Algorithmic Handicapping (MMR) is Wrong for Overwatch

Competitive Discussion
957

I don’t think we are going to agree at this stage ha ha :slight_smile:

1 Like
958

Blizzard pretends that every match is basically a coinflip with a 50% chance for each team to win. These 50% are calculated via algorithms, that TRY to GUESS your skill (MMR/SR). So there COULD be streaks of winning and losing, because there are only two final outcomes (sometimes a third with draw), but in reality the difference between winning and losing should be much more “random”.

So there is a gap between reality and theory. Therefore I see some possible flaws in the “math” or “algorithms”:

Unpredictable Skill
Either Blizzards “math” is horrible at “guessing” a player’s skill or we as players are performing so inconsistently, that it’s impossible - even for the best algorithms - to forsee any outcome of a match.

A 50/50 winning percentage leads to theoretically balanced games, which are absolutely not balanced at all (user experience).

Maybe after some lost or won matches, the math starts “running wild” because it can’t really estimate your “real” skill level (you are leaving your calculated skill floor). Then it tries to find “better” or “worse” enemies and mates for your next match. Eventually it fails in doing so.

I think it’s a common Blizzard fail: The pendulum swings TOO HARD in the calculated direction or isn’t even checked after every match.

Example: Your secret MMR is 2500. After 3 matches, you have 2600 MMR. Blizzard logic thinks you are too strong for the 2500 bracket and challenges you with 2750 enemies. You get rekt. Blizzard logic isn’t sure what to do and needs more data. It places you again against 2750 MMR enemies. You fail again. Your MMR drops to 2490. Blizzard logic is putting you back in the 2500 bracket where you get some wins.

The problem is: You have to perform OUTSTANDINGLY good for your estimated bracket to overcome blizzards match making and “guessing” logic. And with outstandingly I mean that you have to overcome a) the possibilty of unskilled teammates b) the possibility of extremely skilled opponents and to win a lot of games in one playthrough.

Edit 2: The problem is, that not everyone is able to always perform at their best. That leads to swings even within 1000 SR +/-. But in this system you either climb HARD and hit like a truck, or you are “forced” to play in your roster until you git very gud. At least for a winning streak that gets you to the next bracket. It’s possible to climb “slowly”, but it doesn’t feel like that, because your MMR and SR are so volatile.

Edit: the more I think about it, the more it makes sense to say, that blizzard isn’t using live data for match calculations. It doesn’t use your current MMR / SR for every new match. It seems that it uses some sort of threshold and stored data, which isn’t working very well and generates weirdly balanced matches.

2 Likes
959

I’d largely agree with you here. There’s so many things that can change. A player might have a good or bad game, a good or bad day, filling their favorite hero on a map good for them, or flexing to something they don’t play because it’s needed. These all can seriously shift a players performance.

If you expect every game to be to the wire, it won’t happen. However, in a larger sense, the game does a good job of sorting out the tiers. If you put two teams together, one with an average 500 SR higher, the higher team will crush them 98%+ of the time. Could you imagine how mad players would be if they, at 2250 SR, played a team all at 2750 SR? They’d be furious because the game was completely unfair. Which shows that overall, the rankings are good.

This, however, is factually incorrect. Your ideas about how the matchmaker work aren’t accurate. I encourage you to look up ANYTHING about how Elo/MMR systems work. But I’ll explain the basic issue.

So you’ve climbed your internal MMR up to 2600. Where you go wrong is the idea that Blizzard “challenges” you against different ranks. You will attempt to be matched against other players at your current MMR, 2600. Not only is this the best way to create a “balanced” game, because it’s the systems best guess of your skill, but it’s also the best way to find your true rank. Balanced games provide the most information. This is intuitive if you think about the extreme case, like a masters team playing a silver team. The masters team will always win. What does that tell you about each teams ranks? Absolutely nothing, obviously.

So again, all any MMR system does is match you by your current MMR, against other people with the same (or as close as possible) MMR.

I’m not trying to be rude, but this is just the outcome of people being terrible at understanding probability. You acknowledged there’s a lot of randomness in any players performance. This means badly balanced games and streaks will happen, even with perfectly even MMRs between teams. Human brains are almost TOO good at pattern matching, and have a well documented bad habit of finding patterns that don’t exist. So people tend to observe random patterns and come up with incorrect theories about underling causes.

There are people on this forum that are CONVINCED that if you play well, the system will “handicap” you and make your games harder. There are others that are convinced if you play well, the system will realize you should be a higher rank and give you easy games to climb. These are comically total opposites, but illustrate the issue well. Players see the same thing and draw opposite conclusions by flawed rationalizations of their games not being due to the randomness of human performance.

I really think a players SR should stay in a range of ~400, or 200+/- your true rank. About 200 SR you start to feel a notable difference in players, which will make it easier or harder to win. If you are swinging 1000 SR, you have some serious problems with consistency.

2 Likes
960

I disagree with you about a couple of things here. I think that MMR is actually good at predicting the performance of players, and assigning them so that either team has a 50% chance of winning each match. I also think it is not something the game should be doing, at all. Handicapping/balancing is wrong for competitive play.

1 Like
961

well if it WAS good, then there wouldn‘t be so many stomps?

my post was also just a guess. no scientific evidence included :laughing:

962

it’s possible to flip a coin on heads 10,000 times in a row. Improbable, but possible.

1 Like
963

MMR doesn’t predict anything, it just tells you where you are in relation to the playerbase, and this game does not have handicapping.

2 Likes
964

I love how everything in OP is purely denial to accept reality as anyone that actually takes time to study the system realizes all of OPs claims to be false.

3 Likes
965

I love how you ignore the popular vote, which I’ve won 4 times over. And if you think any of us can really “study the system,” you are delusional because Blizzard has revealed very little about how it works.

How can MMR “tell you where you are in relation to the playerbase” when it’s completely hidden, and nowhere in the user interface?

As always, challenges to my argument fail basic tests of truth and logic. I’ll see you in 2019, Blizzard fans.

Yeah MMR is designed to make matches more narrowly contested and not “stomps.” But in a game with human players, there’s bound to be exceptions to the experience Blizzard is trying to enforce.

966

“MMR works very similarly to SR. There are some minor differences that make it feel worse though, when you just watch that number. For example, it’s possible to win a match and not gain any MMR. We make it so that if you win a match, you always gain SR – even if it’s just a little bit – to feel psychologically rewarding. But MMR’s entire goal is creating fair matches – which isn’t always fun to look at and certainly not “rewarding” for players looking for pats on the back or a sense of progression. So SR “chases” your MMR very closely, except in a rare case of severe SR decay at GM/Masters/Diamond level of play.”- Jeff Kaplan

What you have tried to do, however, is attempt to argue that MMR is a differentiator between “good” and “bad” players at the same SR by using a Scott Mercer quote, which doesn’t even make sense considering the actual source of the quote.

There’s a reason the “REFERENCES” section of Kaawumba’s topic is extensive, while you refuse to actually post the link to the quote of Scott Mercer.

See, Blizzard is willing to discuss the workings of the matchmaker, but with rational players. They are NOT willing to discuss anything with intellectually dishonest people like you. So don’t start talking about truth and logic when your entire argument is 100% based on a lie.

3 Likes
967

the german’s were wrong…

968

You mean a vote that means nothing?
And is only in your favor because you take information from the devs then twist and manipulate it from what they actually said to fit your own narrative so that it co-insides with the sentiments of the majority(Who are mostly low ranked and blame anything but themselves).
And don’t forget tha part where multiple people who have actually took the time to study it have proven you wrong, time and time again. But of course you refuse to acknowledge this since you think a vote from humans is the end all be all.

Of course I’m going to ignore it as it means NOTHING

1 Like
969

_https://imgur.com/On08cLt
_https://imgur.com/aLPxXFI
_https://imgur.com/onzwjv

I just have to carry right?

970

Can you elaborate what you are trying to say? Also you kinda need something proving you right so it’s clear you aren’t just bullsh*tting!

971

Happy New Year everyone!

Voting is often the best way of making the best decision, and it’s a good tool for some tasks that would surprise you. If you’re jaded on the concept of democracy, I recommend reading “The Wisdom of Crowds” by James Suroweiky.

972

Only bronze and silver want mmr to be removed.If mmr wont exist Top500 will play with bronse on placements

1 Like
973

I thought this was the basis of competitive: A match of SKILL. After reading your post it does explain a number of anomalies that are discussed in and out of the forums.

I agree – the queue should exclusively look for SR’s as close together as possible, as I thought it had already done.

974

Rofl. If you wanted to make huge impact you should have convinced the lucio to go main healer or go main healer your self. With that kind of comp your just padding ur stats and you didnt even have 4 golds. Smh.

0 golds as spamrat? I mean medals dont really matter, but I cant see how you think you carried here.

This image doesent even work lol.

1 Like
975

Convince another player to play something they dont want to play?

I switched to rat towards the end as we were going to lose a 5v6

976

Thats not from the same game. And how about you flexing if you cant convince another guy to flex? Oh no I cant swap hog, Im farming medals.