A More Nuanced Take on MMR Bashing

Competitive Discussion
1

So we all see the threads bashing on MMR, and common rebuttals to these MMR bashing threads are “how come a Top 500 player can take a brand new account and get a 90%+ winrate and get back to GM no problem?”

Well, a Top 500 player would be successful no matter what. My issue with MMR is for the little guys (arbitrarily drawing the line at Plat and below).

I currently go between mid to high Gold on tank and support (sometimes breaking into Plat before I go on a 10 game losing streak), but unlike the stereotype, I don’t think I belong in GM. If I had to truly guess, I’m thinking I’m probably mid-Plat.

SKIP TO HERE FOR THE POINT OF THE THREAD:
My criticism of the MMR system is that it isn’t conducive to actually learning what you did, and didn’t do, right.

This harkens back to the common complaint of MMR where one game, you could be an absolute non-factor and still win (admittedly, this only happens in like 5% of my games), and another game, you’re popping off, and still lose.

To add more substance to this, about half the games (sometimes more) I play are straight up blowouts. Either my team blows out the other team or visa versa. And it’s even worse in Silver (I’m a Silver DPS) where I’d say 80% of games are blowouts (I HAVE noticed a slight uptick in actual competitive games, so that’s a plus)

So my question is how am I supposed to learn from my mistakes when most games are blowouts? Obviously, there’s something to be learned from every experience, but it makes it very difficult to truly gauge what you’re doing right or wrong when doing VOD reviews, when one team just steamrolls the other.

MMR is so hellbent on getting you to that 50% winrate, that it stacks the deck too much in one direction. I get it, when you’re doing well, it wants to make things more difficult to see if you can handle it. But unless your skill is well above the rank you’re currently in, it typically just ends up with your team getting blown out.

So again, my issue with MMR is that it makes it difficult to get better at the game (and I’m not using SR as the key metric for “getting better”) due to how it stacks the teams against or for you, based on how well you’re currently playing.

Speaking of SR, here’s a side note. I used to track my SR to see if there was a pattern and here’s what I’ve seen:

  • Process is typically as follows: win streak, trade wins and losses for a few games, losing streak, trade wins and losses for a few games, win streak…rinse and repeat. UNTIL you surpass your career peak SR, and you go on a MASSIVE losing streak. (It was a slow grind from Bronze to Gold as I don’t really have time to play a lot; but when I hit Plat as a tank, I IMMEDIATELY went on a crazy losing streak and deranked all the way to mid-Silver, where I had to claw my way back to mid-high Gold. Happened when I hit Plat on support too)
  • It’s tough to climb even when your winrate is 60-65%. Because the algorithm tends to punish you more for getting blown out relative to it rewarding you for doing the blowing out (even if you did very well during the game)…don’t get me started on SR lost when a teammate leaves vs. SR gained when beating a team down one. I’m all for the small uptick in SR if you beat a team 6v5. But I don’t think you should lose a lot of SR if you’re not the one who left.
6 Likes
2

There are serious issues with the system they currently use.

But you need to remember it is a very similar system to many games out there now.

1)Player engagement metrics influencing the matchmaker is BAD in what is supposed to be a “competitive” game. (Keeping little Timmy ‘engaged’ in Gold when he should be tanking in Bronze is bad for the game)

  1. Punishing players for swapping heroes is BAD, in a game of counter swapping (I am more in favor of no swaps in a game and better actual hero balancing… but Blizz clearly isn’t heading that direction).

  2. Random shuffling of 12 players with close MMRs into 2 teams is BETTER than trying to achieve a “close game” by “balancing” the teams as is currently done… players will actually get sorted to appropriate ranks BEFORE the heat death of the universe with random teams. Instead of “floating” in ranks they do not belong in.

3 Likes
3

Well put. You would enjoy this game way more if MMR was removed for competitive and learn so much more in the realm of real competition.

This is the brilliant study of the dopamine effect to make this game even more addictive. It gives you this false sense of winning (dopamine drip, which will get you hooked) and then when you start on a losing streak, you keep playing to get that “high” back.

MMR is designed to do exactly this and your tracking of SR is on point. My account alone can testify to this too. You’ll notice that most of these higher than 50% win rate percentages are on accounts that have pooled in maybe 15 games played? 25 games? Definitely not even close to 100 games played.

Removing MMR and allowing genuine competition to take it’s rightful place will allow for optimal development of the Player’s performance psychology.

#RemoveMMRfromCompetitive #DefendYourSR

2 Likes
4

you should post a replay code and ask for a review from other users. sometimes having an outside perspective is helpful vs reviewing yourself.

5 Likes
5

This is not how MMR works though.

Dev quote:

3 Likes
6

Why can I only give this post one like?

Well said.

2 Likes
7

Hmmm this is a very common rebuttal to the whole MMR debate. I know what the dev says…They can design something and have a certain intention for how they want it to work, but the end result is something completely different.

It’s tough to truly criticize since they keep their methodology a secret. But what’s most likely happening is their methodology is flawed. Whatever parameters they try to use to match people constantly gives games that is nowhere near a 50% win rate (both wins and losses).

4 Likes
8

A lot to unpack here. A few things I’d like the devs to explain:

  1. Are they admitting to handicapping games in order to give both sides 50/50 chances?

  2. How can they say a vast majority of games are 50/50 when there are so many team diffs? Not just stomps which can still happen if teams are fair but games that you could replay 10 times and the same team would probably win at least 9 of them.

  3. Are they saying that unranked to GM players get 50/50 games all the way until they hit GM? Then why do they win 96% of their games? Do they need to fire the bookies?

  4. Why are they admitting to knowingly making some matches that are not close to 50/50 and shipping them anyway? Dead game?

3 Likes
9

Bc the games aren’t really forced 50/50 I don’t think. I used to think the matchmaker was against me until I hired a former OWL player to coach me and realized I was indeed a huge part of the problem. Went from bronze to plat season 31 and was climbing in plat before the season ended, my win rate well above 50%

6 Likes
10

the matchmaker can‘t handle the biased data that is fed to him by smurfs and alt accounts. so the matchmaker THINKS the match is balanced, but most alt accounts have totally inconsistent data (on purpose), which more or less „hides“ the actual skill level.

it‘s a problem of multiple levels.

2 Likes
11

Well… I’d say you’re thinking correctly.

Some light reading.

2 Likes
12

I think it tries to make a 50:50 game but with the insane amount of Smurfs soft throwing combined with people who literally get triggered and throw after the first fight it definitely does not feel like it

4 Likes
13

how much did you pay him?

2 Likes
14

Apparently Muma does 30 mins for $15?

But interested to know as well.

15

Goes through a secondary website that has some discount codes therefore it varies. He’s unreal good though at coaching and obviously at the game

5 Likes
16

Which site? Which coach?

17

I am coached by Muma. Website is metafy

4 Likes
18

I have Muma on discord. He’s pretty cool. He gave me the replay codes for the tournament final

:smirk_cat:

2 Likes
19

Great guy! Incredible player and coach too.

5 Likes
20

The goal is to get 12 people into a match who are as close in MMR to each other as possible.
The “alternatives” would be to intentionally create uneven matches with an advantage for one of the sides, or to have no matchmaking whatsoever and simply throw 12 random people into a match.

The way ultimates work in Overwatch heavily increases the chance of a steamroll for one of the sides even if the teams are very close in skill to each other.
The ult advantage carries over to the next fight and often this is enough to stop the other team from ever gaining ground again.

On top of that there are many factors that the matchmaker doesn’t, can’t (and maybe shouldn’t) take into consideration.
Have a Ball one-trick on one team and a good Sombra on the other?
Pharmercy on one team and no good hitscan player on the other?
One team with players whose hero pools don’t synergize well with each other?
One team with two one-tricks for the same hero?

Tough luck, but I don’t see a viable approach to fix this through matchmaking without causing a whole lot of other issues that might be even worse.

The system obviously has issues to deal with people who are manipulating their Rank and MMR through different means.
They already confirmed that they have plans to better deal with in in OW2, though they haven’t said anything about how they are going to approach this issue.
I guess we can just wait and see :woman_shrugging:

It’s basically in the quote.
Not enough players in the lowest and highest skill brackets and issues with matchmaking when it comes to groups (probably especially larger ones with a high SR range).
The first one isn’t that much of an issue since the amount of affected players is extremely small, but they definitely have to improve the matchmaking for groups.

I hope that they finally offer a way to create/join clans in OW2 to make it easier to find people to play with.
That way more people would play in groups and they could probably also reduce the SR range. Both would increase the matchmaking quality quite a lot.

4 Likes