I don’t think the matchmaking system is perfect and has no issues. I think that your words don’t mean anything because you misinterpret your “clear evidence” due to personal bias and call yourself an expert on matchmaking systems.
If you wrote a poker matchmaking system, great for you! Even if that isn’t a lie, that in NO way makes you an expert on matchmaking systems, especially one as complicated as the overwatch matchmaking system needs to be.
I’m not arguing that the matchmaking system is good or bad, I’m only arguing that your opinions based on personal experiences are biased and unreliable. All my proof for that lies in your posts on this thread.
None, your suggestion is optimal (seeding players through Quick Play performance.) But perhaps there is something that is unknown to us that makes your suggestion impossible to implement. That’s why it’s difficult to intelligently discuss MMR related topics. Because everything is so cryptic.
I think the needs of the of the well established 2350s should trump that of Overwatch as a whole. Why lessen their experience for the “greater good?” Seems tyrannical imho.
Very much this idd. It is the same across all activision game, they use the same MM system. Do you know why? CAUSE THEY HAVE A FREAKING PATENT ON IT!
Really can not believe people still denying its existence. Only way to beat the system is to play the game once or twice a week. Coz than your internal MMR is soft rested. Tried it multitabal times. In all the latest CoD games as well. My overall win % is 50% in Bops4 for example, but if I do not play for 1-2 weeks, my first game will be against people that do not even know how to aim, my second a bit harder, and my third is me with non prestige players against prestige 2+. It gets really wacky you do not play for a long time, I mean its almost funny, if you are a good player that is, how bad your opponents will be.
Do not believe me? Do not play Overwatch on one account for 1-2 weeks and see what happens.
You want real data. Youre the bottom % of players in this whole game. Its almost impossible to find fair matches for the worst players in the game. Makes sense right? There is almost no one in this game as bad as you players, so how is the game going to find fair matches? Youre the statistical anomaly by being so bad. You want fair games? Play AI on easy, might make you better.
I honestly think the matchmaker is as noisy and terrible as it is due to cheaters. It’s trying to figure out the mmr of them and balance around them in all elos. It’s generating toxicity and people are misattributing it.
In all honesty at this point obtaining a better anti-cheat will lower toxicity and the knockdown effect on what the mmr system devolves into.
Easy fix – don’t put me vs. anyone with an SR more than 100 higher than mine. Don’t say queue times will be too long either, I don’t believe it and don’t care. I’d much rather wait 30 minutes for a FAIR fight than play two insanely unfair matches in that same amount of time.
this is an easy fix. you should be able to set SR parameters and it should give you feedback on expected queue times.
if we had QoL for the LFG you could 25 other people that play during your time and similar to you mindset. once you have these 25 friends, the match maker only needs to find 6 more.
The silver bullet for the matchmaker, is not to use it. but the tools they give us to cast off the burden of matchmaker are lackluster
and this is on the LOW end, the very bottom as I counted in hundreds instead of being exact. The average level I saw was 400 for my high end account and 200 for my low end account. The low end account also had substantially more leavers where my higher end had none.
“Role MMR is a hidden number that goes up when a player wins, and down when that player loses. How much it goes up and down is a complicated formula, based on many factors, but the essential truth of wins are good and losses are bad is certain.”
complicated formula, based on many factors, huh?
why don’t you just show some good faith and let us have peek at it?
oh, no?`
why?
because its CoMpLiC4t3d?
weak.
STOP HIDING THE MACHINE, IF THERE’S NO FUNNY BUSINESS; ONLY BORING MATH - LET US SEE IT!
Exactly. I’d love to audit that kindergartenkode. My 5 y.o. learned to make a matchmaker on her tablet in Scratch today. It sure doesn’t use TrueSkill/ELO because even kids know people who run that kind of system may as well run themselves out of a job.
Thanks for the data. It looks like your dps account is seeing players of about 75 levels higher than your tank account. Your dps account is also ranked a bit higher (about 130 SR). My hypothesis is that higher SR players have higher level, on average (because players become better with experience), and that is why you are seeing higher levels in your dps matches.
The alternative hypothesis, that Blizzard is matching on level, but only a bit, is just kinda weird. Why would they do this? If they decided to match on level, I would expect it to be a lot less subtle, especially for accounts under 100.
It would be nearly impossible to get exact SR comparisons and I don’t particularly believe that it’s that much of a factor. I can do game as support (1100: 1669, 67: 1632) and probably see the same result. [EX: Tank and support accounts starting match at +/- 300 instead of +/- 100 like it was before rolequeue. See Also: Overwatch Forums where I talk about the SR divide between players and overall being in the <50 of each other with only a few spikes into the 300sr range. Recently tanks and supports are commonly out of their rank, for faster dps queue times.]
With MMR the way it is - the closer to the middle you are the closer to mid level the players you should see as there are simply more people at a wider range of levels, if it were just random groups.
It might have something to do with the age of the account. My 60 account is fairly new (less than 6 months) while my 1100 is over 2.5 years at this point.
When I queued for tank (1700) on my dps account I also get really high level groups and it’s not an isolated incident. This happens ALL the time.
https://imgur.com/a/WWjMRS6
I tell ya, I’m not just talking out of my butt when I say things.
[edit:] The main reason I believe they do this is to try and mitigate the amount of smurf/throw accounts placed with older, more established, accounts to give a sense that there are less smurfs and throwers as people play more. I come to this conclusion because my newer account has a substantially more volatile experience than my older account. Don’t get me wrong my older account has problems sometimes but not nearly as much as my lower level account.
[edit2:] It would also explain the phenomenon of older accounts seeing more teammates leaving on the OTHER side and newer accounts seeing them leave at random or on their side.
I am pretty sure they are matching people based on levels too. But mmr has higher priority. You can very often see low level accounts on both sides as game is trying to balance levels too. But sadly one team have high ranked player on alt and other team new player often too.
I apologize for my slow response. Holidays are very busy for me.
It is often difficult to gather conclusive data, yes, especially if you are trying to explain subtle effects.
I guess you just didn’t want to play support? This would have been a significantly better test (and I indicated before hand that I would be suspicious of results where the SRs were not matched). It also helps that they are the same role.
People have reported all sorts of funny business with tank matchmaking. I would not be surprised if many of these reports are true, as the game struggles to make matches without enough people queuing for tank.