Why Handicapping (MMR) is Wrong for Competitive Play

Competitive Discussion
Prev 1 38 39 40 50 Next
02/08/2018 10:35 AMPosted by ForceComdr
While i don't think the system is "rigged" i do think it's broken at worst and unnecessary at best.

All this back end hidden voodoo accomplishes two things.

1: Foster conspiracy theories because no one truly knows how it works.

2: Encourages people to "play the system" in as much as they perceive it.

Look at some of the posts above. People saying they refuse to switch even if they're being hard countered in an attempt to control their MMR. It leads to people playing for themselves instead of playing for the team.

To say that the current system prevents one sided matches is laughable, and since that's the bulk of it's purpose there's really no reason to keep it.

Match 12 people according to their SR. No voodoo, no other complicated back end algorithms.

You get + / - X amount per game. This way there is no "playing for my mmr" you work together or you all lose the same amount, period. This is a team game, win by the team, die by the team.

Sometimes you'll get matched against people who are climbing and lose.
Others you'll get people who are about the same skill... coin flip.
Sometimes you'll get people who are falling and win easily.
... just like you do now.


All true, bang on.
Allow me to expand a little.

To all those saying people blame the system for their bad play...
This is true, it's easier to blame others for your failures; but this will occur regardless of what type of mm you have.

It doesn't matter if it's "rigged", "handicapped", or "working fine"

The simple fact is that it changes the way people play in a negative way.

I remember a big thread a while back that advised people,
"if you're gonna lose anyway, play badly so the system doesn't flag you as a carry in the next match."
It had pages and pages of posts agreeing and people saying they do it. Even if the entire theory was 100% wrong, it influenced how lots of people played.

I point again to the people in this thread that said they try to influence their MMR by changing their play (for the worse).

When you have this shadowy system that lurks in the background that no one understands, you will get a steady stream of nonsense like the two examples i just gave.

Eliminate the entire thing and you remove the incentive for people to play selfishly; and that will only improve the game for everyone.
02/08/2018 11:48 AMPosted by ForceComdr
Allow me to expand a little.

To all those saying people blame the system for their bad play...
This is true, it's easier to blame others for your failures; but this will occur regardless of what type of mm you have.

It doesn't matter if it's "rigged", "handicapped", or "working fine"

The simple fact is that it changes the way people play in a negative way.

I remember a big thread a while back that advised people,
"if you're gonna lose anyway, play badly so the system doesn't flag you as a carry in the next match."
It had pages and pages of posts agreeing and people saying they do it. Even if the entire theory was 100% wrong, it influenced how lots of people played.

I point again to the people in this thread that said they try to influence their MMR by changing their play (for the worse).

When you have this shadowy system that lurks in the background that no one understands, you will get a steady stream of nonsense like the two examples i just gave.

Eliminate the entire thing and you remove the incentive for people to play selfishly; and that will only improve the game for everyone.


Good point, thanks for articulating this.
FYI, I tested initial SR placement on experienced 4+star account vs new account. Placement matches on both accounts were solo que using the same support mains. The experienced multi-star account season placement is within 200 of previous season SR (10 wins! Working very hard and always gold medals.) On new account, initial placement is 1400+ SR above the experienced account with less placement wins.
I very seriously believe there is something wrong with the matchmaker and I support this "movement", but to be perfectly honest your opening post is extremely convoluted. I think you should re-think how you explain this, I understand the concept but it's so wordy and there's so many parts that just seem contrived and lost me.

I think you want the idea to be stated extremely clear, in a concise and digestible way so that more people can be on-board with you.
02/08/2018 03:47 PMPosted by S23
I very seriously believe there is something wrong with the matchmaker and I support this "movement", but to be perfectly honest your opening post is extremely convoluted. I think you should re-think how you explain this, I understand the concept but it's so wordy and there's so many parts that just seem contrived and lost me.

I think you want the idea to be stated extremely clear, in a concise and digestible way so that more people can be on-board with you.


Thanks for the feedback! I've edited the original post hundreds of times, and it has gotten a lot longer than it was in the original draft, but I have always tried to keep it as succinct as possible.

I think that most of the sections are necessary to my argument. But what sections did you not like? I'm especially interested in feedback on the opening sections.

I went ahead and shortened the bullet list at the start of the OP, because it was too long and there were a couple of weak items.
I've made more edits to the original post. I found some stuff to chop out of the opening sections, so I think it makes the whole thing easier to get down. I want to thank S23 again, because this is the kind of feedback that makes me take a more critical eye to my work. That is always the starting place for improvement.

S23, I am still curious which sections of the OP stood out to you as weak or unneccessary. Where do you think things get "convoluted?"

Also made edits to the sections on "Performance-based SR." I see this is a very important subject to players, and it is intimately related to my argument, because performance-based SR exists as a direct consequence of handicapping/MMR. But I don't know how I can bring the sections further up in the post, because I risk derailiing the main subject. If anyone has ideas let me know.
When a person with an English degree explains a math related problem but doesn't understand math.

I'm 100% sure all of the people that agree with Cuthbert must also have English degrees.

See what I did there? Correlation versus Causation.
If I had to choose between numeracy and literacy, I probably would choose literacy. But I don't see any value in choosing between them.
02/07/2018 10:28 AMPosted by Eclipse
I have lost nearly 30 more matches than won. Even though I have lost nearly 30 more than won having a terrible win percentage I still manage to stay between 2400 - 2500. Doesn't that seem a bit odd?


What level did you qualify at?
This makes a lot of sense and explains most of my frustration with this match maker.
While OP makes some good insights on certain aspects, he's wrong about handicapping.
Here are my win % each season:
S1 48% (475 games played)
S2 43% (224 games played)
S3 39% (173 games played)
S4 58% (60 games played) - I basically gave up on competitive from S4 onward.
S5 43% (86 games played)
S6 41% (41 games played)
S7 58% (12 games played)

Can't see the handicapping here. Also despite being often way below 50% win rate I did not drop severely in ranks (major drop was S3 where they artificially cut people's SR) - and NO, I never one tricked and I'm nearly always filling.

What OP proposes is anarchy mode - matching team of worse players against team of better players to ensure that better players do not lose - this way there will never be a close game - it would be complete stomp 100% of the time.

Now I can't tell what shenanigans are going on but it's definitely not handicapping as OP describes it. Just check your % wins each season (wins/games played) and you would see that it is nothing alike what OP is claiming.

Games feel like sh!t because MM principle is BS. It doesn't cover roles you play - so you get 4dps mains on your team, 3 mercy mains on your team etc. Team that gets more luck in this slot machine - wins, because they have more balanced composition with more mains in each role.

Fact, that there is no role pre-select makes whole MMR system pointless (MMR systems as such are not bad as OP says). For example - lets say I'm Genji one trick and my MMR is mostly based of my Genji plays. But if someone else picks genji (or I need to fill role) I basically play hero I'm not good at (or maybe even suck) - so my MMR is completely bollocks - because it overall it suggest I should manage in this game and play on the level of everyone else, but it doesn't predict I might have to play hero I suck with.

tl;dr - MMR systems as such best system you can get, but implementation of it combined with how MM works is the problem here. System requires more work, the problem is that Blizzard is not in a rush with that despite it's S8 already.

Also placement games are completely pointless (except for new accounts) - they serve no purpose here whatsoever. Also there is no soft reset (which is the case for example in League of Legends) which basically makes seasons pointless (because you pick up where you left - so what's the point?). Actually LOL had similar issue until they introduced role pre-select. Since that games were way less random, because each role has competent player for said role (not like ADC main having to play jungler and ruin game for his team).
As much as I understand MMR for creating competitive matches, I agree that MMR is being abused/missused by Blizzard. It feels like two scenarios are at work. Scenario one is your such a good player they balance it out by putting you on a team with lower skill. Scenario two the mmr is horribly calculated. Either way I support the removal of MMR and hope that it gets removed. PS Blizzard your minor changes per season to competitive have been a disappointing to say the least.
02/10/2018 06:39 AMPosted by scapegoat
This makes a lot of sense and explains most of my frustration with this match maker.

I blame this guy.
Outcast, you gave a thoughtful response and there are other things I want to reply to in there, but first I have to clear this up:

02/10/2018 07:34 AMPosted by Outcast
What OP proposes is anarchy mode - matching team of worse players against team of better players to ensure that better players do not lose - this way there will never be a close game - it would be complete stomp 100% of the time.


This is an over-extension of my proposal; a *reverse* function of Match Making Rating, to systematically create one-sided matches as opposed to handicapping matches as it does now. I would rather remove MMR entirely.

One-sided matches would happen more often under impartial matchmaking, yes. But those matches would be *proving* something about the merit of the participants. Players would have mobility in the SR system, and players would start to separate into ranks according to their skill.

With players and groups winning matches according to the effectiveness of their strategies, we would get better ideas of what actually works in this game. You would see more one sided matches, but you would also see less ridiculous strategies prevailing because games would not be placed artificially on edge.
02/10/2018 12:27 PMPosted by iamhappyso
02/10/2018 06:39 AMPosted by scapegoat
This makes a lot of sense and explains most of my frustration with this match maker.

I blame this guy.


Lol! Now I get it.
Cuthbert, I have found video evidence of a Blizzard matchmaking programmer admitting that he has written code where teams are split in order to make a 50/50 match. He did not say it was for overwatch specifically because the video was taken before overwatch was created, but we can assume if he has written it and it has been used it is possibly being used for Overwatch.

In other words, matchmaker specifically arranges teams to keep one team from being stronger than the other. This is awful, a disaster really for leveling up. It means no matter where you are you will have around the same difficulty leveling up. If you get much better and improve, matchmaker will simply find someone as good as you to place on the other team to offset your improvement.

Only if you suddenly increase in skill so high that there is no player available to offset your skill will you rank up, and even then it could place you with potatoes to offset your skill. But I am not so sure this is all there is to it, I think it looks at individual skill accomplishments to see if it wants you to rank up.

I'm going to be watch the video again and get the specific times and leave it here, I can't do it right now.
@S23, Jesus Christ! You must also have an English degree like Cuthbert, you guys classmates? Basically what you're saying is that we should make pro NBA players play versus high school players. Your argument is that you want unbalanced teams, that rank (1 2 3 4 5 6) should always face versus (7 8 9 10 11 12).

The other thing you probably never consider is the ratio to players per bracket. In Masters to Silver the ratio is 1:10 for 1 point of SR. Meaning that you're more likely to find someone of your equal MMR/SR number at the low brackets because of population density.
If they only cared about Competitive Mode as much as they do about OWL. I don't see any OWL team getting 1.3 points for a win. Seems like a pretty static one up and one down to me. If Overwatch does have a legit Competitive Mode it's OWL..
Woah this is not true. Well, if it was, who would play? If I was put against a GM of course that is handicapping, but this is not exactly right. Think about this logically. In a hypothetical sense, there is a Widowmaker on the enemy team. The Widowmaker has 70% accuracy and gets head-shots a majority of the time. They win. Well when it puts so said player in another match and the player's phenomenal character is Widowmaker, there is no way the algorithm could honestly suppose that they would play Widowmaker. The algorithm (unless its an instalock) can not know for sure if that player would be Widowmaker. Maybe they would bite the bullet and be healer and die multiple times.
If your handicapping accusation was correct, well, it wouldn't work effectively. Maybe the player is amazing at a select few, but there is no way possible that the algorithm could actually pair this person against you, since it would have to assume that the actual character would play that. Say your a Winston main, the other skilled fellow is a Reaper main. The system has no way of knowing that either of you will play that character. Then you could just switch heroes if necessary.
Therefore, you are also assuming that the system has something against you, I mean, YOU you. Well, how would this so said handicapping pick a target? It would not be possible to handicap everyone.

Join the Conversation

Return to Forum