1. Hello and welcome to the brand new home for PlayDota!
    Please read through our Welcome thread to see what's new!
    Dismiss Notice

Rating works as intended

Discussion in 'Dota Chat' started by kamukag3e, May 18, 2018 at 8:43 AM.

  1. kamukag3e

    kamukag3e Member

    Jun 11, 2009
    I want to talk about on a topic which is no longer actively discussed here "matchmaking doesn't work, rating is not accurate" etc.

    While me and a lot of people knew it work as intended, a lot of people were arguing against it not understanding it and not actually taking the experiments done by good players on it.

    First - let me tell you how Rating is supposed to work.

    After a player is assigned a position in the ladder, he should have a 50% win-rate against his peers (or people with the same points) - that's all of it.
    You are not supposed to know what all heroes do, be mechanically competent at specific level or w/e, all you need to do is if matched with your peers in 100 games, to get around 50 wins.

    A lot of people were stating that their peers are bad at the game, not seeing their own shortcomings and when they are under performing or use excuses (in front of themselves for those games), while not even understanding that the other players are in the exactly same position.

    Now on my personal experience with the different ratings and players.
    My steam account is linked and easily accessible for registered members which can after that search for the Dotabuff stats of it.
    Summary - 1,700 games, ranked been 1,017 of them with 53.6 win-rate in the European servers.
    My solo rating is around 5,500 and party 4,800.

    I had been given access to another low-rated account from friend of mine, because my party rating was to high to play with them and I actually did quite a big number of games on it (can give information only on those which ask for it in private massage for safety reasons).

    But to summarize it's stats. It was with 665 games and calibrated with 1,500 solo and 2,000 party rating.

    After I took ownership to that moment it has 1,363 games (or 698 are mine) and I while playing casually and not with the intention to win no matter what - also playing all roles - while being terrible with them (and if done on my the high-rated account will lose quite a lot of games due to under performing severely) when my team need them filled I still manage to get 60% win-rate and climb the ladder and after that number of games I am at 3,600 solo and 4,300 party.

    While I do not have the specific number, I estimate my solo win-rate to be around 70-80% while party 60%. The reason for that is because on solo I usually pick heroes which can carry the game, while on party I play with bunch of friends and just pick w/e seem fun, one more reason to have such a low win-rate on party is due to inflating my friends' party rating as well, once we started playing they all get around 500-1,000 rating each after the first 100-200 games and from them being severely outplayed by the enemy team due to not accustomed to the difference, slowly catching up and getting better at the game.

    What I wish to say is that despite me not being as good as multiple players which have done such an experiment and grabbing the core roles manage to climb extremely fast, just because my overall performance was still much better then my rating shows I was surely climbing the ladder little by little.

    So to all of you, who are eager to get higher rating - the system is working, don't rush to your aim, take it slow and enjoy the ride.
    If you wish to get fast results - just work on yourself and stop watching/blaming teammates but instead try to outperform them every single game, despite the outcome.
    Skulls_n_Souls likes this.
  2. ManOnTheCan

    ManOnTheCan Member

    Aug 18, 2013
    I'm at least 7k MMR, but my MMR is 4.5k, clearly Valve system is broken

    Daed game, I stopped playing years ago, gg noobs
    slipkwar and kamukag3e like this.
  3. GoLD-ReaVeR

    GoLD-ReaVeR Member

    Feb 19, 2014
    You managed to completely miss the point as to why MMR is broken. Good job.
  4. kamukag3e

    kamukag3e Member

    Jun 11, 2009
  5. GoLD-ReaVeR

    GoLD-ReaVeR Member

    Feb 19, 2014
    I've done this plenty of times in both the old forums and the new ones. In short:

    1. You should not be aiming for a 50% winrate as that creates situations where players become unable to change the outcome of the match. They can't see anything because all they see is opponents playing differently from the players they are playing with; which stunts the learning process.

    2. There's other variables at work here, such as behavior score, which interfere with the MMR balancing process. A badly behaved 3200MMR can be significantly stronger than a 'properly' behaving 3600MMR quite easily. This is double standards in action and it removes ANY validity that MMR has.

    3. The ELO system (and by extension MMR) was designed for dueling games. The score differences there and the reward differences there are built to accommodate weaker players playing against stronger players; which is exactly the thing MMR isn't doing. In Chess communities and fighting game communities, it is far more common for weaker players to prey on the stronger players as it is the best way to improve your own play and the ELO system will reflect that accurately. DotA players are kept in a slum and those fighting to get out are held back by players that are insanely stupid.

    4. Your evidence is "If I play a little better MY MMR goes up and if I play worse MY MMR goes down." This is anecdotal at best and cannot be used to prove that that is the norm. That is the advantage in having to prove something doesn't work, because you only need 1 counter example; and plenty have been provided over time.
    frigidsoul likes this.
  6. kamukag3e

    kamukag3e Member

    Jun 11, 2009
    @GoLD-ReaVeR, thank you for your insight.
    1. 50% chance of winning with player with the same rating is desired outcome, it's player's problem if they lack the understanding to see that's how it's supposed to be.
    There is no learning process if people are not aiming for it.

    On the topic of people (players) and not the system.
    There are plenty of people which are stagnating below 4,000 rating after more than 1,000 games.
    Which is OK if not aiming to become good and improve.
    But if you think that you are trying and still there - you are not doing it right.
    Most players below 4,000 rating are good at certain fields of the game and bad at others, all the things are easily accessible in internet and no more than 100 hours of learning theory and around another 100 hours of practicing any player could easily achieve 3,000-3,500 rating. A bit more (or like double) would easily get you to 4,000 rating.
    That's why the game is usually perceived as one with slow learning curve.

    Then again - most of the players as we know statistically are below 3,000 rating, some of them having more than 2,000, 3,000 and even 4,000 hours played. Why is that ? Because while a lot of them are not aiming to play better - a lot of them are just plain idiotic and their thinking pattern is at false.

    Being put in a game with relatively similar skilled players is good, much weaker players will hinder your growth (goes both ways) and higher players will crush you like a bug, not giving you a chance to learn anything. In such games having players floating around your skill level - a bit higher, a bit lower is just fine.
    If you are improving - your rating will be more than 50% and gradually you will meet better players.

    I know a lot of friends which after learning just few simple things increase their win-rate to respectable levels (55-60%) and reach heights which they didn't believe they can after some time.

    2. Indeed, there are plenty of side factors outside of rating, but they do not interfere as much as you think.
    Know plenty of toxic players with low behavior score, it does not impact their ability to impact the game. Slight deviation of the scale (rating) are expected, can't put the blame on behavior score entirely.

    A lot of players think themselves are being higher and just that their allies are making them fall (a.k.a. your average toxic player attitude). Then if they are still trying to improve and reach to someone (have few guys like that from those I coach), they still learn new stuff every single day, despite thinking they know everything. Things, which completely change their performance.
    Recently a guy who was around 2,800 rating after 4 hours of coaching and some clear lines manage to climb 700 rating in about 400 games. Keep in mind that those games were both party and solo while I'm talking about his solo rating exclusively and that he still didn't stay strict at my advises. So-o, outside of everything I specifically tell him, he stopped watching his allies and focus on improving his impact over the game. He still play really bad when he start playing a hero we did not talk about in specifics (a.k.a. showing his 2,800 rated skills), but when he is playing the roles and heroes we've talked about - easily outperform his peers at 3,500 rated games (his dotabuff , he review it to public after he started coaching with me, we make no more than 10 hours of coaching until that point and he stopped booking anymore, so his progress will perhaps be slowed down at some point).
    Basically - he was toxic but he did not perform better than his teammates while being in 2,800 and clearing his toxicity indeed give him a chance to raise his rating - but those were not actually related.

    3. Indeed, it is true. The thing you are forgetting about is that - there is no better suited system to that point and it's adapted to some extend. While not perfect is still good enough and much better than anything else on the table to that point.

    As for your example of weak chess players getting a rating boost of high rated players - it's true, but then again just to some extend.
    1) their skills must not be represented by their rating (a.k.a. they don't have a lot of matches or have a private tutor for some time), which also happens in DotA. After certain number of games they are still in the proper place in both cases. Just that Chess usually requires less because of it's type of games, where you can basically challenge everyone in a tournament - but then again - in DotA you can do that too, just very little players are skilled enough to enter tournaments.

    2) low rated chess players can only gain rating from players they can actually win against, as a skilled but new player still won't stand a chance playing against a grand master, so getting that dreamed +100 points will still be just a dream.
    In DotA, playing as 3,000 rated player against 4,000 rated player will still grant you more than 25 points, but the chances of winning are low. The chances consecutive wins are even lower. Then again - meeting a 5,000 player as 3,000 is just useless and the system nicely represent that.

    4. It is not anecdotal, and it's proven multiple times by examples (experiments), causal players rising in rating after learning new things, boosters and players with multiple accounts.
    If someone is delusional enough not to see how the system works - probably is delusional enough to not valuate properly his performance as well.

    It's easy to see that if you perform good every single game - you will win more then you lose.
    See your own games, count the numbers which your performance was better then your allies and see how many times it was worse (don't even bother looking at your enemies).
    KDA status is usually more then enough - and no - position doesn't matter as much.
    In a game where the 2 main cores have 10 kills, a lower score is expected from the others, if cores have 10 deaths, slightly higher score is expected from the others, assists should be the same, but supports being higher and cores lower.

    Then again - in most games supports thing having double the amounts of deaths of a core and the significantly lower number of kill participation (so basically add kills and assists) is just playing bad.
    As easily observable at higher level games and competitive matches - supports have on average slightly higher number of deaths - something like 6 (for cores) to 8 (for supports) and not something like 4 to 20, which usually happens in public games - when the supports are severely under performing.
    Now - can you tell me how many games statistically you were over-performing (having significantly better stats than your peers) and how many you were under-performing. Just pure statistics, no score or anything like that, no excuses as well.
  7. slipkwar

    slipkwar Member

    Dec 11, 2009
  8. frigidsoul

    frigidsoul Member

    Feb 10, 2015
    MM system doesn't use Elo system, It uses Natural selection so your mmr slowly goes up , you need to pass darwin test and strongs only will be selected for higher mmr.(I'm crusader so don't hate me)
    slipkwar likes this.
  9. Oesile

    Oesile Member

    Oct 8, 2016
    If dota used a darwinian system, we would have completely random matchmaking and if you lose you get sent to the shadow realm
    MrFrank and slipkwar like this.
  10. frigidsoul

    frigidsoul Member

    Feb 10, 2015
    you can see , your game and teammates wont change , you must do somthing about it, you must be the alpha guy in your team.
    see it as single player mode with 4 bots then you will improve cause you focus on yourself.strongs only pass mmr hell.
    simply below 4k mmr you must see this game as 1v9

    yea your team somtimes are good,bad , trash, smurf... and i accept that you can't trust your team , but if you want rely on your team you will never Improve.

    note : I hate current mmr system and don't defend it.
  11. kamukag3e

    kamukag3e Member

    Jun 11, 2009
    True to some extend, but completely false that teammates are bad, trash and smurfs, nor the game is solo versus 9.

    Most of the times allies are as good as you, until you put some effort to get much better.
    Even when a smurf/booster is raising rating, he doesn't play "1v9", he play "5v1+4meatshields", unless intentional game ruiners all allies are usually doing everything in their power to win - it's just they are bad.

    note: I don't think the current system is the best there is, just managable and perhaps the best there in any MOBA game so far.
    frigidsoul likes this.
  12. vorsybl

    vorsybl Member

    Mar 17, 2015
    All that stuff MIGHT be true, with the numbers. And you are only one person K, which is not representative of the entire sample size.

    Like gold reaver said, MMR is not the only variable. This absolutely absurd thing called "behavior score" completely skews all that statistical stuff. They made it so that a person has to spend months earning a "normal" behavior score so that they can play with people who do not INTENTIONALLY THROW, and TROLL.

    You're talking about winning games with no allies, against 5 people, even if they are mediocre, who are working together. Bro, if you're really as good as you say you are, win a match with people I have 10% winrate with? Show us a match where YOU are able to carry the game with NO allies, against 5 people who ARE working together. If you can show that, then your argument is valid.

    I agree with you that to climb you have to be better than your opponents, AND your allies. That's fine, but to be honest, that's not why I got into the game, and quite frankly, it's not enjoyable anymore. To "improve" in this game means fighting your opponents AND your allies. Every single fucking game in 1k mmr is lanes that don't make sense, supports offlane and carrys in the jungle or offlane instaed of safe lane, just retarded ass fucking conditions that should have no bearing on your "rating" or "skill" at the game.

    So the point is: TO be pro at dota, you have to win 1v9? because that's the jist im getting, and quit frankly, that's a shitty game. DOTA was dope because of 5v5, if to have high rating is 1v9, fuck this game. Waste of time honestly no offense.

    Maybe it's different in 6k, maybe it isn't. I'm never getting there so I'll never know, and there's more to life then sitting at my computer desperately trying to be a "professional player."

    My dad showed me an article that I just read about how free to play games ARE DESIGNED TO PREY on people. This dota pluss shit, leveling up heros, its all rigged to keep people playing, and feeding money to the developers

    Its pretty fucked up actually, regardless, still a good time playing with ya bro.

    It's just me honestly, I'm just tired of playing this kind of dota. If 90% of my games were normal, a support and carry going safe lane, an offlane going offlane, and a mid going mid, instead of two children going mid raging at each other for no reason, I would gladly accept my rating. My mechanics have gotten OD trihard recently; I've never tried this hard at dota before, and my rating is actually 1k lower than what i calibrated at, BEFORE i had any of these mechanics or triharding happening. Either I am severely autistic, or theres something else at play here...

    All that being said, hate to say it DOTA is still dope. The lore, the graphics, the sound, all the best in any game ive ever played. Unfortunately, my ranked experience has been absolutely horrible, and traumatizing as well. For the last month or two, about two months, everytime i win 2-5 matches in a row, the same players show up on my team, over and over. One ally, I have a 10% winrate with over the course of TWENTY ONE matches.

    21 x 40 min avg play time / match = ~ 14 hours of time. If I spend 8 hours a day playing, thats nearly TWO DAYS worth of time WASTED.

    And for some reason, that individual is allowed to continue playing in ranked, without being punished. In that clip, I'm not giving up, im not raging, im not even saying anything, im trying to play the fucking game. And what does that HUMAN BEING do? Takes my tranquil boots, probably thinking its funny, laughing in their chair, while im sitting here with my dick up my ass expending every last ounce of energy to win and not cancel out my previous win with a loss becuase of this disgusting peice of shit. I ended up losing that matchm 6-1

    Last ten games, I focused only on myself, with expectation of these confounds/variables mentioned. ONLY because they are all wins, and because my cousin hooked me up with a job at a coffee shop nearby, I will continue ONE LAST FUCKING TIME. One more month. If I don't wield 5,000 rating by the end of June, I will quit playing dota forever.
    Last edited: May 20, 2018 at 8:12 PM
  13. MrFrank

    MrFrank Member

    Oct 8, 2016
    I'm not sure if that is the most hilarious or messed up thing I've ever read ^

    Good luck either way :thumbup:
  14. Blarrg

    Blarrg Member

    Jun 12, 2009
    Ya the shadow pool it already exists dude lmao