General DiscussionSome conclusions about Matchmaking and Brackets

i thoguht soo too and it would make sense especially if u consider that sing sing was highest rated player at one point while he always farms and go for kills even with randomed supports

also few players on devs dota 2 forum said u can reach very high bracket without winning a single game only thing u had to do is reaching good kda

i believe it was ranaki that tried that 0 win thing with new account

I play like 3x my allies every game and i still play in high only. Nor that i want to play very high (me solo against stacks, no thanks), but just bringing up that this brackets sometimes are very random imo.

i am sure mm has same shit alghoritm as old report system L:OL @ We are looking at patterns of behavior over time only
2 reports in 24h put in nice words
pretending to be competent is what valve does

so what if u troll a bit ig when u won and suicide sometimes it makes u unskilled for valve ?
i doubt they are that retarded ;>

yeh clearly report/mute system didnt show that

and other things they designed that didnt work at all like hero performance bars ( both editions ) hall o fame and guilds

and even gayben said they have shit rating system

im not talkin to you ? ;o

yes you are

what most people don't understand is that in some games you are actually expected to do really well especially when you play a carry

for mm having a win rate significantly higher than 50% means that you are underrated
mmr can be seen as an expectation of your performance in the actual game and when you feel like your game is rather low rated (by your own standards) and you're being one of the best players you should accept that it's expected for you to crush the enemy players as they'll be worse on average as well

I've been working my way up in the brackets.

When I finally get to the very high:
http://dotabuff.com/matches/308329784

Guy playing his first lion feeding ultra hard, a poor mid by me and a clueless alch.
In the next game I was back all the way to the NORMAL bracket.

Wtf? Is the team performance taken into consideration?

Maybe it considers the past few games?
I don't know, it seems to be a very random system for me.

Here are my last 4 games:
http://dotabuff.com/matches/303738122 VH game with one leaver in my team. Lost
http://dotabuff.com/matches/303764084 Another VH game but this time with 2 leavers in my team. Lost
http://dotabuff.com/matches/303779666 Third VH game in a row. This time against a team who didn't know how to deal with a Huskar.
http://dotabuff.com/matches/303805383 This game followed it. Compared to the first 2, I think I did a decent job in the third game, but I was still put back into High bracket, which I wasn't put into after 3 loses in a row.

So...unless I was demoted from VH for picking Huskar, I could say that it either considers not your last game but the last few, it is a very random system or the High bracket and the VH bracket are no different at all.

So when I play naga support my MMR decreases because most of the people play carry naga and my KDA GPM XPM will be lower than the average?
BULLSHIT

When you are switching between high and very high, it probably means you're in the high end of high and the very low end of very high.

let me tell you something. I got a losing streak few days ago and went down to the normal bracket.
http://dotabuff.com/matches/299956404
this game started TERRIBLY, godlike viper godlike QOP, The score was like 2-22 on minute 20 with my teammates feeding ( I had 0 deaths at that point) . But I kept on farming calmly, got 2 kills on my lane and farmed desolator and blablabla. so the plot is, I carried the team from the bottom of the ocean. Their win looked safe but I still carried them hard.
So after that game I was put in the very high bracket from normal. one game shouldnt affect my mmr drastically because I had 100+ games at that point. But it did.
So I mean that there are a lot more things that MM uses to calculate your mmr.

another example
http://dotabuff.com/matches/308597828
safe-lane medusa. we won early game and started pushing. but after some point I left them and went to farm jungle. so they were playing 4vs5 from that point. So even I didn't die a single time and had respectful KDA. I still got down to high bracket game after this one. maybe because my K+A was lower than any of my teammates. who knows.

Was there a leaver in that game? As soon as there's a leaver, any game will be considered normal.

Like I mentioned earlier, you're probably in the high end of high and low end of very high. This is why you're bouncing back and forth because its virtually the same.

if you solo queue you might end up being used as a filler to balance 4 and opposing 5 man stacks or two 2 man stacks
especially when you queue at odd times
if you are in the middle of high or something i could imagine that you could be used to balance both normal and very high games

your mmr actually changes rather quickly since you have a very small amount of games and kda is an important factor in that matter

I wonder if they consider the number of wards you buy if you are a support, that would be interesting. Or the Apm, usually pro players have a higher apm, right?

The highest apms I have seen are often on very bad players who click a ton of times to move and attack for no reason. When you watch the POV of pro players they use very few clicks because they only do what is needed. Their apm would actually be lower than average....and that is why dota does not track apm. It is not a game where there is often more to do than it is possible to click.

I think OP is right that KDA does have an impact on MMR. I don't believe that it should but there are just to many examples indicating that they do use it. I suspect that it is used to change the uncertainty variable that determines how much change in MMR you get for each game.

As for GPM...what is needed to test that is a smurf account of Alchemist who farms like crazy game 1 and instead of winning right away just goes for 1000+ gpm. See if he jumps to Very High like a huge KDA score would. To be an independent test you have to have super high gpm and low KDA. Just PVE it and then get some towers.

KDA is probably the only thing affecting MMR. A friend of mine made smurf, went 41/0 with pudge the first game and was in very high ever since. His winrate is even lower than mine at the point of 60 wins and I haven't seen a single game in high.

Sometimes he wins a game with his smurf and drops down to high. Then he loses and gets back into very high again. The MMR in Dota is very, VERY weird.