Just 60 games so far excluding placement matches. I’ll get back to you when I hit a 100. My assumption is based on a comparison of multiple factors: my performance (overbuff and oversumo especially which compare you to other players of the same rank), the SR difference between the two teams and my own SR compared to the game average. I might be wrong. Further analysis with extended data should help getting a better idea.
1 Like
This thread is like the master of confirmation bias. The author simply denies facts and views or experiences that do not fit in his theory. The author has very little experience in the game from different situations where he would have seen strange phenomena happening with matchmaking.
Example of this is being a diamond/masters player picking up an established silver account. After only six to eight games which you could win by playing using your feet on keyb & mouse you suddenly notice that games seem harder and harder. You really have start trying to win but you are still at gold. You have to carry hard to win. 80% kill participation might not be enough.
No one on this forum has the slightest idea how matchmaking actually, in reality works. This thread can give you some hints on the basics but lacks any real information.
1 Like
Hey i have two things you could implement:
-
if someone from your team leave and will not go back in the time, others in team including you, can safely leave without leaver penalization and with - 25 sr. Even when everyone leave and you are left alone against 6 team, its better to stay and not leave, because if your MMR is higher than your SR you will lose less than 25 SR. In last occasion i had this, i lost only - 18 SR. Others who left the game - 25. I am not completely sure if it only affected by mmr to, i am pretty sure better stats in that game will matter too. So i suggest to farm kill from spawn with Widowmaker. Leave only if you know you underperformed or you know your mmr is low. Others wife, never leave games.
-
Other thing is not really practically usefull, because it can harm your team but its interesting. Honestly i have no clue why is it in game for so long and its not fixed.
If someone from one of the teams leave game and timer for penalization is out, you can safely leave game just with - 25 SR and without competitive time ban. You think this is normal right? Here is the twist, you can leave after the time limit even if leaver was on enemy side. It doesnt have to be on your side. You will lose only 25 SR. No competitive time penalization. You can go next game instantly.
Which part of his stuff is wrong exactly?
I cant say i agree with everything he wrote but based on my own experience and experiments most of it is right i think.
I for example completely disagree that players bellow Diamond are matched together in games based on mmr.
Most of the things in this topic are obvious but what I disagree with (like you) is how the matchmaking works. At one point Bliz implemented a change where the SR difference inside a team and between teams was significantly reduced. I don’t think there’s any other efficient enough method to implement that except for including SR directly as a factor. It might have been there all along but they just narrowed the range. It was even said in the blue posts that SR is used but the author chooses to ignore this (as he wrote himself).
In my opinion the matchmaking is more complex than we believe including numerous factors, most probably AI/machine learning etc. In some cases (like the dia/masters playing a silver account) the 50-50 win probability after a few games can only be achieved by a very efficient handicapping system. This leads to unwanted behavior for players performing above average at their current SR level. Also as a side effect it helps underperforming players stay at their current rank.
If you are better than your current rank you will eventually climb. No doubt about that. The mm just makes it harder and take longer than it should.
Read more here:
1 Like
This isn’t data. Link me a twitch stream or a record of things like SR, heroes played, who was playing the account, etc. and I’ll give it a look.
However, if your statement is just, “You think climbing should be easier” that isn’t really proof of much of anything. Especially when you can simulate a simple, unhandicapped, unrigged system that shows similar difficulty in combing: https://www.reddit.com/r/OverwatchUniversity/comments/aatezy/why_match_quality_is_frequently_poor/.
For this diamond/master player in gold, he is still winning much more than 50% of his games, yes? This forced 50/50 thing isn’t a thing.
Because of the leaver SR bug (see the bugs section) I generally recommend that people never leave, for any reason.
Why Bug? Its normal loss why it should be Bug?
Did you read the bugs section of the original post?
I dont think so, you mean in this thread?
Yes.
20 characters…
But its something different, you are talking there about next match after game with leaver.
What happened to me wasnt bug, its how the game works. People just dont know that you can lose less by staying and not leaving like others. Thats why i said it here so you can maybe implement this in your post so more people know it. Its another reason not to leave games. Well, if you performed well and your mmr is not lower than SR 
. I didnt find this in your post, but its possible i missed that. I dont i ever experienced the Bug you are talking about btw.
It was for the diamond placed acc. I didn’t tryhard for the last 5 games since 5th game had 2 masters during placements. If went full try hard mod, I’m sure it’d place me at 35xx-37xx
Unfortunately, I need much clearer data (or a developer statement) to put a theory like this in the main post. There are too many random things that people say, and too many random things that can affect SR gain, for me to go on anecdote alone.
IF this is true it’s explained quite simply by Kaawumba’s guide. After six to eight games the MMR has risen substantially but the SR hasn’t risen at the same rate - it’s still chasing the MMR. Given that matchmaking uses MMR rather than SR the matches get harder faster than the SR rises. Your SR is still gold but you’re playing with and against much harder opponents.
Could you please link that. I can’t find it.
EDIT: If you’re referring to the post by Scott Mercer, then I agree with Kaawumba that it is simply sloppy language. There is so much evidence for MMR still existing. The fact that this one post doesn’t even mention MMR suggests he used “SR” interchangeably with / instead of “MMR” - possibly as a simplification for the layman or possibly just being lazy.
If Scott was referring to a change to the system then he most likely would have talked about both MMR and SR.
I have no way of knowing how fast MMR moves. I know that when MMR is more than 50 away from SR, SR gains on a victory will be significantly different than SR losses on a defeat. As this usually doesn’t happen, we do have verification that MMR and SR are close for most people.
But yes, one possible exception is bronze to GM type series, so you may be correct that MMR moves faster in this case. The long term effect will be zero though, because SR gets dragged behind it.
Yeah, the stated scenario was diamond/master player playing on a silver account. I consider that to be the same as bronze to GM…except it starts at silver.
Also note that it was a big “if” I began with. I’m not convinced the story is realistic in the first place.
What do you mean by “theory”. Both things i presented can be tested. That thing with leaver on enemy side can be tested anytime when enemy team has leaver, just try it. I wish i recorded other thing because i really didnt expect you would not know it, i just thought you forget to implement it.
Btw you had no problem to write there theory about how are people matched in games by mmr, while blizzard said people are matched by SR.
I will record both these things if you want next time i will have them. And i already have like 500 recorded games on my Channel, you can see pretty much in every game that SR of both teams is practice ly same so i dont know from where is your theory about matchmaking by mmr coming from. With exception of placement for sure, placement. Placement matchmaking is working with mmr.
For the first one, I’d have to see enough data for it to be significantly significant. It could take 20+ games for this to be the case. And all the data has to be written down. If I believed every time someone said they saw something, my post would be a bunch of random contradictory statements about how things worked.
For the second one, I guess I could test it. But why would anyone want to leave a game when they are are up a player? Even if true, it’s pretty irrelevant to how anybody plays the game.
MMR is usually close to SR, so if matching is done by MMR, then SR will be close as well.
The one time that the developers have confirmed that MMR and SR are not close is decay. In this case, the SR of the two teams are not close. See Streamer Data - Google Sheets → Seagull 10, Team SR compared to Enemy SR.
Blizzard has also said that matching by ability is by MMR only. My statement on the topic is in the original post → “But I just read this post from a developer, and it said matchmaking is based on SR, not MMR”.