Yes. The mmr right now is a little bit wack because according to the mmr equation miracle wud have a 100% chance to win against sumail. It needs to be adjusted but the concept is extremely sound. Eu mmr is extremely inflated.
If performance is the only thing that matters in calibration games then why do you have to win games to win MMR post calibration?
Hell even CSGO's calibration system takes win or lose as consideration
I agree with COME ON VOLVO. MMR is pretty accurate, but calibration can be abused to get higher than you deserve. Although eventually you'll drop to your true MMR so in the long run its a good measure. It might be slightly inaccurate at the very high end of the spectrum but for most players your MMR pretty much says how gud you are.
@OP, Technically MMR is just a number. That doesn't mean its meaningless though.
calibration is an issue.
if only KDA ratio, GPM and hero damage count, support players are the ones that are on the loosing end.
one can think that it should be that it should be possible to calibrate properly as support.
MMR does matter, when the difference is big enough. A real difference is noticeable at 1k pts gap at least.
I've used to play in lower 2k after the calibration and the people there definetely play way worse on AVERAGE than a 4k or upper 3k, for example.
I say average, because there will always be outliers, people, who tryhard against you, even if their MMR is very low. Those, who play better than their pts. But don't lie to yourself, they are rare.
All 4k and upper 3k is the same. 5k's definetely play better, but I guess upper 4k's can also be really good rarely, because some of them may be future 5k's, hehe.
but its possible to callibrate properly as support, just like its possible to get vhs by playing only support.......
Doing your job as a support doesn't mean you'll end up having sick ass KDA or GPM
I don't think the system can calculate your performance by warding positions, spell timings, rotations, or whatever that differentiates good and trash supports
How many people abused the system to get to vhs by picking core, and compare that to the amount of people who got to vhs by abusing the system by picking support
Probably like 1000 to 1 in comparision
@desolated
lol no. As a support calibrating you arent going to reach high gpm or kda. The metod does not takes in account things like the impact you have on the game, lets say, saving your teammates ass as dazzle, keep your partners informed of what is happening in the other lanes or other kind of things that I think they dont have a way to measure. I dare to say that if you play your calibrating matches as a support you are going to be 300-1000 under your real mmr.
Yea is possible to climb as a support. But in normal and high skill you cant rely on other player to carry your ass to a win. You better, pick a core OR pick a high impact support (not a babysitter, some shit to fuck the other team game, like a BH for junglers or easily gankeable mid or a silencer for channeled spells) and try to win the game not by direct help on improving your teammate game but with messing with the other team.
The reason because you find high skill cores that cant even get 50 lh min 10 in a safelane with uncontested farm is because of the metod they use to calibrate. Its just not accurate and is full of exploits.
Sorry for my bad english btw
i played most of my smurf calibration games as support and lost them, calibrated 4.4k
?
That's because you deserve to be there
Do you think diox can reach VHS by spamming pos 4/5?
@positivity
well you are 5.5k
The cap wasnt somewhere close to 4900? if thats the case you didnt get your real MMR
Thats what it takes to be a pro
Anything less than that means that the player is too shitty
Lowest solo mmr pros that come on my mind are loda and akke, both are about 6k and they rarely play solo ranked
I feel like there a slight difference on skill on average about every 500 mmr.I went from originally 2.2k and now I'm 3.3k . The biggest thing to note though is that there are dumbass who make it impossible to win at every mmr. Watch a 7k streamer, there are still people who make 1k miatakes, albeit only once or twice vs the whole game. But it still loses the match.
@sneaky
Thats because 5k+ punishes mistakes and weaknesses
While below that people just do what they're good at (or even worse, what they think they're good at, LUL HAHA ENEMY HAS 5 SLARK COUNTERS ILL PICK SLARK WHEN MY TEAM NEEDED SOME CC HAHAA)
before 6 or 7k
for example 5k player will shit on 3.5k player but miracle but the difference isnt nearly as big sumail (7.5k) vs miracle (9k)
My smurf only spam support pos 5 all the way and still able to hit vhs. So it doesn't matter whether u are using core or support to climb skill bracket.
as support you can have good kda, and even 450 gpm on a support is amazing. and if you are a really good support player worthy of max calibration, you will be getting that kind of stats.
also kda/gpm/xpm is not even what matters most when you callibrate(or at least play unranked raising your hidden solo mmr) , its the smart decision making and reaction time shit like that. and i base that by observing my own gameplay on the games which increased the skill bracket. and im perfectly well aware that im not capable of that all the time or else id never fall to NS.
Like I go 6/13 on Furion in NS, but my next game is HS because i did not even once failed to murder the slark with a simple orchid click.
what dendi said is applicable to 7-8k players
if u r 3k, u suck at the game in any case, be it team coordination or mechanical skill
being forced to win after calibration is not a weakness of calibration system, the cap was ntroduced to reduce the rate of account selling and abusing
mmr is like looking at the records of a sports team. it gives you an idea on how they match up against everyone else but it does not do anything to determine a game. A higher mmr player is more likely to win against a lower mmr player in the same way that a 7-1 team is more likely to win against a 4-4 team.
id say a 7-1 team playing against a 4-4 team has smthng like 80% to win max, while a team with average of X mmr has 95%+ chance to win against a X-1000 av mmr stack (not considering 8k vs 7k games)
well yes the skill gap between mmr is much greater and is probably more like a difference between leagues. But it was just an example
mmr not entirely accurate. you get +5/-45 mmr if the game has a 400 mmr gap so basically valve thinks a stack with 400 more mmr has a 90% chance to win which is bullshit. its probably 60% maybe 70% if its low mmr but 90% is false.
The mmr number is kinda skewed. The general idea is perfectly valid, it was first invented for chess and was made by legit mathematicians. However according to the current equation, which with testing matches up to the dot with irl mmr shit, miracle (9k) shud have a 100% chance to beat sumail (7.6k) in mid. The mmr inflation in eu along with the general equation fucks shit up a lot. Not to mention if miracle matches up with some 6k's, the whole team loses/gains the same amt of mmr despite the fact that miracles higher mmr (by 3k) shud make him more responsible for victory. He shud gain/lose more mmr than the 6k's. So yeah it needs some tuning but its otherwise very representative of skill. The scaling is just weird.
TL;DR: read above post
often people just act retarded, so predicting the difficulty of a game and thus how many points should really be won/lost isnt very possible
i would bet a 5.4 team with necro+abaddon offlane ember mid and weaver/luna/ld safelane with +1 (venge cm) is highly favored over a 5.8k avg team where the lowest mmr decides to pick qop mid and other garbage shit that happens all the time
matters in ranges of 500-800 mmr.
there is really no difference between a 2.8k player and a 3.5k player
it does matter in player skill. i used to be ~4,5k almost 2 years ago and now i just came back in normal skill. those guys are bot-like even im rusty af
https://www.dotabuff.com/matches/2964279767
this game i simply split push and took their rax. tped base when they came for me and won fights. easy
in most of my games, all carries get 4-5 cs/min at their best. i think the slark above was the best farm ive seen.
https://www.dotabuff.com/matches/2932988283
another example. my shit farm 314 cs/36 mins. enemy carry, free farm, less than 200cs.
also the mechanics are bad af. no tps and other stuff like that.
My boy u can pick any high mmr na player and compare to high mmr in eu, it doesnt make a difference if its miracle or sumail. And ur right, 3k's do have a huge advantage over 2k's, but the equation says that a 3.1k average team has a 64% chance of beating a 3k team, only 100 mmr difference, and that is clearly wrong.
And btw i did fucking google what skewed means
Skewed: make biased or distorted in a way that is regarded as inaccurate, unfair, or misleading.
I believe the current mmr system is all of those things.
Don't believe that retarded video, it's full of bullshit. Miracle is just as responsible as his 6k teammates for his wins/losses, you are all in the same game together and everyone has to do their part. If I get in a game with Miracle and for some dumb reason the system decided because I am shitter than him I deserve to gain/lose less mmr than him I'd just do run around and do whatever because who gives a shit about +5/-5. Not to mention that as a player who regularly gets games with low averages compared to my mmr it would be insanely punishing to lose extra mmr in those dumb 1v9 games that you basically have to solo carry anyway. Also in some of those games occasionally some of my 4ks step up and play out of their skins, do they deserve to get less mmr than me because they are lower rated?
If Sumail and Miracle met in a ranked pub and Sumail decides to random/run around as ES (reflecting his 7.5k rating) then ofc Miracle is probably going to win almost every time tryharding meta mid heroes. Anyway sub 5ks shouldn't even be allowed to have opinions about Dota 2 let alone matchmaking.
As a measurement of your ability to win pub games with randoms the current mmr system is basically perfect, higher rated players are better at winning pub games with 4 random players than you and lower rated players are worse. As a measurement of actual skill at Dota mmr is not very great but I'd say short of getting expert human judges to watch a bunch of your games and rate you as a player there is no system that could accurately do something like this. Also below a certain point ability as a dota player == your ability to win pub games, it is only when you get to the pointy end of the mmr scale that issues start to emerge.
A 1k can keep up with a pro game, but implementing it is their problem.
It's always funny when I see people say "watch a 6k play"
You need to practice, not just watch.
If I watch rtz play I understand everything he's doing, but I cannot (in game) think as fast, and implement all he does because he's practiced.
Anyway sub 5ks shouldn't even be allowed to have opinions about Dota 2 let alone matchmaking.
Mmr is ethos. It is there to validate ur argument, but it doesnt help if u spout bullshit. In your case i think everything except for this quote and one other thing that u said makes sense. However if u were to start randomly saying complete horse shit it doesnt matter what mmr u have. Its just wrong. Having a lower mmr just increases the probability of saying horse shit because ur less experienced and less skilled. But it DOES NOT mean u r always wrong. That is complete and utter BULLCRAP. I absolutely hate it when people say that.
If Sumail and Miracle met in a ranked pub and Sumail decides to random/run around as ES (reflecting his 7.5k rating) then ofc Miracle is probably going to win almost every time tryharding meta mid heroes.
Thing is sumail may fuck around, but look at the highest mmr in na, i think its QO but i dont remember. Its somewhere in 8k meanwhile eu has broken 9k. And out of all the 8k na players, dont tell me theyre all sumail copypastas who dont give af abt dota matchmaking. This argument wasnt supposed to be sumail centric i guess i just phrased my original post wrong. My bad.
Regarding ur counter argument abt weighting mmr gain i think ur right actually. Dota just has too many variables, in a perfect world each player wud be individually judged and assessed and whatever, but using stats is highly abusable, and valve obviously cant just watch every replay to judge peoples contribution. I guess the current system will have to stay because there arent better alternatives, however i still think it is flawed.
@kr thx for invalidating all ur points by devolving to an autistic animal who can only call people names cuz he cant formulate proper arguments. Makes my life easier.
Pro psaní komentáře se přihlaste.
Does MMR really matter ?