They put people into a match that is predicted to be 50/50, not forced to be 50/50. The game is then used to update the prediction for future games (that is, winners rank up and losers rank down).
Are you talking about visible SR? If yes, then this only happens when you play in a region/platform/SR/day/time with low population. I’m in high gold these days and I never see diamond or above players.
If you are talking about some theory where the visible SRs match and the invisible MMRs vary by 1000, then you are incorrect. This does not happen.
Again, predicted, not forced.
Mostly fair. SR is accurate to +/- 250 in normal situations, +/- 500 in unusual situations, for players that are active and not manipulating their SR.
So, in QP, placing me against all diamonds and masters at 2400 is predicted to be a 50/50 win rate?
What exactly is the difference between predicted and forced? The System forces a prediction of a 50% win rate does it not?
If you win to many games in a row, the system predictably forces you to go against better players until your performance rating drops.
And FYI I see games like this all the time because again I know how to manipulate the MMR system.
So is there any other kind of SR? Last time I checked everyone’s SR was visible until they locked profiles? I happened to go against all diamonds and masters last night as a matter of fact, and that 3700 wasn’t just a 3700, he was actually 3749 for the current season.
I’d be happy to explain to you how to manipulate the MMR system to ensure you go against these people more frequently, but you are basically telling me that it’s not possible. (Not possible for you at least, but entirely possible for me of course)
In fact, here’s a game with nothing but plats and diamonds and I almost NEVER play Zarya but you wouldn’t know by the card I received for that game: i.imgur.c0m/Lp3pzzY.jpg
(Off topic: that mercy on my team was beast mode and only a 2600 - But as we all know mercy is unplayable)
Here’s another example of that Forced…erm… I mean Predicted 50% rate. I was the highest rank on my team at like 2300-2400. Yes I actually felt bad for that 3100 too after I made him rage quit, I can’t imagine what his team comp must have consisted of if mine was the way it was. Guess he and I just won to many games in a row? And Matchmaker just didn’t like that, so it tries to force…erm… I mean predict a 50% loss.
i.imgur.c0m/BYlKgZy.png
And I suppose you think Elo hell doesn’t exist too?
What happens if I get matched with other people that has had their system be “mostly fair” to? Stompy?
Wow. I was going to make the point that “follows closely” is used as an argument as if anybody knows what “closely” means. Or “follows”. For Blizzard, 1% could qualify as “very closely” and that is 50 SR. Add that over 5 players and all of a sudden a high gold is matched with a silver that forgets that Orisa has a barrier or that Soldier can heal. For 1%.
For some roles, a difference of class (Gold vs Plat, e.g.) is a lot. For example, in Plat I can mostly expect that everyone can mentally count cooldowns and can and will use abilities correctly, whereas in Gold you still get tanks that back away from the payload or stay outside of the capture point, shooting in.
Add that to the fact that MMR doesn’t take into account your skill on the class you are going to play and it’s no wonder it does a bad job. And a lower player is more likely to insta-lock you out of your class, increasing the MMR difference between teams.
QP has separate MMR from competitive. If you try hard in QP, you can have a substantially higher “rating” there than in competitive. Overwatch Forums
If a player’s true rating is 3000, but his skill rating is 2500, he will be placed in (what the matchmaker thinks is 50/50) a 2500 match. He will then actually have a greater than 50% chance of winning, and will rank up over many games played, and slowly get harder and harder games until his true rating and his skill rating are in alignment and his true win probability is 50%. See Streamer Data - Google Sheets → Rainbow 10 to see this in action. This was Kabaji leveling up an alt (he filled each game). When he started, the game thought he was a gold player and placed him in gold games. With each game that he won, his games got harder and his rank got higher.
Umm. This is what it is supposed to do. No one wants to be your designated punching bag.
Just stop talking about quick play. It is separate from competitive and will only confuse things.
Elo hell is a somewhat subtle topic. Certainly, nothing is preventing you from ranking up except your own play. It isn’t your teammates that are holding you back. But if you expect silver players to be able to pull off grav/dragon and can’t handle it when they don’t, then you need to learn to adjust your play style for the tier that you are in.
Dude you F^$#in killing me to say that your team mates don’t hold you back. You must be playing something else lol.
People throw all the time especially in the Lower 2 tiers. What about those who do the Bronze to GM BS ? How about those who derank ? Combine all of that in the lowers tiers with the losers who just want to ruin your game and you have a game that isnt enjoyable anymore. Blizzard focuses their attention on the toxic players and not the trolls. FK it though let this game die its their fault.
100 or 10 difference would NOT constitute a kind of variance that requires handicapping. 100 is like 4 games difference on average. Think SR1500 and SR1600, SR2350 and SR2450, SR2950 and SR3050, and SR3700 and SR3800. SR4000 and SR4100 is where it starts to get iffy, according to streamers and pros that play there. You’d need someone like a SR2300 that has MMR2900 to require “handicapping”. (I still maintain that MMR doesn’t even look like a number that resembles SR, all according to Blizzard posts and Stylosa, btw)
Keep in mind, Cuthbert’s main point appears to be that players are matched by SR and then balanced by MMR.
QP MMR and Competitive MMR have always been and will always be separate. This includes your very first placement matches, where you start at approximately a Comp MMR-equivalent of SR2350-2400ish
You sound like the kind of players that post on here that want to have a competitive 85% or higher win rate and climb from Silver to Master while playing with and against the same Silver players the entire time.
You keep talking a lot of crap about manipulating MMR and beating higher-SR players without even actually playing comp. Maybe you should hit that Comp button then because you don’t sound believable at all.
Throwers, boosters, and trolls make games unpleasant, but they don’t hold you down in the rank you are in. To rank up, you need to be better than the average player at your rank, and you need to play enough games to overcome random factors. The average player in bronze and silver is quite poor and those are the players you should be aiming to be better than to rank up.
A loss done on purpose is a loss that shouldn’t happen your making excuses for a good game that isn’t good anymore. A hidden MMR is hidden because they don’t want you to see how flawed it is. Don’t be a sucker and drink the kool aid .
Yet when people on other threads complain “I won 5 games in a row and then I got dumped 6 games” the response, almost overwhelmingly, is “You were not good enough and got put in your place”
Which is it? 4-5 games “close enough” or “Worthy of a losing streak?”
Its still a sliding scale from lowest possible figure to the highest possible figure and can be compared to 0 through 5000 SR.
As Celsius and Fahrenheit.
GradeA’s weird juxtaposing of winrate and win prediction. (One is a ratio of all matches one is a direct prediction of the immediate match)
Technically better then the average player compared to your perceived MMR. As you are paired based on that with a SR range limit.
My goal here is not to give a complete description of the system. That is at How Competitive Skill Rating Works (Season 11). My goal here is to boil it down to the easy to understand essentials, and only answer the questions at hand.
I want to play heroes I can’t play in competitive. Because this is the only environment where you can test. So if I derank and people think I am trolling because I don’t play my mains, is this accurate? We are in season 11 and I made diamond like a 1000 times, just because I play a hero I can’t play for a few hours, I go on to a lose streak and then I get the hang of the hero and win again and try to get diamond again. Or I get bored with that hero because imo not every hero is fun to play. For example tanks, if I play tanks, its just trolling, because I never play tanks. I don’t like playing tanks so if I would start to pick tanks I would just be a troll. There will always be trolls without role based queue. And when someone insults me in my team, I don’t try to win neither. You could call that troll, but I am the last guy who would insult in a competitive game. For me the coms are for constructive feedback and making strats/synergies nothing else.
Your goal is to pose as an expert on these systems, which you are not. None of us have complete knowledge of Blizzard’s Match Making Rating or SR systems. Not so much as a mention of MMR is Overwatch’s user interface or terms of use. Only the glib statements that Blizzard has leaked out on these forums.
Blizzard’s use of big data, and their handicapping policy remain largely a mystery. When players demand an explanation, they’re not asking for yours Kaawumba, they’re asking for Blizzard’s. And from Blizzard we get no comment for a full year!
It might not be fun to play, but that still doesn’t hold your rank down. You’re slightly more likely to have a thrower on the other team, not your own.
Saying “my teammates hold me back” is like saying you lose at poker because “you get unlucky cards”. You might get unlucky for individual games/hands, but that all averages out quickly as you play more.
They have answered, though. That’s the thing. Kaawumba’s post links to multiple long threads explaining things in detail.
You are just ignoring that because they say what you don’t want to hear. That Blizzard just uses a normal MMR system, which everyone who understands how MMR works knows that it doesn’t “handicap” players.
Shockingly, Kaplan has better things to do than repeatedly tell you the same thing over and over while you ignore the truth to protect your video game ego.
The analogy is closer to a relay race team than poker. If one person on the team doesn’t want to win, or is weak, others have to make up that gap in performance.
Kaawumba’s post, detailed and referenced, points out that they have kept MMR hidden.
The formulae for both gains and losses (SR and MMR) are hidden.
The formula and limits for “Acceptable MMR distribution between sides” is hidden.
They have the right to keep them hidden. Just as consumers have the right to ask for them.
Blizzard has given a venue (The forums) to ask for change or information, semi publicly. Cuthbert has the means to ask the questions and request for changes.
It doesn’t mean Blizzard will answer the questions or address the requests. What is the harm to do so in your eyes?
Then why have it hidden? It is proprietary and they don’t want people “gaming” the system. Don’t need to get mad if people do want to see it though.
That and all the flaws it has maybe they should unhide and let the community guide them . The community has a better track record when it comes to things like this IMO.
Hidden MMR is just SR without decay. Nothing else. At diamond+ you lose and gain same amount of sr every time and matchmaking just makes a team with average SR on each side (ignoring decay). That’s it.
A relay race could fit if you ran hundreds of relays with random teammates. That’s the key in both poker and Overwatch. Sure, for any SINGLE game/hand you might be unlucky, but your success is not determined by a one event. (Unless you go all-in in poker, which isn’t a thing in Overwatch).
This simply comes down to understanding basic probability, and not letting your own bias cloud your judgement. No one is specially given more bad teammates than other players.
I’ve read far too much of Cuthbert’s writing and the vast majority of his energy is not spent asking for public code, but claiming the system is rigged and holds him down.
Which we can prove isn’t true even without public code.
This really ignores the fact that the two have separate gains and losses. MMR can stagnate with wins and losses, SR cannot. Making them two separate things representing skill, and connected in that MMR can manipulate SR gains and losses.
No, there is no 2 magical factors that influence matchmaking, its just SR or “sr without decay aka hidden mmr”. Its a very simple ± system that just adds and subtracts to make two teams have even average SR. One of the devs that worked on matchmaking tried to explain exactly that while sounding “smart” and making it seem as if he was doing some big brain work for the entire time he was on payroll which got a lot of people confused and the whole hidden mmr myth started. All that “hidden mmr” ever was is just SR without decay for people above diamond which is used in the matchmaking algorithm instead of displayed SR that can be decayed.
Its also rather amusing how all the conspiracy theorists on here blaming their lack of skill on teammates or other imaginary factors are around silver.