I think this factor has so low influence that it's negligible. For a researcher like me I'd say KDA has the most influence amongst the other factors mentioned.
Well but it's unknown and hidden, who knows..
I'm curious what would these factors be for an omni player, except heal. Repels count? Ulties count? Meaningless repels count as plus or minus? At KDA/LHD usually omni sucks, so he would have a disadvantage usually.
Naah it would be for sure based on average numbers of the hero but even then... let's be honest, it's impossible to make it perfect.
Win/Loss is probably a big component and when you win a game you should have lost because the enemy team was stronger concerning rating you get probably a hole lot of points. Everything else can be misleading and shouldn't play a huge role. So when your hole team is feeding and then you start dying too because they are freaking fed you will mostly not drop a lot of points because the system probably expected this team to lose anyway.
I could imagine that they take the average stats of each hero and compare your's to them.
For example lets say that CM has an average KDA ration (counting assists and kills together divided by the deaths) of about 2.3. Then they check what your's is and add points or take points. They could do the same for GPM and XPM but then they would need to compare those values the to matches with the same result (long story short, you cant compare the GPM of a winning AntiMage to the GPM of a losing AM, so you compare lost game average to lost game).
Thats how I imagine it works
would kind of suck for hybrid heroes that are played as hard carries OR supports/offlaners for example naga siren or mirana.
Ranked play in itself encourages selfish play and basing your mmr entirely on match w/l ratio wouldn't be accurate. Listen to @opinionomnomnom
Exactly what opinionomnomnom said.
And also, I guess, that if you focus on trying to abuse system and getting good stats, you hurt your team's chance to win and thus, actually you lose more than you gain through abusing.
so lets say you play support and you have another support on your team, you would always want that other guy to buy wards, smokes etc. so you can focus on getting a lot of utility items like ghost and force staff to help you survive and thus improve your kda.
then, if the other support agrees to buy all wards and you win the game, the support who didnt buy any wards and went for utility items instead will most likely get the highest increase of mmr cause he died less.
how is this system even remotely fair?
Well, there are certain supports that need the items (tide, visage...) and others who literally need brown boots and a wand to be effective (Bane, CM....). Tide and visage are generally gonna have better KDs due to them building items/tide's insane amount of assists while CM and bane just die of they get focussed. Should even itself out with the averages that I tried to explain earlier.
Still, as long as I gain MMR I dont rly care if its 5-10 more or less :)
They probably don't focus much of KDA. KDA is mainly something player focus on to bring their performance.
@sano : I don't think the system does that, otherwise the system sucks badly since anyone can abuse that.
No, it's not easy to abuse that. I'm no programmer but it's obviously not hard to develop a system that knows that when someone gets 20 dusts and 20 sentries in a row it's not a normal situation...
When there are several invis heroes you buy 10 sentries in a row? lol
One thing is always having sentries up, the other one is buying like a maniac to inflate your MMR, but w/e, the system probably doesn't count this
If you're about to die and have 800 gold on you, might be a good idea to quick buy 4 sets of sentries.
i can tell you it seems they compare your kda (or some similar k/d/a number) to the other players in your team according to role (supposedly) and avg for that hero
Also that "individual performance" I guess is either playing your hero better than YOUR average stats or playing it better than the GLOBAL stats of the hero.
Not sure how game should measure that you play Omni better then Riki. I think it's about GLOBAL stuff
^No I mean you have average 3 KDA 250 GPM and 400 XPM with a hero. If in your next game you get 3.5 KDA 300 GPM and 500 XPM then your MMR "can" increase. By YOUR I mean that.
Yeah, but there are also heroes you did not play at all and heroes you have 1 game with. If I play my first game with Invoker now, you think my rating should get higher if my team carries me?
Bitte melde dich an um Kommentare zu posten.
http://blog.dota2.com/2013/12/matchmaking/ states the following:
Win/loss is the primary criteria used to update MMR, but individual performance also plays a role, especially when our uncertainty about your MMR is high. It is possible for an individual MMR to increase after a loss or decrease after a win
does anyone have any clue whatsoever what factors are included in this 'individual performance'? if its stuff like kda, gpm, last hits and xpm like people suggested before, the system is highly punishing for people who only play support and dont care about these factors but only about winning (even if it means they buy 20 sets of wards during the game and go 1-15).
tldr; i dont think this focus on 'individual performance' is the right way to go as there are not enough viable ways to measure it atm. we would be better off with a mmr that was based on win-loss only. as things are currently selfish play is promoted way too much.