Should matchmaker use MMR or only SR?

I’m sorry but as it sits right now it’s not hard to smurf so I dont think that matters. It makes it easier for them, so what…if so.eone is going to smurf they are going to smurf no matter what. I dont know the solution for that…its one downside I’m willing to accept for the greater upside that having the clan system, reputation system, and SR dependent match making would bring.

1 Like

Here’s what I’ll say and I’ll leave it at that since I doubt you or I will ever be able to convince the other on this topic.

If you’re arguing that they should matchmake based on SR instead of MMR I can see it, use MMR as a part of PBSR to help push players out of brackets they are either under or overperforming in.

However, even if they did that I would still argue that they should split the teams with the goal of them being as even as possible. In an SR matchmaking case that would mean putting the highest SR player on one team then the 2nd and 3rd highest on the other, 4th and 5th on the first and so on.

Obviously I don’t know (and no one really knows) whether it balances teams with the draft style I just described or with the goal of making average SR/MMR even but the point is that there will always need to be some sort of system like that that guides the teams to a theoretical fair matchup.

1 Like

Delete POTG. Delete medals. Make registering a team a requirement for comp mode only. Developers a good reputation system to keep uncooperative players in check and create a good tutorial to replace the disgrace of a tutorial we have now

3 Likes

I’d be all for clans and the updated tutorial (keep POTG for the fun of it, players who want big numbers will still hunt them no matter what you can’t suddenly make them team players) but this is just unrealistic unless you make a separate solo queue mode. Not to mention that if you were always playing with the same team then you’d be open to just having everyone go up or down together since all variables on their team are consistent.

However currently or in any mode where solo is an option MMR and the systems that come with it are necessary.

That’s perfectly reasonable, because everyone’s SR should be ± 50 to begin with and we’re not incorporating trends to bias. I’m not suggesting that games should be deliberately unfair.

My issue with MMR based balancing results is solely with how it pertains to outliars. By pairing people on a winning trend with people on a losing trend, your games become more about ‘are you good enough to offset the thrower you got saddled with or smurf you have to face’.

If someone is 2200 SR, that says nothing about whether it’s their lifetime high and they’re ready to drop, their lifetime low and ready to climb, they’ve thrown their last 5 games, etc. It is just 2200 SR, and that’s all it needs to be. This makes climbing more akin to ‘are you better than the average of your lobby’, which honestly feels more fair to me. If you’re better than the average player in your lobby, you should be able to impact enough games positively to climb. If you’re worse, you should harm your team enough to drop.

1 Like

I think you’re wrong with that kind of thinking though. MMR does not radically change based on how you performed in the last couple games and if definitely does not take win or loss streaks into account either. All it cares about is how you perform on average compared to other players at your rank on the heroes you play in a per 10 setting.

It’s a black box, we can only speculate as to what it cares about. Even if it’s just average, having the highest per 10 stats of players in your rank should mean you’re due to climb to a rank where you are closer to the middle. Having the lowest per 10 stats should mean you’re due to drop to a rank where you are closer to the middle.

In neither case should it mean you get the opposite to balance you out, as that will always slow ‘correct’ transitions by pushing your stats back toward the middle.

1 Like

That’s the thing. I think people argue it from the standpoint of “they specifically look for good/bad players at your rank to balance you out” when in reality they grab 12 players and just split them. If you are truly due to climb or drop then you will by virtue of winning/losing more games because of how you perform. A 2100 player is not being given a 2400 player to balance them out at 2350.

I agree, whether the pool is MMR or SR this is what they’re likely doing.

This just depends how much MMR really varies from SR, and we can’t really know that. If MMR is hyper accurate, someone who is playing 2+ ranks above their current SR can only reach a balance by being placed with the 5 worst players in the lobby against the 6 next-best players.

Some lobbies will be more fair than others, but without hard numbers to indicate what percentage of players are playing outside their SR it’s pretty hard to say how many of them are able to balance a given player. Skill discrepancies within a given rank aren’t exactly small.

1 Like

Like we both have said we can’t know for sure. I believe they split teams in a draft style where it goes back and forth splitting players like team captains in middle school. But we have no way to know.

Especially since groups make it impossible to split some players.

I agree with the clans and reputation systems. I do realize people are going to smurf regardless of any changes made, sure what I said would make it easier for them, but they wouldn’t have to throw and ruin even half as many normal matches happening…

1 Like

They should remove SR/any type of points below GM and just have ranks and MMR like CS.

1 Like

One thing i don’t say often is that i actually think it’s kind of admirable that the Overwatch devs insist SO hard on making even matches. They don’t want their players getting spawn camped. The individual games are often pretty close. A lot of my matches go right to Overtime. It’s uncanny how well this game can cancel out our influence, whether i’m playing bad or good, it finds a way to hand me that even match. Granted, we’ve got an afk guy who’s probably wondering why it’s taking so long to throw the game, but hey… EVEN MATCH!

What’s frustrating to me is that i then lose the same SR as this AFK guy.
When i feel the reason he was on my team and not the other team is because my duo partner and i balance him out. Sure, we’ll make this an even match playing with an afk guy or the worst support player i’ve ever seen. Thus, the extremely close match despite the afk troll thrower or actual brand new player on our team seemingly 90% of the time.

If they used SR only, i think truly random games based on your SR alone would feel better in the long term. However, there would be a drop in match quality. You would have spawn camp games. Because it’s more likely that the 4 best players in the match would end up on the same team. Right now, Blizzard makes sure that doesn’t happen. I get why.

Ideally, they would KEEP MMR. Make the even matches. Fine. It’s probably more fun that way. BUT… crank the personal gains up by a lot. That way, you don’t lose the same SR as your literal thrower.

I just think there should be more personal SR to win (or lose if you’re playing bad). I guess overall, the game is most frustrating right now because it feels like the matchmaker and your random teammates have more power over your SR than you do. This could be because i have aged accounts that seem almost set in stone no matter what i do.

Even matches are fine, but you gotta give individuals a way to rank up or down faster. This endless stream of even matches that seem to counter balance your play gets really frustrating.

That is not the matchmakers fault though. Yeah you can argue that it’s harder to move an older account (no one knows for sure how heavily it weighs older stats compared to new ones, though I’d argue it should only look a the last 2 seasons or 100 games played in a category and ignore everything else). It’s the nature of a game that is so team dependent. You may make a key play here or there but rarely will you be the only reason that your team wins the game.

I’m not a huge fan of mixing MMR with SR. If the game is making match making decisions based off the invisible stat then the visible stat is superfluous and more over wrong. I have been in games where 3 ranks were represented and if MMR brought that match together rather then SR that shows a 1000 point discrepancy between MMR and SR. This deflates any argument of “you’re in the rank you deserve” because your visible rank doesn’t actually represent your skill only how many comp points you can potentially get at seasons end.

MMR is fine enough for placing you on the proper rung of the ladder at the very beginning but if MMR acts like a shackle to that rung then there’s no reason to attempt climbing. Anecdotally speaking my experience with the marriage of MMR and SR is days worth of SR gains negated in 1 - 2 matches. Why was I allowed to climb at all then? The MMR is saying I was boosted while solo queuing and throwing once the win streak ended. Why give me any SR at all for the wins? How was I even placed in those games if my MMR and SR are so far apart?

Invisible stats that dictate whether you’re playing the game right regardless of win or lose only encourage you to play the MMR not the mode. Which if I’m not mistaken is exactly the excuse for PBSR stopping at Diamond.

1 Like

I am a big supporter of this discussion, thank you for posting

2 Likes

You’re begging the question here. This is simply not the case. It’s one method, it’s not the only one. In professional sports in collegiate sports or in any league for that matter, they do not hand pick teams to ensure fair matches, so don’t say the way Blizzard does it is only way to do it.

This argument is far from an accurate comparison. Tell me in which of those leagues do they just collect a random group of players before every game? That’s right none. They are all preformed teams with none being strangers to each other. If OW had an all premade teams mode than it would be fine but get out of here with that “bUt iN ReAl SpORtS” fallacy. So don’t say the way professional teams are organized is anything like a group of random players thrown together.

You’re missing the point. The point is that in virtually ANY system of competition they do not pre-assemble and handpick the opposition. And you only find pre-selected matchmaking in gaming. That’s the distinction I’m talking about. What difference does it make if they’re pre-formed teams or not? The individuals on those teams still had to compete to establish a pecking order.

Everywhere else, they let competition itself sort out the pecking order. The matchmaker in Overwatch is in my opinion nothing more than the same type of grind mechanic you see implemented in other games to artificially pad playing time and prolong engagement.

And it would be “fine” even if Overwatch didn’t have pre-made teams. That’s the whole point here.

1 Like

Creating balanced match for two teams is matchmeker’s job. To create fair games. Meaning its up to players if one or other team win. But both teams should have chance to win. This fails only if players are boosted or smurfing which is not fail of matchmaker or MMR.

What do you mean by this? Completely random system will always be a lot worse than system which is matching players based on their skill. Which kind of comp game is not matching players based on skill or rank?

1 Like