If SR was in any way an actual measurement of idividual skill then MMR is unnecessary. Simply fill teams with players of relatively equal skill…the SR! The MMR makes SR a bad represenation of individual skill. It’s not a conspiracy theory it’s elementary.
3 Likes
Actually a lot of people agree with this. Several of us would rather have one of two systems…
- Remove SR and show either MMR or (if the number is just too messy) a 1:1 function of MMR.
- Remove SR and only have tiers, similar to LoL. Each tier would be basically a set of 200 SR (or whatever makes sense).
Either way, though, you’re going to have MMR and “handicapping” in the way the OP describes it. People get so hung up on the idea that MMR is unknown they forget that even if it was known, no 12 people are likely to have the same MMR/SR so some balancing must be done.
A balanced match is the only way to show if the estimate of skill is right or wrong. It makes no sense to put a Bronze team against NYXL…you’re not going to be learning anything. That reasoning goes all the way down in differential to the point where you must not be able to know who will win using data.
2 Likes
This doesn’t mean the match is handicapped, it simply means they balance the teams for experience, in addition to skill level. It doesn’t mean they are NOT balanced by skill level, it’s both. They take players of similar overall skill level, and then sort by experience afterwards.
Often, given a specific overall skill level, the more experienced players will have better game knowledge, and the newer players will have worse game knowledge but better mechanics to make up for it. Given this, it makes sense that balancing teams for experience in ADDITION TO skill (to reiterate, not INSTEAD OF skill) will result in less chaotic, random matches, and will give people a better experience overall.
To put it simply, this controls for more random factors, allowing your skill to actually have a larger effect on the outcome of the game.
I hate playing a comp where the Dva doesnt understand that she is a pharra counter. How do u main a character and not know what they counter. Its like playing with retards in quick play
I always counter the Phara.
I think I know what Cuthbert is talking about, but I’m happy to be corrected. If you take the OP himself as an example here - with a gazillion hours invested in the game and firmly placed in mid silver as it seems - it’s fairly safe to say that unless something drastic happens this is his true rank and that’s where he will always be (until he maybe shifts some focus from spreading misinformation on the official forums to practicing and improving). Then when he gets matched with less experienced players at the same rank there is a likelihood that they are just at that rank because they’re still new and a fairly significant portion of those people are still improving and finding their way towards their “true rank”. Handicapping in this case would be the rising star with a true skill higher than their current MMR being matched with someone who is firmly placed at their rank after many hours of playing this game.
3 Likes
What’s elementary is why they have two separate scores. SR accepts limitations like decay and not changing too rapidly or slowly per win/loss; this makes it good for a visible system that rewards are based on. MMR works on less user-friendly rules that are ultimately more accurate at forming fair matches, so it is good for the actual matchmaking but not so great for rewarding players and would make rank gain/loss sometimes feel arbitrary.
Both tend to be rather similar values, but there is clear benefit to having both. And the people keeping this stupid thread going are very much conspiracy theorists.
2 Likes
They try and make each match a 50% chance to win for both teams. This means it’s a coin flip that is actually determined by random factors (someone throwing, someone having an unusually good game, someone disconnecting etc) and not skill.
2 Likes
You’re going to lose a game if a thrower or a disconnect ended up on your team randomly anyway, and I think someone having a good game would fall under the umbrella of skill.
1 Like
You know, after reading the comment above i had a thought; since tilt is a negative skill factor, how about the mm make players that have lost 2 in a row wait 5 minutes to re q.
I came in fresh today in a good mood, turned voice chat on for the first time in a long time looking forward to having fun and winning. Well 2 matches later, both were a toxic mess, i decided a long break from comp is in order.
The problem as I see it, is when someone is 500-800SR lower than their true rank and are placed in a match they have a 50% chance of winning, they are relying on factors that are mostly out of their control in order to rank up instead of the way it should be, which is ranking up because they are better than almost everyone at their current tier.
Before the white knights jump in: Yes, I understand that you believe if a match is 50/50 then they are already at their true rank, I don’t agree with this and so your point will go nowhere with me.
1 Like
I think it’s weird that the game tries to make fair matches, while matches should only be fair when all participants are at the same actual skill level.
If I were rated below my actual skill cap, I’d expect a greater than 50% win percentage, until I’ve attained the proper rating.
1 Like
That is correct and it’s the reason that people climb. The matchmaker doesn’t know your actual skill cap which is why as long as you are just a little bit better than your current MMR you will have a greater than 50% chance of winning (of course assuming that the other 11 players are actually ranked where they should be).
1 Like
There’s a saying, “When you assume, you make an ass out of u and me”
Don’t assume. Verify.
1 Like
Why can’t the matchmaker simply match based on your Elo rank, like the games of old did? Nobody ever complained about those.
1 Like
Sorry dude I don’t understand what you are saying. I said - “given the premise that the other 11 players are ranked exactly where they should be and you are ranked lower than your true skill the match won’t be 50% even though the matchmaker might think so”. However it would be silly to assume that in every game you play everyone else is ranked correctly wouldn’t it? Everyone knows that lots of people are ranked both higher and lower than their true skill so this is not what I was talking about.
But then again the matchmaker doesn’t know this either and all it can go from is their current MMR.
Not trying to get into the debate here again - I just answered your question and I think you misunderstood with what I meant by using the word “assume”. If you are interested in knowing why a random matchup is a disaster compared to a 50% pre-balanced matchup you can dig among my older posts in this thread and find one where I presented the results of a 100 000 player computer simulation that proved how a random approach to team balancing would be the worst possible idea ever conceived.
2 Likes
The problem is that, in my opinion, it should look no further than the MMR (or SR, if you prefer to rank based on that). It’ll be rare for people to have the exact same number, so you’ll try and find the closest match. But don’t, if the gap would be large either way, compensate by going to a more extreme on the other side of the spectrum.
Say we’re matchmaking for a person at 2750 SR. According to the rules, the limits are “can group within 500 SR”, meaning we end up with a range of 2250 SR, all the way through 3250 SR to pull from. There’s plenty of people in that range, so we’ll most likely end up with 11 other people, most of which will likely be less than 100 SR away in either direction. Just divide the people found between the two teams in a way that balances the two out. No rocket science, there.
Now, it gets a bit different when we factor in stacking, but it’s not that much different from what I described above. Instead of taking the person’s SR, we now take the stack’s average SR, and apply the same rules as before, except, we also try and find a similar stack, because it would be reasonable to assume the stack has better coordination, and is thus more likely to win. Failing an equal stack, we look at smaller stacks, and if we can’t find one for some reason, we’ll just fall back to throwing in people according to the SR limits specified earlier.
In theory, these will all be fair matches, assuming all people are at their max SR.
Now, what the current system does, and BS like decay and such doesn’t help, is match based on arbitrary criteria (as we can’t see MMR, but we know it’s not simply matching based on MMR, either).
1 Like
I don’t have a problem with taking that stance, and I’m glad you clarified with that last part.
I honestly don’t believe that the divergence between SR and MMR you speak of is commonplace to such a degree that it prevents players from climbing. I don’t think we have a lot of 3k MMR players that are stuck in gold, and are routinely being forced to play other 3k caliber players that are also stuck in gold.
I will concede it is theoretically possible, so I’m not about to call you some crazy lunatic that’s off your rocker, but from what I have experienced in combination with the information the devs have shared, it strikes me as something that would be highly unlikely.
That said, if I operate under the assumption that it is true that this divergence is common on the ladder for those that regularly play, then I will freely admit it would suck as a player, and I would give you my support in requesting that it be changed.
3 Likes
It’s not weird, actually. It’s how people are supposed to get to their proper rank.
The system doesn’t know how good you actually are or how well you’re about to perform in your next game. It can only make a guess (usually a pretty good) of what you could reasonably be expected to do. It puts you in a match with other people that it thinks are generally in the same ball park.
If you are a truly a Plat player that can perform consistently, but find yourself in a Silver match, then it obviously wouldn’t be an actual 50/50 match, even though that’s what the system was trying to accomplish. If you’re the only outlier, then you should win pretty easily, and then your MMR would increase accordingly. Then the next match it would try to make another 50/50 at a slightly higher point than your last, and the process simply continues until you end up where you’re supposed to be.
The same process applies when someone has an MMR/SR higher than they should have. Their 50/50 games aren’t truly 50/50’s, 'cause the system thinks they’re better than they are.
TL;DR: The 50/50 matchmaking philosophy is based off of what the system thinks you’re at, and not where you’re actually at.
3 Likes
If you’re playing a hero with carry potential, sure. But I’m a team player. I exclusively play heroes that depend on their team. My actual skill is therefore rather badly represented, and not at all of any meaningful influence on the games I play.
Seriously, a dumb Elo-only matchmaker would be so much better for this game. This is the first game with ranked play which has me question the matchmaker. I’ve played tons of Halo before, and didn’t have half the trouble ranking up and actually sticking to my skill cap, once I reached it. In Overwatch, it’s a neverending roller coaster ride. People who can’t be arsed to improve should get stuck, not go on this roller coaster that pisses off everyone that’s on it, and produces nothing other than toxicity.
2 Likes