population, rather than a fixed rate like it is right now.

Australian premades like @damarky 's group are probably the ones feeling the pain the most.

ecru

Thogard wrote: »

Regarding your edit - see my post about "What can we do to fix it"

If there are only 12 people queueing, even if those 12 people have different MMRs it's a bit ridiculous to make them all wait 30 min to fight...

The interval at which the system gradually increases its tolerance of mismatches needs to be shorter when there are fewer people queueing. It needs to be a function of the BG population, rather than a fixed rate like it is right now.

Australian premades like @damarky 's group are probably the ones feeling the pain the most.

It does seem ridiculous, but that's the way these systems are designed to work. ZOS could adjust it so that after a certain period if time your MMR is weighted less and less so you can get into a match, but it's hard to say if they will ever implement something like this. I pushed for a system like that in Rift for well over a year and the devs would not budge, insisting that it would just lead to me getting matches during low population periods (like in the middle of the night) where I would eat the other team for lunch (and they were probably correct), so it comes down to whether ZOS is okay with matches like that happening or not.

This is also a factor of population and how long the current matchmaking system has been running, and might adjust itself as time goes on to be better for higher ELO players. Or.. it could get worse, if ZOS didn't implement a k-factor and your ELO is shooting to the moon, making it impossible for you to get matches.

That said, if players are waiting 20+ minutes consistently during primetime, I think the system is probably a little broken. It should not take that long to create a match since BG population is not that low. If this is happening, I'd get together with a group of players who are also having the same issue and try to get in contact with a dev or make a thread about this specific issue so that it's addressed.

edit: I should add that the issue I mentioned above was in regards to players solo queueing. If you're queueing with a full premade of very high ELO players, then it's not unreasonable to have to wait 20+ minutes for a match. The matchmaking system has to find two teams of players who have equal ELO, and that probably takes quite awhile. Sometimes it may not be able to create a match at all. Stacked high ELO premades in Rift would wait about the same time, or sometimes you would just never get a queue. The way we solved this issue was to find another group to form a premade and queue at the same time, and then we'd get queues instantly.

ecru

BTW, I think the above issue would be more easily solved by putting more players into consideration. If you were in a high ELO 4 person group in a system that was 8v8 instead of 4v4v4, it would be much easier to match your 4 person group with 4 low ELO players on your team, and then 8 average ELO players on the other team.

If when queueing with a high ELO premade the game is trying to match you with two other groups that are equal to your ELO, I can definitely see why it's taking a really long time for those groups to get matches.

tl;dr right now (i assume) it has to find twice as many players with equal ELO as the high ELO premade queueing, lol.

BNOC

So, if there's two teams queuing with a 200 MMR and a third queuing with a 140+30(170), but 3 solo players queuing with a 180 each - The game would start as A v B v 3Randoms before it ever got to A v B v C?

Solo:
180 + 180 + 180 = 180 Average.

PM:
140 + 150 + 150 + 150 = 147.5 Average

PM with penalty:
170 + 180 + 180 + 180 = 177.5 Average

How do you determine whether the randoms combined are stronger than the 177.5 premade?

Even if the penalty is being applied this way (+X MMR for 4 etc) - That team of 4 is more powerful, surely? - I guess the answer is, you still couldn't properly assess it. Are the PM using group builds and have a pocket healer? Are they 4 glass cannons that don't synergize well etc.

Given your assessment of MMR being a cumulative total of medals and score, MMR is not actually MMR - At least, not an indication of skill.

You're right that the interval system needs addressing if you're right about it.

Thogard

BNOC wrote: »

So, if there's two teams queuing with a 200 MMR and a third queuing with a 140+30(170), but 3 solo players queuing with a 180 each - The game would start as A v B v 3Randoms before it ever got to A v B v C?

Solo:
180 + 180 + 180 = 180 Average.

PM:
140 + 150 + 150 + 150 = 147.5 Average

PM with penalty:
170 + 180 + 180 + 180 = 177.5 Average

How do you determine whether the randoms combined are stronger than the 177.5 premade?

Even if the penalty is being applied this way (+X MMR for 4 etc) - That team of 4 is more powerful, surely? - I guess the answer is, you still couldn't properly assess it. Are the PM using group builds and have a pocket healer? Are they 4 glass cannons that don't synergize well etc.

This is one of the things I am 100% sure about because it comes straight from the patch notes:

"Battleground match making now adds an additional multiplier to the value of players who are grouped. This value scales differently between 2, 3, and 4 player groups"

But obviously we are still getting solo queues matched up with premades. And i know almost all of the premades on PC NA and have at least one of their members on my friends list (or can watch them stream), so I can keep track of this thing pretty easily. Soloers are definitely getting matched up with premades, even though other premades are queueing before the premades that go up against soloers... and remain queued for far longer.

Maybe I've tried to stream snipe fengrush's 4 man once or twice with my own four man and have queued before him but still not gotten matched up with his 4 man because that team is relatively new to the BGs. I have a hunch hehe.

But yeah you should know that I'm big on synchronizing queues and used to do a lot of sync'd queueing to get set matchups with other groups. I'll time queues with other premades and ask them how long they were queued before the match. @Galalin @del9 @Aliyavana @Gibbs and all the other BG regulars can attest that i often whisper them asking them questions about whether or not they're a premade and how long they were queued for

BNOC wrote: »

Given your assessment of MMR being a cumulative total of medals and score, MMR is not actually MMR - At least, not an indication of skill.

You're right that the interval system needs addressing if you're right about it.

Agreed on the MMR thing. Right now my group i smostly playing alts so we dont have to wait as long. It's ridiculous.. we get the same matchups we'd get on our mains, but because we're on new alts we dont spent 20 min waiting.

Darkmage1337

Very interesting read @Thogard, thanks for taking the time to put it together.
I'm an alt-a-holic so I find some characters getting into Battleground matches much faster than others (like my newer characters compared to my veteran BG PvP characters).

Hopefully @ZOS_BrianWheeler @ZOS_GinaBruno @ZOS_JessicaFolsom or anyone else at ZOS could share more information on the specifics of the BG MMR system, and implement some of the suggestions and changes here to alleviate the long queue times.

Adding a queue factor for more frequent time-intervals on MMR mismatches based on the current queue population sounds good, especially during off-hours.

I can't tell you how many matches in a row I face the same names / same people over and over and over again, it just shows how small the BG population queue is at certain times.

Thogard

Darkmage1337 wrote: »

I can't tell you how many matches in a row I face the same names / same people over and over and over again, it just shows how small the BG population queue is at certain times.

Yup! and what's interesting is that under this new system, even if there are a ton of other people playing, you probably won't fight them if those characters are more AGED than yours.

Even if those more aged characters are closer to your MMR than the vet's alts youre getting matched up with.

GrigorijMalahevich

give us MMR ladder

Gilliamtherogue

This new iteration has really taken a heavy toll on my personal enjoyment in Battle Grounds. BG's going base game was one of my biggest excitements coming with Summerset, but the reorganization of MMR for "fair matches" has resulted in the exact opposite for me.

Not only do I generally have to wait 20+ minutes to get into matches when I solo queue (and even mostly when I group queue too), but they're always against premade groups of players who are equal and above my skill level, giving them the huge advantage of organization. This has lead to some of the most unfun experiences in BG's since their release in Morrowind. Added onto the fact that we have to guess and experiment all of these systems on our own and have to clue where we're placed in the system, and how it relates to other players or groups, it becomes extremely discouraging to continue playing.

It would be great to see some adjustments brought to the current match making system, as well as some transparency so players can actually figure out how their progression is going with the rest of the player base. At this point it seems if you play BG's a lot and do somewhat decently in them, your end game participation will just be being ran down by 4 man groups all wearing the latest over performing sets (looking at you Sload's & Durok's) and ultimately hating your time spent there.

ecru

Gilliamtherogue wrote: »

This new iteration has really taken a heavy toll on my personal enjoyment in Battle Grounds. BG's going base game was one of my biggest excitements coming with Summerset, but the reorganization of MMR for "fair matches" has resulted in the exact opposite for me.

Not only do I generally have to wait 20+ minutes to get into matches when I solo queue (and even mostly when I group queue too), but they're always against premade groups of players who are equal and above my skill level, giving them the huge advantage of organization. This has lead to some of the most unfun experiences in BG's since their release in Morrowind. Added onto the fact that we have to guess and experiment all of these systems on our own and have to clue where we're placed in the system, and how it relates to other players or groups, it becomes extremely discouraging to continue playing.

It would be great to see some adjustments brought to the current match making system, as well as some transparency so players can actually figure out how their progression is going with the rest of the player base. At this point it seems if you play BG's a lot and do somewhat decently in them, your end game participation will just be being ran down by 4 man groups all wearing the latest over performing sets (looking at you Sload's & Durok's) and ultimately hating your time spent there.

I think this is a symptom of the small team sizes which allows a team to consist of a full premade rather than a partial one. Again whether it's 4v4 or 4v4v4, there's no way to fix this unless you throw more people into the mix to balance it out (8v8 or something similar). Since you have a very high ELO you're more likely to run into premades, but while facing that premade or premades you're still stuck with only three teammates who may or may not be anywhere near you--you'll never have the advantage of superior numbers or that premade having other "less skilled" teammates.

In no game have I ever seen a team that is an entire premade, regardless of team sizes, matched up against pugs. The potential advantage of being an organized group is simply too great when it consists of your entire team.

CO_Gibbs

Great work Thog, this confirms a lot of my suspicions.
I solo q’d most of the day yesterday, with 20+ minute waits, and ended up going against Feng’s premades 4/5 times in a row. When I grouped up I had even longer waits against similaraly skilled premades.

It seems like the best way to counter this is table my main until a patch is figured out and Q with an alt. BGs have largely been ruined for me based on how much I’ve played on my main, which kind of defeats the entire purpose of the Summerset changes. I didn’t have a single game yesterday that didn’t have my solo Q facing at least one premade.

Bfish22090

can confirm this is a thing.
queued with high MMR players yesterday and kept getting matched vs fengrush's premades (after a long wait, 5+mins)
then i jumped to an alt with low AR and low mmr and got queued up immediately against bad players who were not very good.

i wouldnt be that mad about this system if it want for high MMR equating to 5-10 minute queue times, even solo

Thogard

ecru wrote: »

Gilliamtherogue wrote: »

This new iteration has really taken a heavy toll on my personal enjoyment in Battle Grounds. BG's going base game was one of my biggest excitements coming with Summerset, but the reorganization of MMR for "fair matches" has resulted in the exact opposite for me.

Not only do I generally have to wait 20+ minutes to get into matches when I solo queue (and even mostly when I group queue too), but they're always against premade groups of players who are equal and above my skill level, giving them the huge advantage of organization. This has lead to some of the most unfun experiences in BG's since their release in Morrowind. Added onto the fact that we have to guess and experiment all of these systems on our own and have to clue where we're placed in the system, and how it relates to other players or groups, it becomes extremely discouraging to continue playing.

It would be great to see some adjustments brought to the current match making system, as well as some transparency so players can actually figure out how their progression is going with the rest of the player base. At this point it seems if you play BG's a lot and do somewhat decently in them, your end game participation will just be being ran down by 4 man groups all wearing the latest over performing sets (looking at you Sload's & Durok's) and ultimately hating your time spent there.

In no game have I ever seen a team that is an entire premade, regardless of team sizes, matched up against pugs. The potential advantage of being an organized group is simply too great when it consists of your entire team.

I see this all the time. It is extremely common.

But due to the low pool of players to draw from in the high MMR brackets, it might appear that they're a premade. Most of us know each other. But regardless of the fact that we know each other, a lot of the time we're solo queueing.

I make a point to ask people to verify my findings. Any BG regular on PC NA can tell you they've got the "hey did you queue solo or premade?" from me.

High MMR pugs going up against a premade is extremely common. It's just that the same pug group keeps getting groupe dtogether because of their MMR, so it gives the illusion of a premade.

Thogard

CO_Gibbs wrote: »

Great work Thog, this confirms a lot of my suspicions.
I solo q’d most of the day yesterday, with 20+ minute waits, and ended up going against Feng’s premades 4/5 times in a row. When I grouped up I had even longer waits against similaraly skilled premades.

It seems like the best way to counter this is table my main until a patch is figured out and Q with an alt. BGs have largely been ruined for me based on how much I’ve played on my main, which kind of defeats the entire purpose of the Summerset changes. I didn’t have a single game yesterday that didn’t have my solo Q facing at least one premade.

What we've been able to do is still get short queue times with 1-2 high MMR characters as long as there are also 1-2 low MMR characters. We still get put in the very top level matches when we do it, but at least this way we don't have to wait 20 minutes before the system becomes OK with the "mismatch."

Here's a great example of the kinds of hoops we have to jump through to get matches:

www.twitch.tv/videos/266004918?t=05h56m29s

Background: we have 6 people and we want to do a 3v3. We balance the teams based on classes and making sure each team has as close to the same healing / purify / DPS as possible. Then i realize that even though we queue at the exact same time, we're probably not going to get matched up... and i accurate predict what happens next... then we adjust the teams based on the BG games played of some of the characters.. and we fix it.

I talk a lot more about the queue in this clip, but this is the kind of thing we have to do to get matches... plz check it out:


twitch.tv/videos/266004918?t=05h56m29s (not my stream but you can hear me talking.. i'm the guy that doesn't shut up lol).

www.twitch.tv/videos/266004918?t=05h56m29s

Copy & paste that ^ into browser

Fake Remedy

To this @Thogard we should see population / wait times balance out for those with 15 minute+ queues, once players values has increased overtime due to success and games played?

edit: typo

Thogard

Fake Remedy wrote: »

To this @Thogard we should see population / wait times balance out for those with 15 minute+ queues, once players values has increased overtime due to success and games played?

edit: typo

That is a very good point that I had not considered...

I'd be worried that as new players are raising their MMR... many of the highest MMR players are no longer queueing up... But with the new bonuses we should see an overall increase in population.

What you're saying makes sense and is definitely something to think about. It holds up, logically.

NyassaV

Jules and Arya were in their own group and they got grouped with me and a friend against for 4 man premades

Thogard

NyassaV wrote: »

Jules and Arya were in their own group and they got grouped with me and a friend against for 4 man premades

Makes sense. Jules and Arya are both hard to gauge because they have so many characters and the names of them all look pretty similar to me But i know you probably have a mid to high MMR on your magblade so it makes sense you'd get thrown into that match against premades rather than a new player.

Fake Remedy

also when are your new values implemented? after each game? when the leader board resets? I feel the queues are better than last week.

venzzini

Great explanation Thogard! I am very casual and so like the quick pvp of battlegrounds. I solo que a lot unless asked to join a group. This patch has me to the point where I don't even feel like playing much.

I agree with everything Gilliam said about the situation.

Today I entered a BG after 30 minutes and was 3v4v4 and then had load screens for half the match.

I have been logging off before ever getting a que to pop. It's disheartening since I cannot play enough to have a chance at the boards at all anymore.

I just hope that the ones getting in quick que pops with lower MMR has thier MMR up quickly to help me get in faster.

ecru

Fake Remedy wrote: »

also when are your new values implemented? after each game? when the leader board resets? I feel the queues are better than last week.

If they're like every other game, the values are adjusted after every single match. I also still find it really unlikely it's based off of anything other than win/loss because a system like that would just be way too complicated to code and implement correctly.

If people care enough they should be tracking their win% to figure out if the system is actually creating good outcomes or if it's broken. If you're winning half of your games (it should be closer to 1 in 3), especially if you're solo queueing, it's probably slightly broken. If you're winning more than half or 2 in 3, it's very broken. Organized premades may be outliers when it comes to their win percentage at this point so they probably shouldn't be included in your individual win percentage.

Anethum

Thogard, great work on this thread, absolutly agree with suggestions

Thogard

ecru wrote: »

Fake Remedy wrote: »

also when are your new values implemented? after each game? when the leader board resets? I feel the queues are better than last week.

If they're like every other game, the values are adjusted after every single match. I also still find it really unlikely it's based off of anything other than win/loss because a system like that would just be way too complicated to code and implement correctly.

If people care enough they should be tracking their win% to figure out if the system is actually creating good outcomes or if it's broken. If you're winning half of your games (it should be closer to 1 in 3), especially if you're solo queueing, it's probably slightly broken. If you're winning more than half or 2 in 3, it's very broken. Organized premades may be outliers when it comes to their win percentage at this point so they probably shouldn't be included in your individual win percentage.

I track everything. Win percent has nothing to do with it.

Vapirko

Gilliamtherogue wrote: »

This new iteration has really taken a heavy toll on my personal enjoyment in Battle Grounds. BG's going base game was one of my biggest excitements coming with Summerset, but the reorganization of MMR for "fair matches" has resulted in the exact opposite for me.

Not only do I generally have to wait 20+ minutes to get into matches when I solo queue (and even mostly when I group queue too), but they're always against premade groups of players who are equal and above my skill level, giving them the huge advantage of organization. This has lead to some of the most unfun experiences in BG's since their release in Morrowind. Added onto the fact that we have to guess and experiment all of these systems on our own and have to clue where we're placed in the system, and how it relates to other players or groups, it becomes extremely discouraging to continue playing.

It would be great to see some adjustments brought to the current match making system, as well as some transparency so players can actually figure out how their progression is going with the rest of the player base. At this point it seems if you play BG's a lot and do somewhat decently in them, your end game participation will just be being ran down by 4 man groups all wearing the latest over performing sets (looking at you Sload's & Durok's) and ultimately hating your time spent there.

Getting pitted solo against streamer groups is rough. It appears you don’t even have to do all that well to get into that slot, or in other words there’s likely a large gap in the player base and most people either are very very inexperienced or very good, and those of us who consider ourselves decent often get pitted against the very good players probably because of the small population. My opinion of it ranges from I don’t mind because I’m a fan of so and so and I I know they’re a fair fighter and it’s fun to try to take them down once in a while, to streamer groups running multiple sloads and Zaan. It’s either that or playing against people that don’t have much of a chance. BGs is just incredibly hit or miss. A good close match between all three teams is very rare.