Why matchmaking sometimes seems broken.

[Deepbluediver]

Other people have crunched the math far better than I can, but I'd like to try and give some perspective from the human-personality side of things.

First, full disclosure: I'm relatively new to the game. That means I'm still learning all this and figuring it out. But I've had the opportunity to play in both PuBs, as well as in groups with far more experienced players, and I think it's given me a decent experience with gameplay involving different groups.


One of the biggest things that I've come to realize, is that the ease of the game for any one person can vary highly depending on whether your team is winning or losing. In games where my team is winning, even by only a small amount, my mistakes get papered over and the effect of my good actions is magnified. In games where my side isn't in control of the battle, every error makes the next one more likely, and the game becomes harder and harder to turn around. Essentially, the same team can look better or worse depending on factors largely outside their control.

This is why, I think, some people claim that the matchmaking system fails when it sticks them with certain kinds of people, mainly those that have less experience. It causes players to think to themselves that, since THEIR skill level hasn't changed much from game to game, it must be everyone else's fault. Yet I know this isn't true, because I've seen it myself as the effectiveness of my gameplay changes, even over the course of only a single match that swings back and forth.

There are lots of threads about people who go on losing streaks, but where are all the threads by the people on the other team, laughing about how they keep face-stomping all the noobs matchmaking pairs them against, or complaining that the game isn't as much of a challenge?

Because of the variance in the combinations of player-skill levels, hero-interactions, and player-skill with any given hero, IMO matchmaking isn't broken, so much that it's very VERY complicated, and exceedingly difficult to perfectly align every time.


If someone is interested in doing real statistical analysis, it might be worthwhile to track every single game a player plays, win or lose, and the kinds of teammates you end up with, as judged by hard data (games/hours played, wins, etc) not by how noobish they look.

[SymbioteZ]

;)

[Taweret]

its normal practic that loser blame others on his team because he isnt self any better but cant understand it and creating thousand of useless threads on forum how bad everyone are but him. people even flame dendi on pubs that hes noob etc. just ignore them.

[pacifist]

Or because the team is iddling to much

[Lictas/ YouAreANoOB]

I Dont blame my teammates , i blame the mmr. How is it fair that i'm matched with people with 200-700games on my team , and the opposing team? One of the players on the other team will have simliar games to me ( 2k). Its pretty much a case of 'who can snowball first and rape the noobs'.

Theres 2 major issues with the matchmaking .
Firstly , Too many newish players are in a higher skill bracket then they should be.
Secondly the balanceing is bad , just because you have 1 simliar skilled player in each team , doesnt make it balanced .Its a race to snowball , If i mid with a low win game team/enemy , and snowball we win , If i go support/low impact hero , we more then likely lose and the other teams 'higher skilled' player will snowball.

[Taweret]

lol then why i saw players with 500 wins "snowballing" 2k ones? this logic.. there is no difference beteewen 500-2k, most of players with 500 are dota1 veterans. so they really are 10k wins (like myself), you cant balance this by any matchmaking.

[Deepbluediver]

YOUAREANOOB eredeti hozzászólása:

I Dont blame my teammates , i blame the mmr. How is it fair that i'm matched with people with 200-700games on my team , and the opposing team? One of the players on the other team will have simliar games to me ( 2k).

As the previous poster alluded to, skill is not directly correlated to games played. In fact, I suspect that the longer you play, the less each game adds to your skill-level (you improve more rapidly early on), and by the time you've played several hundred games, the system probably has a decent idea of how you usually play.

But, like I said, it matches you up based on the assumption that you will do as much to affect the outcome of a game as you possibly can. If it operated on the opposite metric, games would still be imbalanced, just in a different direction.

Also, the game ranks players based on overall skill, which could be quite different from hero to hero. A system for more balanced play might ask people to pick a role and hero before it organized, but then you would might still have problems with enforcing the roles people chose, as well as eliminating the entire concept of counter-picking (as well as needing to objectively evaluate hero match-ups so you don't see tons of players complaining about the game always assigning them their hard-counter.

Like I said, it's almost infinitely complex.


Next time you play a few random PuB matches, look up all the players you play with and against, win or lose. Maybe get a friend to try and figure out which ones he thought you won or lost just based on who your teammates where, and see how it turns out.

[Taweret]

also everyone have good and bad games, you cant deny that, skill is not stable thing, it vary and depends for example how much you playing one hero (dota2 have above 100 heroes 110? and most of them have differently and unique game style) you cant be good as someone who is playing his 100 game with only one hero while you use him once per week. everyone makes mistakes, cant blame just mm because isnt perfect, ask yourself, you are perfect?

ps i found rarely that ONE player dominate whole game, in most cases is team job, because its team game.

[Lictas/ YouAreANoOB]

Oblivion comes... eredeti hozzászólása:

lol then why i saw players with 500 wins "snowballing" 2k ones? this logic.. there is no difference beteewen 500-2k, most of players with 500 are dota1 veterans. so they really are 10k wins (like myself), you cant balance this by any matchmaking.

I hope you are kidding .
Yeah , people with 500 games that only snowball with pub basher heroes and cant play all heroes, excatly the same skill level as a 2k games player.
Everygame i check all my team mates stats/dotabuff. I get matched with people with LESS games and LOWER winrate . At 500 games you know everyhero /build/counter/jukes? Very doubtful.
The mm is just broken , dont hue hue hue 300 wins player iz good yo. Most arent, or can play 1 hero well.

[Taweret]

i have played dota for 8 years, and i dont know every hero, i have my favourite heroes which i master. and i debout if professinal dota2 players can say otherwise.

[Lictas/ YouAreANoOB]

Oblivion comes... eredeti hozzászólása:

also everyone have good and bad games, you cant deny that, skill is not stable thing, it vary and depends for example how much you playing one hero (dota2 have above 100 heroes 110? and most of them have differently and unique game style) you cant be good as someone who is playing his 100 game with only one hero while you use him once per week. everyone makes mistakes, cant blame just mm because isnt perfect, ask yourself, you are perfect?

I know everyhero build , every counter ect due to the amount of games iv'e had , its unfair to be matched with people that DONT know this . Its unfair on me AND them.. An easy fix would be just to limit the difference in total games. But naaa them 300 game players know everything , can micro visage perfectly , and im just a ♥♥♥♥♥♥.

Ofc people have bad and good games , but not knowing how to play gyro , or visage , Doom with vanguard radiance ags refresher as the hardcarry , its just gg at start

[Taweret]

maybe they never played dota1? its not that 300win player can master dota, or even be half good as pros, i bet there are comp players without any micro skills. most of players choosed dota above warcraft 3 because of controling just one hero...

[Lictas/ YouAreANoOB]

Oblivion comes... eredeti hozzászólása:

maybe they never played dota1? its not that 300win player can master dota, or even be half good as pros, i bet there are comp players without any micro skills. most of players choosed dota above warcraft 3 because of controling just one hero...

I missread what you said - A 300 win player just is nowhere near the same skill bracket as a 2k game player - So why match them with 2k game players.

I can semi understand the ' high ell smurf' stuff , but i check there dotabuffs , they usually have 45-49% win rate.

As i said - i dont blame the players - i blame the mm

[Deepbluediver]

@ everyone
Let's try to keep this discussion civil, please. There is no reason to get personal with the insults.

YOUAREANOOB eredeti hozzászólása:

I know everyhero build , every counter etc due to the amount of games iv'e had , its unfair to be matched with people that DON'T know this . Its unfair on me AND them.. An easy fix would be just to limit the difference in total games.

According to Dotabuff, you've got a little less than 1,700 matches played. What number of matches would you say is the minimum that someone needs to be able to play on your level? 1600? 1500? 1200? I'm curious.

Also, remember that the matching system tries to find the best fit in a limited amount of time. How long do you wait for these matches? How long WOULD you be willing to wait? If you had 20 minutes in between every game because that's how long it took to assemble 9 other people with 1600-1800 matches (no more and no less) and similar win-rates, would you still be willing to play the game?


Statistically, its impossible for everyone to always play with people who are better or more experienced than you; at some point you will need to be the better/best person on your team, and support the others to some degree.

[Taweret]

dota buff dont track all account names the player had in his dota2 career anyway, only recent one, probably there arent many "2k games" players around, that why you are matched with lower ones with similar win rate.

[Lictas/ YouAreANoOB]

Oblivion comes... eredeti hozzászólása:

dota buff dont track all account names the player had in his dota2 career anyway, only recent one, probably there arent many "2k games" players around, that why you are matched with lower ones with similar win rate.

The mm is just bad , theres enough players , it should range total games so i'm/others are matched with people with between x-y games ie 1.5k-2.5k games , or something else realtive to your games.

Instead i assume it gos on an ell bracket ( Between x-y 'skill level')

Which then gos back to my first issue. Low game accounts being in the wrong skill bracket/ell rated higher then it should.
I don't know why people deffend this bad mm , and valve doesnt adress it correctly.

Its a complicated matter that will never be perfect but it could be improved greatly