This game was completely unfair. I played like crap, missed a ton of lasthits, walked into 3 arrows, got selfish items, did not ward even once. Despite this I have KDA of 10 and easily won. The only person on my team who did not do well was storm and he kept afking and died standing still in the lane tanking creeps because he said he was eating.
This game my team was far more skilled than the other team. Matching failure.
this game has no hope. we have smting like 30 15 score but this kunka feeds 24/7 and destroys all our efforts (page 1 game). his last games are 0 -5, 1- 9, 0-5, 1-19 and so on. even game 1 of live games there are muted feeders, i remeber this afk farmer safelane sf feeder. so what is the point of geting higher mmr really?
This game was an absolute stomp. Was matchmaking just giving a free loss to Radiant? Did my team get better players for unknown reasons?
Nope, they just lost it in picking stage alone. Am, storm and invoker picked at the same time and none repicked, forcing storm to offlane. Instead of running dual supports to salvage the situation the fourth guy probably is like, fuck this shit I'll just jungle and do my own shit. Picks Wraith King of all things. Last guy ends up solo support. It was greedy as fuck.
Can we blame matchmaking for people picking like idiots? Player skill on individual heroes =/= picking sense.
@relentless - their lineup was horrible. Throw in how 90% of meepo players suck at him, it's no surprise your team easily wins regardless of minor mistakes. You can't blame MM for bad picks. Meanwhile your team's picks has amazing synergy with BH to scout out kills and a Naix-Storm infest bomb. You literally just had to press one button to guarantee a won team fight.
Stop judging someone's overall performance from one or two games. Imagine someone judged you solely from one of the games you fed horribly (everyone does that, from dendi to 500mmr scrub). Would it be a fair appraisal of your overall skill? If it was the first time you were trying some hero in ranked?
Yeah.. it would have nice synergy. But storm afk jungled or fed almost the entire game. He said he was eating.
We not only won the game easily. We did that while playing badly. Also the enemy Mirrana was pudge in my previous game and fed very badly... all the while he claimed he was 70% win dendi-pudge. Dotabuff shows he is 39% win. That player has lost 12 of his last 15 ranked matches. His MMR is just much higher than it should be.
But that is not his fault. It's Valves fault. MMR doing a poor job of accurately reflecting individual skill is one problem, but its a symptomatic problem. The real problem is Valve regularly allows large ranges of MMR to be on a team. That is the basic problem which leads to inaccurate scores in the first place because so many games are autowin or autoloss. Because so many games are lopsided it takes MMR much longer to converge and in fact can prevent it from converging to the correct score. Since win's are nearly always 50%, everyone tends to get stuck (not completely) at whatever MMR they happen to have after the first 10 games on an account. You can move off it eventually, but it takes a long time. It would take maybe 20 games if they did it right. But it takes hundreds of games to correct an error because so many games are certain losses or wins.
So only after hundreds of games does your incorrect MMR edge downward or upward. I'm not saying all MMRs are way off. Most are close enough. But too many are way off... leading to more imbalances. The root problem is Valve does not value match quality. They consider very low match qualities to be OK because the que time is lower that way.
Wait are we looking at the same game Zenoth? They have Mirrana, Invoker, Axe, Dark Seer and Meepo. How is that a horrible team? It has really powerful teamfight and strong lanes provide axe does not jungle (and he did not)... yeah the meepo probably can't be played by a lower skill player. But the picks themselves are fine.
Yes, my silencer counter pick really kicks butt against them... but not when my storm pays so little attention that he dies afk to creeps in the lane. He went to jungle because there he was safer afking. Most of this game Storm and LS are jungling while BH, WL, and I beat their whole team... and I didn't even play well. Lifestealer came out every time we had WL ult up and we killed 3 or 4 of them. Then back to farming. The other team was just out of their depth, had no idea what to do.... no leader, no direction. But they were the stack. We were the random solos. I still think the game was just unfair. And I know that Mirrana really wanted badly to beat me after the previous game where I pleaded with him to stop feeding.... which did not work. That was definitely his best effort on Mirrana, Valve just put him in a game with MMRs way to high for him.
-----------------------------------------------------------------------
Here is another way to look at it. All the stats are not on Dotabuff but you can get the idea.
Silencer +66 wins
LS +10 wins
WL +31 wins
vs ------------------------------------- Which player does not belong in this game?
See this is Valve's error. Maybe my storm was at -20 wins and had similar MMR to the Mirrana. But neither of them should be in the game then.
-------------------------------------------------------------------------
Too greedy. Meepo and invoker need levels and farm, axe needs farm, seer needs levels. Potm would be the only one functioning as support, but not necessarily a very useful one.
On paper they have strong teamfights, but you need a ridiculous level of coordination and execution to pull it off. In pubs those who can focus down single targets quickly often fare better than those relying on coordinated teamfights.
Of course, you may be right about matchmaking, I'm not denying that, though I've mentioned before that there will still be edge cases even when everyone has identical mmr. You would need alot of data to measure the worth of wait time against match enjoyment, especially when you will have stomps regardless, albeit less commonly so.
Well they were a stack (not the mirrana). They really walked into a lot of WL+ Silencer ult combos because they did try to be together and coordinated... but they didn't have the skill to pull it off. WL and Silencer ults have a long cd and there were many opportunities to punish us while those ults were down. But they waited so long to fight again ults were up.
I agree completely that stomps can happen with equal teams. And probably MMR for the teams was close in total. But it was not close for individual players. I think its unfair to put people into games with teammates of vastly different MMRs. If people put themselves on a stack with a large MMR difference fine...
But if I stack with some of my highest MMR friends...
In this case the random 3400 guy added to match my low MMR is totally screwed and feeds because he has no experience playing a super high level game. So he drags down his whole team. Yes, I'm very slow compared to those 5k to 6k players. But I know warding very well. I stack camps and play a hard support. I mostly hide and then cast my spells in a teamfight quickly before my being slow matters. The random guy has no clue what to do in a game vs truly excellent players.
Is there data that supports that valve matches games the first way over the other? I haven't really played any games where there was an opponent noticeably worse than the others to offset a newer player in my stack, so I cant really comment on this.
Yes, they showed a similar example when they introduced ranked Matching saying it was a "good" match. They also showed another example on the dev forum a year ago. Those matches had only MMR ranges of about 1000. But I have been in tons of games where my friends are way way higher MMR than I am. I'm sure other people do the same, maybe not as much as I do.
If Valve thinks 1000 MMR range is "good". Then they probably think 2k range is acceptable for those longer que times where the search range fills out.
Here are the criteria Valve uses for matches.
1.The teams are balanced. (Each team has a 50% chance to win.)
2.The discrepancy in skill between the most and least skilled player in the match is minimized. This is related to team balance, but not the same thing.
3.The discrepancy between experience (measured by the number of games played) between the least experienced player and the most experienced player is minimized. More on this below.
4.The highest skill Radiant player should be close to the same skill as the highest skill Dire player.
5.Each team contains about the same number of parties. For example, the matchmaker tries to avoid matching a party of 5 against against 5 individual players.
6.Players’ language preferences contains a common language. Lack of a common language among teammates’ language preferences is strongly avoided. Lack of a common language across the whole match is also avoided, but less strongly.
7.Wait times shouldn’t be too long.
This sounds nice but they also say right after this that...
"The matchmaker seldom achieves all of those goals perfectly. For any potential match, the matchmaker assigns a quality score for each of the criteria above and then takes a weighted average. When the overall quality score exceeds a threshold, the match is considered “good enough” and the match is formed."
Then they show a match with 1000 MMR range and call it good. They say "Players’ appraisals of matchmaking quality are highly correlated with their recent win rate." Well of course ppl won't complain about the super easy games they unfairly won.
--------------------------------------------------------------
Despite all this I do think most of the matches are fair. Probably less than 10% of them are really bad. I just think that 10% or 5% being unfair is unacceptable.
If the really unfair games were knocked down to less than 1% of games people would stop noticing them. Considering how frequently I get teams with people who can't or wont use English on US servers despite my "Prefrence" of English only being selected.... their "strongly avoided" must not be that strong.
Wasn't the match they showed the result of a stack? I also remember how they mentioned they balanced the game by the best player, not the worst, due to their data showing that a good player impacts a game much more than a bad player does (presumably he can at least call the shots even if not on a carry), so if a 3k stack has a 5k player there will likely be a 5k player on the other team, but if a 5k stack has a 3k player there wont necessarily be a 3k player on the other team.
As for what MMR range is acceptable, I don't really have much to comment.
Yes, But you Must remember. Party Que and Solo Que are the SAME que. My examples are for a stack also. In fact it is only weeks ago that Valve FINALY added a patch that is supposed to prevent 5 stack vs 5 soloes.
Almost 1 month now I guess, Dec 18th patch "- 5 solo players will never match against a 5 man party in Ranked Matchmaking (and will be extremely unlikely in Unranked)." So it still happens in unranked and has been for 2 years. So many times I tried to tell people that this was happening but they did not believe me... because its so obvious that matching should never do this. People think "valve can't be that dumb".
It is possible to put together a fair group of MMRs, even with the large range on one stack. But Valve tends to not do it. I suspect this is because their algorithm values goal #4 far to highly. Goal #4 is what always puts much weaker players on the team of elite players when they try to solo que. Its just a mathematical consequence of trying to match the average while maintain the high value. You pretty much end up maintaining the low value also.
Goal #4, screwing you over in solo que for two years running.
"4.The highest skill Radiant player should be close to the same skill as the highest skill Dire player."
Presumably from the data it means that if you pit a 3k-6k stack against a 4k-5k stack, the 3k-6k stack will win a lot more frequently than 50% due to the higher impact of the 6k player despite the average mmr being the same. Would it be fair to the losing team then? Remember that this can happen even in 2-stacks with high skill variation, which are quite common in solo queue.
I'm not taking Valve's side here, I'm just trying to see how their actions may be justified based on the data available to them. Of course, it may lead to other problems of its own, which are the problems some of us face on a day-to-day basis. But we never get to see what bigger problems might have been avoided since we never got to see them in the first place.
I think Valve theory crafted themselves into believing that 2 years ago and never tested it.
But what it really does is push the top players into a situation where they know their team will be feeders if they play solo. And any 5k +, even 4k+ player knows they end up carrying a huge weight often times in solo games... Meanwhile lower MMR players are occasionally stuck into high level games to balance some stack and can't help but feed while everyone shouts at them to stop.
As in the example I gave several posts above, it would be much more fair to spread that large MMR deficiency out across all 5 players and keep those 5 players close together in MMR. Then they will understand the same tactics and have the same expectations and skills.
Those players who are superstars at 6k MMR are going to dominate any game no matter what you do. They do even in pro games. Its not smart to try to force the match into a contest totally decided by the top MMR player Dire vs the top MMR player Radiant. This pattern contributes to the way page 1 live games end up.... a bunch of people all trying to solo play their way to a win because that is the experience they had... what they needed to do to win games.
This is why so often people are demanding solo mid in ranked games. Even if they don't understand why, people have that sense subconsciously that if they don't take control of a game... some nub will feed mid and lose it. It's because so often there is a nub on your team courtesy of Goal #4.
And these are only from the 2 matches I played yesterday. I think the removal of solo matchmaking and the split into solo/party mmr has something to do with this.
Zenoth still trying to prove his fail point... interesting, interesting.
"But what it really does is push the top players into a situation where they know their team will be feeders if they play solo. And any 5k +, even 4k+ player knows they end up carrying a huge weight often times in solo games... Meanwhile lower MMR players are occasionally stuck into high level games to balance some stack and can't help but feed while everyone shouts at them to stop."
So true. How many fucking games I had to carry trashes like ones talking here to win - and they are now mad at me cuz I call them bad.
There is no fail point for me to prove because there is no reasoned argument against it. Relentless's best explanation is that Valve pulled the data about better players having a larger impact on the game than noobs out of their ass.
Meanwhile you prefer to bask in the illusion that matchmaking is to blame for giving you noobs on all the games you lost. How could it be that you played any part in those? Blasphemy! Being in 100% of games that you won, all the hard work must have come from you since you were obviously a common factor.
On a more serious note, what you quoted from Relentless is only likely to have any modicum of truth to it if you end up blue or pink 90% of the time when solo queuing, which has been postulated to be the player with longest queue times - a.k.a. the player for which MM is having the hardest time finding a game for.
more of relentless thinking he is the 'best shit tier player'. I hate to break it to you but you are 2k MM for a reason and all your dota theory is great for giving information but all that information is useless if you can not put it into practice. Knowing the exact time to pull is great information but if you mess it up and pull late by mistake all the time you are no different than the clueless player who pulls randomly and pulls late. The result is the same, get off your high horse plz
@Relentless, there is just one big error in your argument. Valve drops you a question after games, rate how much you enjoyed the game. Guess when avg (95%) player press 5. Yep, when they stomp. Most of them will give low points to losses, whether it was stomp or close... so valve found a way to give fast queue time and maximize satisfaction. Not yours ofc, but most players.
Omfg, sorry bros I actually was the reason for the bad game.
Not 0-10 idiot mid, or wd was keep telling me that naga is carry, tiny is support. Or that they both didnt know how to roam and were farming creeps on different lanes.
I love it! "My team feeds in early game really badly? They must be a bunch if noobs, time to build midas to carry them!" Then flame team at the 20 min lose. XD
6k damage on a tiny despite going for a positioning item, when one combo easily clocks 1k+. Impressive. Choosing to go midas against a lineup particularly high in early nuke damage, arcanes instead of treads yet not ganking to relieve the pressure early. ...
No one is saying we are all 100% responsible for our losses. But we all contribute to it in one way or another, be it through mis-plays or simply the attitude we demonstrate to our teammates. To think one is faultless is simply narccistic and responsible for many of the plateau-ing of skill levels.
Sampson,
''The result is the same, get off your high horse plz'',i can tell you the same thing,i am 2,8k MMR player,fighting now for 3K,fighting hard,you can see my list of matches,i read you on forums a lots,always ''high horse'' probably becuse of high MMR after shitty calibration way of starting MMR,and choice of heroes,wow,Slark,BH,LC,Furion,did you chase kills when other team tearing appart your rax,or just run into invis ???
Relentless played 3x more matches then you and you are smart ass ???
Volvo made a lots mistakes,a lots people stuck on cca 3K rating,we play with people who have 45% of wins....They don't have a clue what is team game not to mention others stuff.....
I got winning strikes,loosing strikes,it's depends on other teammates,on low ranking matches diference in skill level is huge.
Point is,people with high MMR think they are really good players,i respect,when they have 2K+ matches and they are level 100+,but when they are level -50,or have new account and chasing high MMR with that account that is really poor and stupid....
You can add me Sampson,chose some you favorite hero,and come to solo line i will show you what 2.8MMR can do.....
Sry for my english if i made some mistakes.
I came in from dota 1 so my dota 2 matches do not reflect my total dota experience (not saying I am great but just saying). My KDA is avg at best so saying I am a kills chaser is flat out retarded.(my last match on BH I got a necrobook to push towers with). All I hear is blah blah blah my teammates are holding me back, you realize there are shitty players on your opponents side too? And lastly I will deal with your 1v1 nonsense, asking for 1v1's shows how retarded you are because you are asking for a 1v1 in a game build for team play.
3 russian idiots (top and mid), wisp that gets absoultely no wards and admits he is first time wisp, axe with meka (when we have wisp), qop building bkb 1st item.
Tell me its a joke. I know I'm good, but not that so I get trashes every game.
PS I just cannot understand how to argue with trashes like Sampson. Giving them all arguments, saying what idiots did wrong, etc. This retarded tells me that my shadow blade lost the game. As if I had agha 1st item that would be pure win game 100%. Really braindeads, learn to think before talking.
And it still looks like the russian stack got more work done then you and your friend. Midas against all their strong early fighters and by far the lowest hero damage among the cores despite being the most farmed. Why not say the Ember Spirit was so bad, died 11 times and got both MoM and Deso despite it not stacking?
I am asking people in each game what their MMR is, most of them are somewhat reluctant to answer, but from what I understood - with a 4.1k mmr I usually get matched in a same team with 3.3k-4k. It is not too bad, but could be better.
If it was a compulsory thing, rather than giving a choice between ranked/unraked MM, we would definitely have better matches, but, well, I think at least some improvement in the are is good.
Nah,
Just saying,MMR is good idea,and one day it will be really good system,but right now is far far away from even decent system,after every match i check all players,stats,win ratio etc etc,and when i see.....Sometimes i wish to play on old way,SD-RD without MMR,without spending nerves...
To explain how things works on low levels of MMR,when you decide to pick carry after you others pick another 3 and one support who play ofc semi carry way,so,the only way to have 80-90% of wins and climb higher i's to play offensive supp/pusher,farm your idiot noob carry so he can right click right click win some kills in mid/late game and say in the and ''comenT me'' Russian ofc....Wait to all pick their heroes,make counter for enemy team,try to make some synergy for yours.....
That is my advice for strugling/stucked people like me,for now it's working,when i get on 3,5K-4K MMR maybe the situation will be beter,right now is hell,also,don't play from 8.00 till 17.00 on EU servers,kids are just terrible.
@Valedithia If you misunderstood me on that point, probably others did also. Let me try again.
Valve says ""Players’ appraisals of matchmaking quality are highly correlated with their recent win rate." And I'm sure you are right that this is in reference to the post game questions on match satisfaction. The fact is people who lost badly are likely to give the game 1 star and people who won badly will give the game 5 stats... I won't, but most people would. This is a poor metric for judging match quality. The fact that Valve is using it at all proves they don't understand reporting bias. But like their other metrics they don't concern themselves with the metrics being accurate. Instead they say its not important as long as everyone is close to 50% win.
Despite all this complaining about matches being unfair... they still are far more fair than in dota 1. I played about 5,000 games of dota 1 and these games were maybe 75% one sided stomps.... some against me, most in my favor because I knew how to see before the game which team would win in various ways. In dota 1 you could switch teams before the game started so all the players who knew how to check stats switched to one team or quit before the game started. The result was that one team had all nubs who didn't know how to see which team was the winning team pre-game. The higher Tiers of ThroneIT and Dotacash random balance shuffled games... but ordinary pub games were all stacked for one team to win.
That stacking behavior was partially countered by smurfs who made new accounts that looked like nubs, but were really some of the best players who would then stomp the moderately skilled stack that set up against them. In dotacash there were some people who had over 100 smurf accounts. I had 13 myself. The stacking was reliable enough I could maintain 75-80% winrate on my main account which had about 2,700 games when I switched to dota 2. Now I can barely get over 50% ==> much more fair.
Valve's current system produces probably 90% fair matches. But some really are still bad. I think they could fix this, but over the past year I have become more and more disappointed that Valve seems uninterested in correcting or even detecting their problems. I had hoped that at least in Ranked games they would stop hiding the stats. The fact that they continue to hide MMRs even for Ranked where there should be no expectation of privacy to protect nubs... its just that they are embarrassed by the matches they are creating. Valve says they are still changing things all the time trying to get it right... But they have been doing this for 2 years and still are failing to improve. It's just getting frustrating.
^ Pfff. Why to talk to people that have wooden brains. Ignoring the fact given above only by deluding themselves that I'm bad (though I could pwn u with my eyes closed), believing in shit like fair mm only cuz it gives u free wins, etc.
Thinking is hard nowadays.
PS 90% fair? I'd say 10% for good players, 90% for trashes like mimiming in this thread about how they are getting carried to wins. But Valve actually did nice job, none of these guys understand they got decieved into thinking that mm is good.
I wanted to damn near cry last night. 4 stack. Single guy got captain. Didn't talk the entire time whilst picking, nor did he during the game. Gave no heroes the other players wanted (the only one we got was me on Lion). Such a pain in the ass.
@ Zenoth. I really do not believe that the "best player wins dota games" concept applies to very many games. I hope I can illustrate that this is true to your and others satisfaction.
There are situations... where one player is vastly superior to all enemies and wins the game almost totally alone. But these are exceptionally rare games. It is extremely difficult, especially in higher level play, to totally dominate a game by yourself. The best player may get an edge in his lane, may be able to turn that into something more with a timely rune pick up... but that cannot compare to the game impact of the worst player consistently feeding.
Pub dota games are much more regularly decided by the worst player... the one who fails consistently to match up to the level of play in the game. Who is out of position many times, who thinks they can make a play when they can't, who tries some stupid build and becomes useless, who farms for a long time missing lasthits, who afk jungles as the rax dies, who fountain dives and throws away the only chance to win... The pub dota game is typically won by the team who fed on the worst feeder, not the the team who's best player made some amazing play.
Goal #4 is based on the idea that the game is controlled by the top MMR players if such player are outliers. But it is not. It is controlled by the lower MMR players in the game if they are outliers. IF everyone is close, then you are much more likely to get back and forth games that are hard for either team to win. When there are a couple players who are much lower MMR... one of them will begin to feed.
They will begin to feed not because of a simple mistake, but because they really can't play against the superior players they face. And so the first feed will become many feeds because they cannot recover. When a player of similar skill to his lane opponents loses the lane, he has lost only a small advantage. He still might outplay them later. When a player of much less skill loses a lane he is just going to keep feeding. He could not win when it was even... with a disadvantage he is totally helpless.
The low MMR player controls the game by feeding first... and then continuing to feed. Maybe the other team also has a similarly low MMR player, but if he does not feed first and makes it to his core items he can still contribute. The low MMR player that starts badly does not recover. He is outskilled AND behind on xp and gold both. For the rest of the game its basically 4 v 5 and the game becomes a stomp as other players are dragged down by the one who began to feed first.
The Low MMR player may cause the loss in other ways that are less obvious as well. He may decide to be jungle axe and do nothing to help the team for 20 min at which point he has vanguard and tranquil boots and his team has lost all 3 lanes. Maybe Someone could then could out of the jungle with more farm (maybe a blink!!!) and start to turn the game around. But the low MMR player cannot. He can only begin to contribute too late, and badly and begin showing his low skill missing calls and not knowing when to attack or retreat, hiding behind his team watching them die, watching towers fall. At the end he has KDA of 2.5 and claims he did not feed. But he still caused the loss because he could not play at the same level as the other players in the game. He still was the proverbial "anchor". The low MMR player decides the game (when MMRs are not close together)
Let's look at this game just posted http://dotabuff.com/matches/470115379 The AM was not actually 1-10, but he clearly did fail far more than all other players causing the loss.
Here is a metric that normalizes the support and carry contributions so you can see how well they played independent of roles. This shows how much each hero was able to contribute to fights given the farm they required to do it.
ES 10k*KDA/networth = 1.15
AM 10k*KDA/networth = 0.38
DP 10k*KDA/networth = 2.65
Rub 10k*KDA/networth = 1.44
LC 10k*KDA/networth = 2.70
We can see that the KotL on the winning team made a very large contribution compared to how much farm he required. While the Antimage on the losing team required tons of farm and basically used it to die before doing much of anything. Was the Anitmage a much lower MMR than the other players? Unfortunately we don't know because Valve hides it. It could be that this AM tried his very best but had no hope of performing well because he was placed in a game far above his MMR.
When a large part of the team networth goes to the lower MMR player the team can't function. Does this mean the highest MMR player has to be the carry? It would not be a problem if all the MMR scores were close and everyone was only slightly better or worse at carry. But if one player is far worse at carry... and decides to play carry. Well now you are all screwed.
Its much better when the lowest MMR player has much less networth. You don't want the less skilled player to be required to make the big plays. So if I stack with near-pro level people I pick supports and we win 70 to 90% of games. I'm going to pick a low skill cap, reliable support like veno, lich, or dazzle or CM, a hero who is just as good casting within 0.25 seconds of the correct time as 0.1 seconds of the correct time... not something hard to execute at a high level like chen or enchantress. But some random person won't do that. They won't even know they are the low MMR player until the game is going for a while so they might take mid or carry and be unable to fulfill their role.
I think the problem is more serious that mmr doesn't reflect player skill. As kawaiisocks said if you ask around in a match or inspect players and they don't have it hidden the range isn't usually that big. I'm like 3200 solo rating and most people I can tell their rating are 2900-3400. Not a big swing either way. However many of them are substantially worse than me from a simple understanding the game perspective. For example they random last picks, get a terrible pick for our team, refuse to repick. Supports fight carries for last hits. Supports don't ward. You don't even need to counter ward they just don't ward at all. I understand not wanting to play support, I get it, I stopped playing nearly as much too but I always pick my carry with 1-3rd pick. If 2 or 3 carries get picked I won't be a dick and go a 4th or 5th one but players in this range do so ALL THE TIME. I had a match first pick was random bounty hunter, I picked viper and said I'll go mid or lane, next pick was dp for mid ok all's fine. Last two are mirana and anti mage cus they wanted to lane that combo together for some stupid reason. Needless to say AM had no farm and we got crushed cus we had 1 stun, no supports and no team fight outside of dp ult.
I just don't get why people who make dumbass decisions like that are worthy of what is supposedly a medium high rating in the 3000s. But somehow these losers have the same rating. Not saying I always make the best picks or lanes either but I generally win captains drafts when I draft.
The term carry is misleading to alot of players while mid/late game is mostly the carries job, it is the job of the supports to carry the carries to that point. While the term 'creating space' is often joked about this is one of the key jobs of the supports which does not get logged into the stats screen at the end of the game.
Generally its because they are more skillful at clicking. Being better at clicking in dota can win a lot of games. Game decisions and knowledge can win games too. And people who win games in either way tend not to understand the other's weakness.
The people who win on decision making think "How can you be so dumb as to do that and throw!!!" While the people who win on skill think "How can you be so bad!!!" There are different ways to get to the same MMR.
There is nothing MM can do about people picking heroes they should not pick. If they do this commonly their MM will go down which will reflect their game success. If I played venomancer over and over my MM would go down not because I became a worse player but because I played something I sucked with. Maybe I have it easy because nobody really fights for the offlane......
I don't have the data to contest what you say, and any evidence I have is anecdotal in nature. You may be right, I do not know. Valve claims they have the data, and it is also in their interest to at least use it, so my assumption is that they are acting on it.
@try
Winning has alot of factors. Mechanical skill is easily one of the biggest factor - how well you lane, how good you are at last hitting, how well you execute combos (e.g. Timbersaw retract chakram while whirling on a chain) - good mechanical skill alone can easily get you to 4k+ mmr. However, game sense, picking sense, awareness, leadership, personality - all of these are significant factors which affect your ability at the game too. MMR merely demonstrates a rough aggregate of everything I just mentioned. Often, if someone is picking like an idiot, you can probably assume that he is relatively more capable in other categories.
I assumed basically the same things a year ago. But over time I've just seen too many bits of evidence that indicate whoever is doing the matching tweaks just is not up to the task of figuring it out. The way they release so many little patches with more and more bugs... like "oops we forgot that" and "oops missed that" and "oops didn't mean to do that" every time they do a patch. It's just sloppy. They don't even put those errors and corrections in the changelog anymore they just slip them in silently, for better PR I guess? I just ran out of "benefit of the doubt". And when they gave me 7 +0 points wins in a row near the beginning of ranked matching it just became very clear that they did not have a handle on putting the matches together.
Now Valve has announced that "- Ranked matchmaking rating changes are no longer dependent on match length." after they just went on and on about this fancy metric (time integral of the gold difference, measured since the last point in the game where the difference was zero) they had was so great. I explained on dotabuff that it meant you would get the most points for a longer match where you were ahead the whole game but did not end the game. Now 2 months later they realize its bad and take it out. I mean, seriously I explained the impact of the metric on gameplay without even seeing the code when it was first announced and it takes them two months to realize that this means longer games are more points?
As someone who plays support the majority of the time, it REALLY bugs me when people are calling for wards & in the same breath, get upset that you take last hits. I generally try not to, but there are items I need to do my job effectively as a support, too.
http://dotabuff.com/matches/469307127
This game was completely unfair. I played like crap, missed a ton of lasthits, walked into 3 arrows, got selfish items, did not ward even once. Despite this I have KDA of 10 and easily won. The only person on my team who did not do well was storm and he kept afking and died standing still in the lane tanking creeps because he said he was eating.
This game my team was far more skilled than the other team. Matching failure.
^
u were losing too much so valve have to give u a win somehow. ur losing streak will continue now
this game has no hope. we have smting like 30 15 score but this kunka feeds 24/7 and destroys all our efforts (page 1 game). his last games are 0 -5, 1- 9, 0-5, 1-19 and so on. even game 1 of live games there are muted feeders, i remeber this afk farmer safelane sf feeder. so what is the point of geting higher mmr really?
http://dotabuff.com/players/87085280
http://dotabuff.com/matches/461049767
http://dotabuff.com/matches/467614436
This game was an absolute stomp. Was matchmaking just giving a free loss to Radiant? Did my team get better players for unknown reasons?
Nope, they just lost it in picking stage alone. Am, storm and invoker picked at the same time and none repicked, forcing storm to offlane. Instead of running dual supports to salvage the situation the fourth guy probably is like, fuck this shit I'll just jungle and do my own shit. Picks Wraith King of all things. Last guy ends up solo support. It was greedy as fuck.
Can we blame matchmaking for people picking like idiots? Player skill on individual heroes =/= picking sense.
@relentless - their lineup was horrible. Throw in how 90% of meepo players suck at him, it's no surprise your team easily wins regardless of minor mistakes. You can't blame MM for bad picks. Meanwhile your team's picks has amazing synergy with BH to scout out kills and a Naix-Storm infest bomb. You literally just had to press one button to guarantee a won team fight.
Stop judging someone's overall performance from one or two games. Imagine someone judged you solely from one of the games you fed horribly (everyone does that, from dendi to 500mmr scrub). Would it be a fair appraisal of your overall skill? If it was the first time you were trying some hero in ranked?
Yeah.. it would have nice synergy. But storm afk jungled or fed almost the entire game. He said he was eating.
We not only won the game easily. We did that while playing badly. Also the enemy Mirrana was pudge in my previous game and fed very badly... all the while he claimed he was 70% win dendi-pudge. Dotabuff shows he is 39% win. That player has lost 12 of his last 15 ranked matches. His MMR is just much higher than it should be.
But that is not his fault. It's Valves fault. MMR doing a poor job of accurately reflecting individual skill is one problem, but its a symptomatic problem. The real problem is Valve regularly allows large ranges of MMR to be on a team. That is the basic problem which leads to inaccurate scores in the first place because so many games are autowin or autoloss. Because so many games are lopsided it takes MMR much longer to converge and in fact can prevent it from converging to the correct score. Since win's are nearly always 50%, everyone tends to get stuck (not completely) at whatever MMR they happen to have after the first 10 games on an account. You can move off it eventually, but it takes a long time. It would take maybe 20 games if they did it right. But it takes hundreds of games to correct an error because so many games are certain losses or wins.
So only after hundreds of games does your incorrect MMR edge downward or upward. I'm not saying all MMRs are way off. Most are close enough. But too many are way off... leading to more imbalances. The root problem is Valve does not value match quality. They consider very low match qualities to be OK because the que time is lower that way.
----------------------------------------------------------------------------
Wait are we looking at the same game Zenoth? They have Mirrana, Invoker, Axe, Dark Seer and Meepo. How is that a horrible team? It has really powerful teamfight and strong lanes provide axe does not jungle (and he did not)... yeah the meepo probably can't be played by a lower skill player. But the picks themselves are fine.
Yes, my silencer counter pick really kicks butt against them... but not when my storm pays so little attention that he dies afk to creeps in the lane. He went to jungle because there he was safer afking. Most of this game Storm and LS are jungling while BH, WL, and I beat their whole team... and I didn't even play well. Lifestealer came out every time we had WL ult up and we killed 3 or 4 of them. Then back to farming. The other team was just out of their depth, had no idea what to do.... no leader, no direction. But they were the stack. We were the random solos. I still think the game was just unfair. And I know that Mirrana really wanted badly to beat me after the previous game where I pleaded with him to stop feeding.... which did not work. That was definitely his best effort on Mirrana, Valve just put him in a game with MMRs way to high for him.
-----------------------------------------------------------------------
Here is another way to look at it. All the stats are not on Dotabuff but you can get the idea.
Silencer +66 wins
LS +10 wins
WL +31 wins
vs ------------------------------------- Which player does not belong in this game?
Mirrana -15 wins <----- yes
Invoker +3 wins
Dark Seer +36 wins
Axe +39 wins
See this is Valve's error. Maybe my storm was at -20 wins and had similar MMR to the Mirrana. But neither of them should be in the game then.
-------------------------------------------------------------------------
Too greedy. Meepo and invoker need levels and farm, axe needs farm, seer needs levels. Potm would be the only one functioning as support, but not necessarily a very useful one.
On paper they have strong teamfights, but you need a ridiculous level of coordination and execution to pull it off. In pubs those who can focus down single targets quickly often fare better than those relying on coordinated teamfights.
Of course, you may be right about matchmaking, I'm not denying that, though I've mentioned before that there will still be edge cases even when everyone has identical mmr. You would need alot of data to measure the worth of wait time against match enjoyment, especially when you will have stomps regardless, albeit less commonly so.
Well they were a stack (not the mirrana). They really walked into a lot of WL+ Silencer ult combos because they did try to be together and coordinated... but they didn't have the skill to pull it off. WL and Silencer ults have a long cd and there were many opportunities to punish us while those ults were down. But they waited so long to fight again ults were up.
I agree completely that stomps can happen with equal teams. And probably MMR for the teams was close in total. But it was not close for individual players. I think its unfair to put people into games with teammates of vastly different MMRs. If people put themselves on a stack with a large MMR difference fine...
But if I stack with some of my highest MMR friends...
What Valve does now:
My team MMRs ------- Enemy
3400 ---- 3400
6000 ---- 6000
5700 ---- 5700
5300 ---- 5300
5200 ---- 5200
In this case the random 3400 guy added to match my low MMR is totally screwed and feeds because he has no experience playing a super high level game. So he drags down his whole team. Yes, I'm very slow compared to those 5k to 6k players. But I know warding very well. I stack camps and play a hard support. I mostly hide and then cast my spells in a teamfight quickly before my being slow matters. The random guy has no clue what to do in a game vs truly excellent players.
What Valve should do:
My team MMRs ------- Enemy
3400 ---- 5400
6000 ---- 5500
5700 ---- 5200
5300 ---- 4800
5200 ---- 4700
This would be a fair game.
Is there data that supports that valve matches games the first way over the other? I haven't really played any games where there was an opponent noticeably worse than the others to offset a newer player in my stack, so I cant really comment on this.
Yes, they showed a similar example when they introduced ranked Matching saying it was a "good" match. They also showed another example on the dev forum a year ago. Those matches had only MMR ranges of about 1000. But I have been in tons of games where my friends are way way higher MMR than I am. I'm sure other people do the same, maybe not as much as I do.
If Valve thinks 1000 MMR range is "good". Then they probably think 2k range is acceptable for those longer que times where the search range fills out.
Here are the criteria Valve uses for matches.
1.The teams are balanced. (Each team has a 50% chance to win.)
2.The discrepancy in skill between the most and least skilled player in the match is minimized. This is related to team balance, but not the same thing.
3.The discrepancy between experience (measured by the number of games played) between the least experienced player and the most experienced player is minimized. More on this below.
4.The highest skill Radiant player should be close to the same skill as the highest skill Dire player.
5.Each team contains about the same number of parties. For example, the matchmaker tries to avoid matching a party of 5 against against 5 individual players.
6.Players’ language preferences contains a common language. Lack of a common language among teammates’ language preferences is strongly avoided. Lack of a common language across the whole match is also avoided, but less strongly.
7.Wait times shouldn’t be too long.
This sounds nice but they also say right after this that...
"The matchmaker seldom achieves all of those goals perfectly. For any potential match, the matchmaker assigns a quality score for each of the criteria above and then takes a weighted average. When the overall quality score exceeds a threshold, the match is considered “good enough” and the match is formed."
Then they show a match with 1000 MMR range and call it good. They say "Players’ appraisals of matchmaking quality are highly correlated with their recent win rate." Well of course ppl won't complain about the super easy games they unfairly won.
--------------------------------------------------------------
Despite all this I do think most of the matches are fair. Probably less than 10% of them are really bad. I just think that 10% or 5% being unfair is unacceptable.
If the really unfair games were knocked down to less than 1% of games people would stop noticing them. Considering how frequently I get teams with people who can't or wont use English on US servers despite my "Prefrence" of English only being selected.... their "strongly avoided" must not be that strong.
Wasn't the match they showed the result of a stack? I also remember how they mentioned they balanced the game by the best player, not the worst, due to their data showing that a good player impacts a game much more than a bad player does (presumably he can at least call the shots even if not on a carry), so if a 3k stack has a 5k player there will likely be a 5k player on the other team, but if a 5k stack has a 3k player there wont necessarily be a 3k player on the other team.
As for what MMR range is acceptable, I don't really have much to comment.
Yes, But you Must remember. Party Que and Solo Que are the SAME que. My examples are for a stack also. In fact it is only weeks ago that Valve FINALY added a patch that is supposed to prevent 5 stack vs 5 soloes.
Almost 1 month now I guess, Dec 18th patch "- 5 solo players will never match against a 5 man party in Ranked Matchmaking (and will be extremely unlikely in Unranked)." So it still happens in unranked and has been for 2 years. So many times I tried to tell people that this was happening but they did not believe me... because its so obvious that matching should never do this. People think "valve can't be that dumb".
It is possible to put together a fair group of MMRs, even with the large range on one stack. But Valve tends to not do it. I suspect this is because their algorithm values goal #4 far to highly. Goal #4 is what always puts much weaker players on the team of elite players when they try to solo que. Its just a mathematical consequence of trying to match the average while maintain the high value. You pretty much end up maintaining the low value also.
Goal #4, screwing you over in solo que for two years running.
"4.The highest skill Radiant player should be close to the same skill as the highest skill Dire player."
Is goal 4 necessarily bad?
Presumably from the data it means that if you pit a 3k-6k stack against a 4k-5k stack, the 3k-6k stack will win a lot more frequently than 50% due to the higher impact of the 6k player despite the average mmr being the same. Would it be fair to the losing team then? Remember that this can happen even in 2-stacks with high skill variation, which are quite common in solo queue.
I'm not taking Valve's side here, I'm just trying to see how their actions may be justified based on the data available to them. Of course, it may lead to other problems of its own, which are the problems some of us face on a day-to-day basis. But we never get to see what bigger problems might have been avoided since we never got to see them in the first place.
I think Valve theory crafted themselves into believing that 2 years ago and never tested it.
But what it really does is push the top players into a situation where they know their team will be feeders if they play solo. And any 5k +, even 4k+ player knows they end up carrying a huge weight often times in solo games... Meanwhile lower MMR players are occasionally stuck into high level games to balance some stack and can't help but feed while everyone shouts at them to stop.
As in the example I gave several posts above, it would be much more fair to spread that large MMR deficiency out across all 5 players and keep those 5 players close together in MMR. Then they will understand the same tactics and have the same expectations and skills.
Those players who are superstars at 6k MMR are going to dominate any game no matter what you do. They do even in pro games. Its not smart to try to force the match into a contest totally decided by the top MMR player Dire vs the top MMR player Radiant. This pattern contributes to the way page 1 live games end up.... a bunch of people all trying to solo play their way to a win because that is the experience they had... what they needed to do to win games.
This is why so often people are demanding solo mid in ranked games. Even if they don't understand why, people have that sense subconsciously that if they don't take control of a game... some nub will feed mid and lose it. It's because so often there is a nub on your team courtesy of Goal #4.
Many players with long losing streaks lately...
http://dotabuff.com/players/143434493/matches
http://dotabuff.com/players/76384373/matches
http://dotabuff.com/players/30597066/matches
http://dotabuff.com/players/114091014/matches
And these are only from the 2 matches I played yesterday. I think the removal of solo matchmaking and the split into solo/party mmr has something to do with this.
Zenoth still trying to prove his fail point... interesting, interesting.
"But what it really does is push the top players into a situation where they know their team will be feeders if they play solo. And any 5k +, even 4k+ player knows they end up carrying a huge weight often times in solo games... Meanwhile lower MMR players are occasionally stuck into high level games to balance some stack and can't help but feed while everyone shouts at them to stop."
So true. How many fucking games I had to carry trashes like ones talking here to win - and they are now mad at me cuz I call them bad.
There is no fail point for me to prove because there is no reasoned argument against it. Relentless's best explanation is that Valve pulled the data about better players having a larger impact on the game than noobs out of their ass.
Meanwhile you prefer to bask in the illusion that matchmaking is to blame for giving you noobs on all the games you lost. How could it be that you played any part in those? Blasphemy! Being in 100% of games that you won, all the hard work must have come from you since you were obviously a common factor.
On a more serious note, what you quoted from Relentless is only likely to have any modicum of truth to it if you end up blue or pink 90% of the time when solo queuing, which has been postulated to be the player with longest queue times - a.k.a. the player for which MM is having the hardest time finding a game for.
more of relentless thinking he is the 'best shit tier player'. I hate to break it to you but you are 2k MM for a reason and all your dota theory is great for giving information but all that information is useless if you can not put it into practice. Knowing the exact time to pull is great information but if you mess it up and pull late by mistake all the time you are no different than the clueless player who pulls randomly and pulls late. The result is the same, get off your high horse plz
@Relentless, there is just one big error in your argument. Valve drops you a question after games, rate how much you enjoyed the game. Guess when avg (95%) player press 5. Yep, when they stomp. Most of them will give low points to losses, whether it was stomp or close... so valve found a way to give fast queue time and maximize satisfaction. Not yours ofc, but most players.
http://dotabuff.com/matches/469726067
Omfg, sorry bros I actually was the reason for the bad game.
Not 0-10 idiot mid, or wd was keep telling me that naga is carry, tiny is support. Or that they both didnt know how to roam and were farming creeps on different lanes.
Ggwp idiots.
^midas into shadow blade
you are right your horrible decisions do not effect the game at all......
I love it! "My team feeds in early game really badly? They must be a bunch if noobs, time to build midas to carry them!" Then flame team at the 20 min lose. XD
6k damage on a tiny despite going for a positioning item, when one combo easily clocks 1k+. Impressive. Choosing to go midas against a lineup particularly high in early nuke damage, arcanes instead of treads yet not ganking to relieve the pressure early. ...
No one is saying we are all 100% responsible for our losses. But we all contribute to it in one way or another, be it through mis-plays or simply the attitude we demonstrate to our teammates. To think one is faultless is simply narccistic and responsible for many of the plateau-ing of skill levels.
Sampson,
''The result is the same, get off your high horse plz'',i can tell you the same thing,i am 2,8k MMR player,fighting now for 3K,fighting hard,you can see my list of matches,i read you on forums a lots,always ''high horse'' probably becuse of high MMR after shitty calibration way of starting MMR,and choice of heroes,wow,Slark,BH,LC,Furion,did you chase kills when other team tearing appart your rax,or just run into invis ???
Relentless played 3x more matches then you and you are smart ass ???
Volvo made a lots mistakes,a lots people stuck on cca 3K rating,we play with people who have 45% of wins....They don't have a clue what is team game not to mention others stuff.....
I got winning strikes,loosing strikes,it's depends on other teammates,on low ranking matches diference in skill level is huge.
Point is,people with high MMR think they are really good players,i respect,when they have 2K+ matches and they are level 100+,but when they are level -50,or have new account and chasing high MMR with that account that is really poor and stupid....
You can add me Sampson,chose some you favorite hero,and come to solo line i will show you what 2.8MMR can do.....
Sry for my english if i made some mistakes.
I came in from dota 1 so my dota 2 matches do not reflect my total dota experience (not saying I am great but just saying). My KDA is avg at best so saying I am a kills chaser is flat out retarded.(my last match on BH I got a necrobook to push towers with). All I hear is blah blah blah my teammates are holding me back, you realize there are shitty players on your opponents side too? And lastly I will deal with your 1v1 nonsense, asking for 1v1's shows how retarded you are because you are asking for a 1v1 in a game build for team play.
Sampson, so u idiot really think that agha/any item would have saved the game? Why are u so fucking retarded, I dont really get it.
And next game follows.
http://dotabuff.com/matches/469794661
3 russian idiots (top and mid), wisp that gets absoultely no wards and admits he is first time wisp, axe with meka (when we have wisp), qop building bkb 1st item.
Tell me its a joke. I know I'm good, but not that so I get trashes every game.
PS I just cannot understand how to argue with trashes like Sampson. Giving them all arguments, saying what idiots did wrong, etc. This retarded tells me that my shadow blade lost the game. As if I had agha 1st item that would be pure win game 100%. Really braindeads, learn to think before talking.
valve has control over what hero people pick? If you do not like allpick do not play allpick.........
I'm pretty sure everyone is not blaming the Shadow Blade but the Midas.
And it still looks like the russian stack got more work done then you and your friend. Midas against all their strong early fighters and by far the lowest hero damage among the cores despite being the most farmed. Why not say the Ember Spirit was so bad, died 11 times and got both MoM and Deso despite it not stacking?
I am asking people in each game what their MMR is, most of them are somewhat reluctant to answer, but from what I understood - with a 4.1k mmr I usually get matched in a same team with 3.3k-4k. It is not too bad, but could be better.
If it was a compulsory thing, rather than giving a choice between ranked/unraked MM, we would definitely have better matches, but, well, I think at least some improvement in the are is good.
MMR is bad
Nah,
Just saying,MMR is good idea,and one day it will be really good system,but right now is far far away from even decent system,after every match i check all players,stats,win ratio etc etc,and when i see.....Sometimes i wish to play on old way,SD-RD without MMR,without spending nerves...
To explain how things works on low levels of MMR,when you decide to pick carry after you others pick another 3 and one support who play ofc semi carry way,so,the only way to have 80-90% of wins and climb higher i's to play offensive supp/pusher,farm your idiot noob carry so he can right click right click win some kills in mid/late game and say in the and ''comenT me'' Russian ofc....Wait to all pick their heroes,make counter for enemy team,try to make some synergy for yours.....
That is my advice for strugling/stucked people like me,for now it's working,when i get on 3,5K-4K MMR maybe the situation will be beter,right now is hell,also,don't play from 8.00 till 17.00 on EU servers,kids are just terrible.
Gg.
And the next game is a free win after free losses.
http://dotabuff.com/matches/469857124
Really no way to play this matchmkaing, it just gives you amount of wins/losses as it wants.
Just like there are people with long losing streaks there are people with winning streaks.. It doesn't mean anything..
http://dotabuff.com/players/24191321/matches
@Valedithia If you misunderstood me on that point, probably others did also. Let me try again.
Valve says ""Players’ appraisals of matchmaking quality are highly correlated with their recent win rate." And I'm sure you are right that this is in reference to the post game questions on match satisfaction. The fact is people who lost badly are likely to give the game 1 star and people who won badly will give the game 5 stats... I won't, but most people would. This is a poor metric for judging match quality. The fact that Valve is using it at all proves they don't understand reporting bias. But like their other metrics they don't concern themselves with the metrics being accurate. Instead they say its not important as long as everyone is close to 50% win.
Despite all this complaining about matches being unfair... they still are far more fair than in dota 1. I played about 5,000 games of dota 1 and these games were maybe 75% one sided stomps.... some against me, most in my favor because I knew how to see before the game which team would win in various ways. In dota 1 you could switch teams before the game started so all the players who knew how to check stats switched to one team or quit before the game started. The result was that one team had all nubs who didn't know how to see which team was the winning team pre-game. The higher Tiers of ThroneIT and Dotacash random balance shuffled games... but ordinary pub games were all stacked for one team to win.
That stacking behavior was partially countered by smurfs who made new accounts that looked like nubs, but were really some of the best players who would then stomp the moderately skilled stack that set up against them. In dotacash there were some people who had over 100 smurf accounts. I had 13 myself. The stacking was reliable enough I could maintain 75-80% winrate on my main account which had about 2,700 games when I switched to dota 2. Now I can barely get over 50% ==> much more fair.
Valve's current system produces probably 90% fair matches. But some really are still bad. I think they could fix this, but over the past year I have become more and more disappointed that Valve seems uninterested in correcting or even detecting their problems. I had hoped that at least in Ranked games they would stop hiding the stats. The fact that they continue to hide MMRs even for Ranked where there should be no expectation of privacy to protect nubs... its just that they are embarrassed by the matches they are creating. Valve says they are still changing things all the time trying to get it right... But they have been doing this for 2 years and still are failing to improve. It's just getting frustrating.
ｊｕｓｓｉ, when you gonna learn what forced win/loss is, kiddo?
lie to me agen fagt when are you going to stop being a delusional prick, kiddo
^ Pfff. Why to talk to people that have wooden brains. Ignoring the fact given above only by deluding themselves that I'm bad (though I could pwn u with my eyes closed), believing in shit like fair mm only cuz it gives u free wins, etc.
Thinking is hard nowadays.
PS 90% fair? I'd say 10% for good players, 90% for trashes like mimiming in this thread about how they are getting carried to wins. But Valve actually did nice job, none of these guys understand they got decieved into thinking that mm is good.
no idea how good you are but if you can't climb out of <4k solo then you don't stand a chance against most people here on dotabuff
I wanted to damn near cry last night. 4 stack. Single guy got captain. Didn't talk the entire time whilst picking, nor did he during the game. Gave no heroes the other players wanted (the only one we got was me on Lion). Such a pain in the ass.
@ Zenoth. I really do not believe that the "best player wins dota games" concept applies to very many games. I hope I can illustrate that this is true to your and others satisfaction.
There are situations... where one player is vastly superior to all enemies and wins the game almost totally alone. But these are exceptionally rare games. It is extremely difficult, especially in higher level play, to totally dominate a game by yourself. The best player may get an edge in his lane, may be able to turn that into something more with a timely rune pick up... but that cannot compare to the game impact of the worst player consistently feeding.
Pub dota games are much more regularly decided by the worst player... the one who fails consistently to match up to the level of play in the game. Who is out of position many times, who thinks they can make a play when they can't, who tries some stupid build and becomes useless, who farms for a long time missing lasthits, who afk jungles as the rax dies, who fountain dives and throws away the only chance to win... The pub dota game is typically won by the team who fed on the worst feeder, not the the team who's best player made some amazing play.
Goal #4 is based on the idea that the game is controlled by the top MMR players if such player are outliers. But it is not. It is controlled by the lower MMR players in the game if they are outliers. IF everyone is close, then you are much more likely to get back and forth games that are hard for either team to win. When there are a couple players who are much lower MMR... one of them will begin to feed.
They will begin to feed not because of a simple mistake, but because they really can't play against the superior players they face. And so the first feed will become many feeds because they cannot recover. When a player of similar skill to his lane opponents loses the lane, he has lost only a small advantage. He still might outplay them later. When a player of much less skill loses a lane he is just going to keep feeding. He could not win when it was even... with a disadvantage he is totally helpless.
The low MMR player controls the game by feeding first... and then continuing to feed. Maybe the other team also has a similarly low MMR player, but if he does not feed first and makes it to his core items he can still contribute. The low MMR player that starts badly does not recover. He is outskilled AND behind on xp and gold both. For the rest of the game its basically 4 v 5 and the game becomes a stomp as other players are dragged down by the one who began to feed first.
The Low MMR player may cause the loss in other ways that are less obvious as well. He may decide to be jungle axe and do nothing to help the team for 20 min at which point he has vanguard and tranquil boots and his team has lost all 3 lanes. Maybe Someone could then could out of the jungle with more farm (maybe a blink!!!) and start to turn the game around. But the low MMR player cannot. He can only begin to contribute too late, and badly and begin showing his low skill missing calls and not knowing when to attack or retreat, hiding behind his team watching them die, watching towers fall. At the end he has KDA of 2.5 and claims he did not feed. But he still caused the loss because he could not play at the same level as the other players in the game. He still was the proverbial "anchor". The low MMR player decides the game (when MMRs are not close together)
http://dotabuff.com/matches/470115379
Russian am 1-10, what to expect lads?
Let's look at this game just posted http://dotabuff.com/matches/470115379 The AM was not actually 1-10, but he clearly did fail far more than all other players causing the loss.
Here is a metric that normalizes the support and carry contributions so you can see how well they played independent of roles. This shows how much each hero was able to contribute to fights given the farm they required to do it.
ES 10k*KDA/networth = 1.15
AM 10k*KDA/networth = 0.38
DP 10k*KDA/networth = 2.65
Rub 10k*KDA/networth = 1.44
LC 10k*KDA/networth = 2.70
Brew 10k*KDA/networth = 1.02
Pudge 10k*KDA/networth = 1.25
Kotl 10k*KDA/networth = 4.42
PL 10k*KDA/networth = 1.79
VS 10k*KDA/networth = 1.94
We can see that the KotL on the winning team made a very large contribution compared to how much farm he required. While the Antimage on the losing team required tons of farm and basically used it to die before doing much of anything. Was the Anitmage a much lower MMR than the other players? Unfortunately we don't know because Valve hides it. It could be that this AM tried his very best but had no hope of performing well because he was placed in a game far above his MMR.
When a large part of the team networth goes to the lower MMR player the team can't function. Does this mean the highest MMR player has to be the carry? It would not be a problem if all the MMR scores were close and everyone was only slightly better or worse at carry. But if one player is far worse at carry... and decides to play carry. Well now you are all screwed.
Its much better when the lowest MMR player has much less networth. You don't want the less skilled player to be required to make the big plays. So if I stack with near-pro level people I pick supports and we win 70 to 90% of games. I'm going to pick a low skill cap, reliable support like veno, lich, or dazzle or CM, a hero who is just as good casting within 0.25 seconds of the correct time as 0.1 seconds of the correct time... not something hard to execute at a high level like chen or enchantress. But some random person won't do that. They won't even know they are the low MMR player until the game is going for a while so they might take mid or carry and be unable to fulfill their role.
OH, MY, FUCKING, GOD, SRSLY. KILL ME NOW, SLOWLY
Would cause 10 times less pain than this fucking game
I think the problem is more serious that mmr doesn't reflect player skill. As kawaiisocks said if you ask around in a match or inspect players and they don't have it hidden the range isn't usually that big. I'm like 3200 solo rating and most people I can tell their rating are 2900-3400. Not a big swing either way. However many of them are substantially worse than me from a simple understanding the game perspective. For example they random last picks, get a terrible pick for our team, refuse to repick. Supports fight carries for last hits. Supports don't ward. You don't even need to counter ward they just don't ward at all. I understand not wanting to play support, I get it, I stopped playing nearly as much too but I always pick my carry with 1-3rd pick. If 2 or 3 carries get picked I won't be a dick and go a 4th or 5th one but players in this range do so ALL THE TIME. I had a match first pick was random bounty hunter, I picked viper and said I'll go mid or lane, next pick was dp for mid ok all's fine. Last two are mirana and anti mage cus they wanted to lane that combo together for some stupid reason. Needless to say AM had no farm and we got crushed cus we had 1 stun, no supports and no team fight outside of dp ult.
I just don't get why people who make dumbass decisions like that are worthy of what is supposedly a medium high rating in the 3000s. But somehow these losers have the same rating. Not saying I always make the best picks or lanes either but I generally win captains drafts when I draft.
The term carry is misleading to alot of players while mid/late game is mostly the carries job, it is the job of the supports to carry the carries to that point. While the term 'creating space' is often joked about this is one of the key jobs of the supports which does not get logged into the stats screen at the end of the game.
Generally its because they are more skillful at clicking. Being better at clicking in dota can win a lot of games. Game decisions and knowledge can win games too. And people who win games in either way tend not to understand the other's weakness.
The people who win on decision making think "How can you be so dumb as to do that and throw!!!" While the people who win on skill think "How can you be so bad!!!" There are different ways to get to the same MMR.
@There is no try
There is nothing MM can do about people picking heroes they should not pick. If they do this commonly their MM will go down which will reflect their game success. If I played venomancer over and over my MM would go down not because I became a worse player but because I played something I sucked with. Maybe I have it easy because nobody really fights for the offlane......
@Relentless
I don't have the data to contest what you say, and any evidence I have is anecdotal in nature. You may be right, I do not know. Valve claims they have the data, and it is also in their interest to at least use it, so my assumption is that they are acting on it.
@try
Winning has alot of factors. Mechanical skill is easily one of the biggest factor - how well you lane, how good you are at last hitting, how well you execute combos (e.g. Timbersaw retract chakram while whirling on a chain) - good mechanical skill alone can easily get you to 4k+ mmr. However, game sense, picking sense, awareness, leadership, personality - all of these are significant factors which affect your ability at the game too. MMR merely demonstrates a rough aggregate of everything I just mentioned. Often, if someone is picking like an idiot, you can probably assume that he is relatively more capable in other categories.
I assumed basically the same things a year ago. But over time I've just seen too many bits of evidence that indicate whoever is doing the matching tweaks just is not up to the task of figuring it out. The way they release so many little patches with more and more bugs... like "oops we forgot that" and "oops missed that" and "oops didn't mean to do that" every time they do a patch. It's just sloppy. They don't even put those errors and corrections in the changelog anymore they just slip them in silently, for better PR I guess? I just ran out of "benefit of the doubt". And when they gave me 7 +0 points wins in a row near the beginning of ranked matching it just became very clear that they did not have a handle on putting the matches together.
Now Valve has announced that "- Ranked matchmaking rating changes are no longer dependent on match length." after they just went on and on about this fancy metric (time integral of the gold difference, measured since the last point in the game where the difference was zero) they had was so great. I explained on dotabuff that it meant you would get the most points for a longer match where you were ahead the whole game but did not end the game. Now 2 months later they realize its bad and take it out. I mean, seriously I explained the impact of the metric on gameplay without even seeing the code when it was first announced and it takes them two months to realize that this means longer games are more points?
As someone who plays support the majority of the time, it REALLY bugs me when people are calling for wards & in the same breath, get upset that you take last hits. I generally try not to, but there are items I need to do my job effectively as a support, too.
Still trying to find that balance.