Lot of people seems to know exactly how MMR works (only going up and never going down, cumulative by medal score, considering only wins and not losses, etc.) which is quite odd considering that it was never shared by Zos any detailed information about how it works.
I would say the opposite: IMHO MMR works too well, because I've played thousands of Battlegrounds, I'm not in high MMR bracket by any mean, and I'm always put against the same people more or less (in every match there are always 5 or 6 people that i know by the names because I've encountered them hundreds of times). And no, I'm not in the BG elite club by any mean, even if I played thousands of Battlegrounds (real number, not exaggerating).
I don't know how MMR exactly works, but if it was up to me, I would have done something quite simple and effective: you just need to look at the APM (action per minutes) or the average damage / healing output of a player, and then put players together based on those average values. After a certain number of matches, you will have a pretty representative average of APM / damage output, and I'm pretty sure the matches would be quite balanced most of the times. I observed that, in the "typical" match (not considering outlier matches, due to the fact that after a certain amount of queue time the matchmaking system start to "relax" its thresholds and matching algorithms become more random) usually most players end up with similar total damage values. I'm usually sitting at 1 million damage per match, more or less, depending on the matches, and my opponents are most of the times in the same range, while I know for sure that there are players that are able to easily go above 3 mln damage every single match or even more, and I'm not usually put against them (luckily for me).
I'm to a point that I queue random group battleground from time to time even if I'm alone, because I realized that the group queue, having to match different groups with very different MMR individual values, has wider thresholds, and there is more variety (I'm put against new players almost constantly, sometime encountering almost complete newbies, closing matches with 20+ kills, sometimes encountering very well organized pro-premades that completely obliterate me and anyone else).
So I would say the opposite: in solo queue MMR works too well, to the point that it becomes boring and stale after a while, because matches become like a small club of 30 / 40 players and you tend to play against the same players over and over and over.

Category: would be nice to have, but would not happen

ELO is literally useless in every game because your true impact on a match is not measurable.

APM is a bad measure as people can just run around accomplishing nothing at all.

Damage means nothing as you can have a build that just deals a lot of AOE and when people are constantly brawling in a pile you can rack up tons of useless not effective damage into the millions.

Kills means nothing as classes like sorcs have access to better executes and are more likely to get death blows.

High healing just means you are countering the large amounts of aoe from the guy putting out 3M non effective damage.

Wins don't matter because people can carry you.

Losses don't matter because you can be a great individual player and get put with 3 people that have no idea what they're doing and have zero chance of winning.

You cannot measure how many times somebody recognized their team mate was getting assisted on and threw down their ult to save them, or streaked overtop of them at just the right moment to give them that extra second to stay alive. Basic examples of what separates skilled teams from unskilled which are not reflected at all in their measurable stats.

Even in games where MMR is considered to be effective you always get that one person on your team where you go "Huh? how did you even get to this rank playing like that". And it happens CONSTANTLY.

Open world is not ranked and why should BG's be ranked? It is just a faster way to engage in quick combat with others, instead of running to an area in a large open world and hoping to come across some people you can fight against. At least you know you will not be out numbered and that gives you a fighting chance.

The attempt to balance what cannot be balanced just leads to needless long wait times as it tries to match you against people it thinks you'll have good fights with, and then after the long wait time it fails to do so anyway.

Even with "proper" MMR how many matches are good? 1 of 5? 1 out of 10 if you're lucky where you go... "DAMN I don't even care which way this goes, this is a good match"... RARE.

At some point you are gonna have to admit to yourself that, while the pvp can be somewhat fun, it is just no where near a finished product and really more of a mini game.

Ratharel wrote: »

To be honest it feels to me like we do not have even the simplest matchmaking implemented at all. Each consecutive match feels like I'm thrown against different tier of players. First deathmatch yesterday - "wow, I'm a God" - something like 15/1/15 K/D/A. Second one, I felt like playing PvP for the very first time, can't even scratch the guys from other teams, ended like 2/15/2 K/D/A with some random kills scored on a guys with 5% HP left. Third one was a breeze and got me achievement for Quadruple kill. Fourth one, got beaten so hard that I didn't even score a kill and whole team ended up with something like 60 points.

this is my general experience from doing BGs casually (before all the ridiculous testing, i havent actually played BGs since october 2021, been doing more cyrodiil lol)

divnyi wrote: »

@Ratharel I don't think they have enough players for proper matchmaking. And situation you described can happen even if it's the same 12 players, just splitted into different teams.

A bad MMR System also reduced the players you can play against and needs enough players.
Atm MMR is the battleground points you earned on this char in all battlegrounds since last Reset. The more battlegrounds you do and the more points you get the higher your MMR gets. If you get 3rd and no XP in battlegrounds and have to do more for your XP your MMR raises faster. Somewhen you are pushed out of battlegrounds because you cant win anymore. Should replace this MMR with a better Indicator for strength. High ApM Players still loose against low ApM Players with Carry gear, so not a good Indicator for Strength. The lack of a good Indicator for Player strength is the Problem.
I usually do 1 battleground a day on average 2 chars to get daily 100k ap(more if need) and get matched against very strong players.

Aznox wrote: »

Just look at the desired output over several matches: winrate.

Winrate? In 3-team games? In objective modes? HAhahaha. No.

divnyi wrote: »

Aznox wrote: »

Just look at the desired output over several matches: winrate.

Winrate? In 3-team games? In objective modes? HAhahaha. No.

DM queue and Objective queue should get separated rankings. (Or DM only if the objective crowd doesn't want one, i don't care really).

What's the problem with 3-team DM ? (I mean "problem that would be specific to ranking based on winrate", let's not go off-topic)

Superior pre-made teams or individuals always demonstrated their ability to achieve better-than-random winrate.

Those individuals or groups should be consistently matched with each other if we want to see a more competitive scene/meta emerge.

Resulting data (ex: dedicated healers topping leaderboards, group-buff sets providing better results, etc..) can then be used to act on balancing based on factual and publicly available data.

YoureWrongImRight wrote: »

Aznox wrote: »

YoureWrongImRight wrote: »

ELO is literally useless in every game because your true impact on a match is not measurable.

Your true impact over a specific match may not be measurable, but that's not a problem at all.

Just look at the desired output over several matches: winrate.

If you have a better winrate than others at your current ELO, it just means you are performing better at achieving the desired output, so the system just pushes you up until your winrate is normalized.

So if I am playing a healer, and healing my ass off, constantly highest healing output in the match, but my DPS all hit like wet noodles so I can never win, that means I am bad and I should de-rank? or is it that my impact is measurable beyond strictly win loss or k/d?

One specific match win/loss is irrelevant, individual k/d is also irrelevant (a good healer will get a far better winrate than a mediocre bruiser in the current meta)

You need to think in average winrate over many matches, and compare yours to other players' in your bracket, let's say John and Paul.

Over 9 BG (or more, it's an example) the "wet noodle DPS" will statistically end up 3 times in your team, 3 times in John's team, 3 times in Paul's team.

If you are better than John and Paul at managing underperforming teammates, you will eventually outrank them.

Or you could all 3 be helpless with the underperformer, but you seized the opportunity better than Paul and John during fair/tight matches, you will also eventually outrank them.

Also, if the "wet noodle dps" really does bring down the winrate of teams getting him, he will quickly be set to a lower "ELO/MMR" and John/Paul/Yourself will stop getting him as a teammate.

Aznox wrote: »

YoureWrongImRight wrote: »

Aznox wrote: »

YoureWrongImRight wrote: »

ELO is literally useless in every game because your true impact on a match is not measurable.

Your true impact over a specific match may not be measurable, but that's not a problem at all.

Just look at the desired output over several matches: winrate.

If you have a better winrate than others at your current ELO, it just means you are performing better at achieving the desired output, so the system just pushes you up until your winrate is normalized.

So if I am playing a healer, and healing my ass off, constantly highest healing output in the match, but my DPS all hit like wet noodles so I can never win, that means I am bad and I should de-rank? or is it that my impact is measurable beyond strictly win loss or k/d?

One specific match win/loss is irrelevant, individual k/d is also irrelevant (a good healer will get a far better winrate than a mediocre bruiser in the current meta)

You need to think in average winrate over many matches, and compare yours to other players' in your bracket, let's say John and Paul.

Over 9 BG (or more, it's an example) the "wet noodle DPS" will statistically end up 3 times in your team, 3 times in John's team, 3 times in Paul's team.

If you are better than John and Paul at managing underperforming teammates, you will eventually outrank them.

Or you could all 3 be helpless with the underperformer, but you seized the opportunity better than Paul and John during fair/tight matches, you will also eventually outrank them.

Also, if the "wet noodle dps" really does bring down the winrate of teams getting him, he will quickly be set to a lower "ELO/MMR" and John/Paul/Yourself will stop getting him as a teammate.

This all sounds very nice in a cute little scenario with only a few players to work with, not with a player pool of thousands where you can get paired with people for 3 matches and they drag you down, then they log off and you get the next 3 that drag you down. Now before you know it I am actually lower MMR than the guys who logged off after only 3 losses and I am dragged down for an additional 6 losses so you are actually lower rank than them, and we can go on and on.

And this still doesn't account for the fact that you can be the best player mechanically at doing your job and still lose, you can lose a lot, you can be the "best" individual player and lose 10 in a row against measurably worse players because there's a team dynamic at play. You can often have players that refuse to play with the team, they refuse to target assist, they won't go for the objectives, you basically never see them.

It also serves no purpose to rank anything in a game where the gear/stats are not normalized. Are you actually better or is your gear just better? 99% of the time it's probably the latter.

MentalxHammer wrote: »

Cuddlypuff wrote: »

How will MMR even work?

+ ELO for win, 0 ELO for second place, - ELO for last place.

Sounds like the system where second and third place would fight each other to the last breath instead of crab-bucket the first one.

It can use the number of team victory points gained in the game as a basis of performance.

@YoureWrongImRight everyone are in the even conditions.
Random screws over? It screws over everyone.
Gear is op? Meta exists for everyone.

Lots of problems are non-existent in case of 4man premades.

Actually, won't hurt to have separate ratings for solo and group.

Vaoh wrote: »

BGs have two main issues imo:
• Some Objective game modes on large maps incentive avoiding PvP to win. This is a big part of the reason you get Objective matches people say F it and turn it into a Deathmatch. I can’t even blame them.

Separate leaderboard for objective, people who like objective will compete for it and find the best objective winrate.

• Not enough players PvPing in this game.... Cyro and IC have the same issue. It may change in the future but currently combined with MMR, this is why many times you get into matches with all the same people or no games at all.

If there were more people it would be awesome to have weekend-only game modes, or larger scale game modes too.

I don't see the problem, sure you'll play against the same opponents most of the time but matches would be a lot more fair on average than they are right now.

How does this factor in that some people just spend more time on the game so they have gold everything vs people that are just part time and have epic gear.

It gives them different rating as an outcome of the fight.
Yes, this is the answer to gear question. Always was.
This is fantasy MMO after all.