EPTI don't get why people give so much importance to a public games MMR anyway. If one guy is playing troll builds and have your MMR, deal with it. Accept you're bad, if you want a serious game don't go in public games, that's like playing competitive on FFA server from CS. Even SC2 ladder which is accurate does not count to determine if a player is good or not at a competitive level. There are inhouse, where you can agree before if you want to play serious, and the only real legit way, get a team, play in leagues.
Ranked matchmaking coming to Dota 2 - Page 69
| Forum Index > Closed |
Please keep the QQ to a minimum if you do not like this update. We are happy to hear your reasoning for not liking a ranked system, but no "OMG VOLVO WHY" posts. | ||
|
nojok
France15846 Posts
I don't get why people give so much importance to a public games MMR anyway. If one guy is playing troll builds and have your MMR, deal with it. Accept you're bad, if you want a serious game don't go in public games, that's like playing competitive on FFA server from CS. Even SC2 ladder which is accurate does not count to determine if a player is good or not at a competitive level. There are inhouse, where you can agree before if you want to play serious, and the only real legit way, get a team, play in leagues. | ||
|
zezamer
Finland5701 Posts
| ||
|
n0ise
3452 Posts
On December 27 2013 19:04 TechSc2 wrote: I see most people only think of KDA when talking about stats, and even then, KDA alone doesn't mean you need the last hit to perform "well", assist count as well. But lets say people will play furion more, the fact that they are paying attention to the map to use ult or to TP in to kill someone...... wait.... can it be.....? THAT'S EXACTLY WHAT A PRO PLAYER DOES WITH FURION TO HELP THE TEAM?!?!?! But lets say people will play Lion more, the fact that they are paying attention to the map to use ult or shadow blade in to kill someone...... wait.... can it be.....? THAT'S EXACTLY WHAT A PRO PLAYER DOES WITH LION TO HELP THE TEAM?!?!?! | ||
|
fearus
China2164 Posts
| ||
|
Andre
Slovenia3523 Posts
wat, totally different systems | ||
|
Erasme
Bahamas15899 Posts
| ||
|
Alur
Denmark3900 Posts
On December 27 2013 20:13 fearus wrote: Think of Lebron back when he was at Cleveland, if the NBA used the MMR system, Lebron would be a 2000 MMR player. Quoteworthy stuff right here. | ||
|
nojok
France15846 Posts
On December 27 2013 20:28 Andre wrote: ^And once again somebody compares dota/online gaming to sports. wat, totally different systems It's not that inaccurate, it's just that pub games (MMR revealed or not) are the sunday football match in your district whereas if you want to play competitive you play in a club, train with teammates 3 times a week and you have a league or cup match the sunday. It's two different approaches to the same game, it's not because they revealed the MMR (which existed but was hidden before) that it suddenly became a competitive way to play. | ||
|
Plansix
United States60190 Posts
On December 27 2013 20:13 fearus wrote: Think of Lebron back when he was at Cleveland, if the NBA used the MMR system, Lebron would be a 2000 MMR player. Getting paid millions of dollars to be with a low ranked team. Seriously, I know people find it hard to accept the fact that they are as bad as the people they are matched with, but it's true. You are not as good at dota as you think you are. | ||
|
Kupon3ss
時の回廊10066 Posts
| ||
|
sCCrooked
Korea (South)1306 Posts
On December 27 2013 15:47 Penecks wrote: If you play consistently at a level above your MMR no amount of noob teammates/feeders/whatever will keep you down. Basing it on anything other than win/loss will eventually just lead to abuse. In the end it's still the same shit, since you'll inevitably have games where you feed and and still win. Do you wanna lose points for that? Sorry but this is definitely not true. When your team picks all wrong for any general sense and even worse when you consider your enemy team's far superior composition and then proceeds to take that already-wrong composition to even more horrible lane combos (double support top with a double melee carry bot that doesn't speak any language you know for example), you end up with a literally unwinnable scenario. What you should have stated was that if you are matched with noob/feeder/whatever that there is consistently nothing you will ever be able to do because not even Dendi himself could pull some of these nightmare-teams to victory against a decent comp. NotYango wrote: No, actually he has the point quite clear. You can have a system that represents peoples' skill very well. It won't mean shit. Because as soon as the rating system is not aligned simply with winning and losing games, the goal is no longer to win and lose games. It's to pad your stats in such a way that you gain the most rating. Any system that uses any metrics other than pure win/loss introduces a route to abuse that system to gain points. It's not that you cannot establish a system that grades your skill very accurately if you're playing normally. It's that once the rating system does not align with the actual goal of the game, "playing normally" doesn't happen anymore. The team sports analogy is very poor because the nature of the game is such that the reward for winning outweighs the intangible reward of being ranked higher so trying to game the system in order to improve one's ranking at the expense of your team or your chances of winning doesn't make sense. In an online ladder where the only reward IS being ranked higher, there's nothing stopping people from gaming the system. I still disagree. Like the poster above you, you seem to think that there's no viable way to implement MMR better than it is right now (and we don't even know what criteria are used or how heavily they weigh in the equation). Why not have the win/loss matter but with other factors able to almost nullify it? If you are a support who ended up healing 20,000+ health, buying 26 sets of wards and placed a total of 52 wards, less than 5 seconds teleport time after recorded damage on any tower and less than a 2 second interval between enemy team heroes recording damage on your team's carry hero(s) and you casting a support spell to heal/move/protect them but you lost the game, there should be some way in the equation to make you not only not lose points, but gain them instead. In the same token, if those same qualities aren't carried out as well, there should be some way of lessening the points gained from a win or even losing points from a win. If you aren't in team qeue, the system should in fact base everything off of your own performance based on your hero, length of game, and comparative performance stats from your other teammates. If the carry feeds like crap, but you had high recorded support skills because the enemy team kept 4 ganking your safelane and your other lanes never called missing, that also needs to be taken into effect (shorter game duration is a factor in these, team comp is a consideration, etc). As complicated as that all sounds, that's the task at hand when you try to make a ranked system properly. If you don't hold yourself to this standard, you should not try putting ranked in because of how shitty and stupid the calculations are (they wouldn't make sense in any other implementation). KDA and Win/Loss should matter, but not too heavily, as they should easily be overturned by other stats weighing in depending entirely on who you picked and who you were playing with (4 carries 1 support calculated significantly different than 2 carries 2 supports and a snowballing semi-carry). n0ise wrote: But lets say people will play Lion more, the fact that they are paying attention to the map to use ult or shadow blade in to kill someone...... No, because if they follow the standards I just outlined, a lion pick would not even be gauged on kills. They would not factor in at all. They would be gauged by support standards. Plansix wrote: Seriously, I know people find it hard to accept the fact that they are as bad as the people they are matched with, but it's true. You are not as good at dota as you think you are. I know people find it hard to accept that the company that makes their game could possibly make mistakes and implement things badly, but its true. Valve isn't as good at implementing proper matchmaking systems as you think they are. Taking a test where the criteria for figuring out your score is poorly figured out will always have large numbers of people getting incorrect ratings. Also if your statement had any truth to it, we would not see such radical differences in peoples' party ratings vs their "so-called single" ratings. | ||
|
Plansix
United States60190 Posts
| ||
|
Erasme
Bahamas15899 Posts
| ||
|
Alur
Denmark3900 Posts
On December 27 2013 23:49 Plansix wrote: I am sure the Valve is using all the data available to them to make sure the match making is working well. If someone feeds hard during a game due to their own mistakes, I have no doubt that Valve is tracking that and factors it into how the MMR is adjusted.. It took them some time to come out with this system and there is no way it just takes into account W/L ratio and that is it. In post-calibration games, the change in rating has always been equal for all the members in my party. I believe the change in rating is purely based on the balance between the teams, and the outcome of the match. | ||
|
rabidch
United States20289 Posts
On December 27 2013 23:23 Kupon3ss wrote: the Cavaliers were actually one of the better teams of the league with decent regular season records and having made it to the finals in 2007. If NBA used the mmr system, they would have been top 5 team mmr because unlike a lot of terrible dota players in this thread, he was actually able to carry by playing well not to mention the lebrons cavaliers would still stomp over a more accurate pool with 99% of foreign teams, d league teams, all college teams, and amateurs. | ||
|
Talin
Montenegro10532 Posts
On December 27 2013 23:43 sCCrooked wrote: I know people find it hard to accept that the company that makes their game could possibly make mistakes and implement things badly, but its true. Valve isn't as good at implementing proper matchmaking systems as you think they are. Matchmaking systems aren't rocket science, you don't need a particular level of expertise or some arcane knowledge to implement one. They are relatively simple mechanisms, and tend to work better the simpler they are. In a game of Dota's level of complexity, the best choice is simply to measure how good you are at winning. Remember that the reason matchmaking exists to begin with is mostly to save players the hassle of having to look for people to play with and set up their own games. Ultimately, the only thing that matters is that a) the system can find you a game in a few minutes or less b) you are content enough to keep playing the game Neither Valve, nor quite honestly anyone else, is going to give a shit that you're not happy with the teammates you get every single game. Besides, the problem you're trying to solve is unsolvable, because it is essentially rooted in human behavior. Modeling a matchmaking algorithm around that is just going to make people play (and abuse) the system instead of playing the game, and that's not going to make anyone happy. Not to mention that any significant patch (ie 6.79) could break it by changing the value of different aspects of the game. | ||
|
Andre
Slovenia3523 Posts
On December 27 2013 23:43 sCCrooked wrote: Sorry but this is definitely not true. When your team picks all wrong for any general sense and even worse when you consider your enemy team's far superior composition and then proceeds to take that already-wrong composition to even more horrible lane combos (double support top with a double melee carry bot that doesn't speak any language you know for example), you end up with a literally unwinnable scenario. What you should have stated was that if you are matched with noob/feeder/whatever that there is consistently nothing you will ever be able to do because not even Dendi himself could pull some of these nightmare-teams to victory against a decent comp. I still disagree. Like the poster above you, you seem to think that there's no viable way to implement MMR better than it is right now (and we don't even know what criteria are used or how heavily they weigh in the equation). Why not have the win/loss matter but with other factors able to almost nullify it? If you are a support who ended up healing 20,000+ health, buying 26 sets of wards and placed a total of 52 wards, less than 5 seconds teleport time after recorded damage on any tower and less than a 2 second interval between enemy team heroes recording damage on your team's carry hero(s) and you casting a support spell to heal/move/protect them but you lost the game, there should be some way in the equation to make you not only not lose points, but gain them instead. In the same token, if those same qualities aren't carried out as well, there should be some way of lessening the points gained from a win or even losing points from a win. If you aren't in team qeue, the system should in fact base everything off of your own performance based on your hero, length of game, and comparative performance stats from your other teammates. If the carry feeds like crap, but you had high recorded support skills because the enemy team kept 4 ganking your safelane and your other lanes never called missing, that also needs to be taken into effect (shorter game duration is a factor in these, team comp is a consideration, etc). As complicated as that all sounds, that's the task at hand when you try to make a ranked system properly. If you don't hold yourself to this standard, you should not try putting ranked in because of how shitty and stupid the calculations are (they wouldn't make sense in any other implementation). KDA and Win/Loss should matter, but not too heavily, as they should easily be overturned by other stats weighing in depending entirely on who you picked and who you were playing with (4 carries 1 support calculated significantly different than 2 carries 2 supports and a snowballing semi-carry). No, because if they follow the standards I just outlined, a lion pick would not even be gauged on kills. They would not factor in at all. They would be gauged by support standards. I know people find it hard to accept that the company that makes their game could possibly make mistakes and implement things badly, but its true. Valve isn't as good at implementing proper matchmaking systems as you think they are. Taking a test where the criteria for figuring out your score is poorly figured out will always have large numbers of people getting incorrect ratings. Also if your statement had any truth to it, we would not see such radical differences in peoples' party ratings vs their "so-called single" ratings. You give an example of a game where your team has everything shitty and the enemy is next level gosu pro team. Tell me how many times does that happen? There's gonna be times where your team is vastly inferior to the enemy's and you're gonna get stomped, the point is over a larger sample size of games the system that valve uses is good. For every "bad" player on your team there's one on the other team. Implementing some complex criteria for gaining rating is dumb, because no matter what you do people are gonna find ways to abuse it and not play the game to win - but for rating. Only thing that matters is your own impact in any given game over large number of games. edit: also, what's this about radical differences in party rating vs solo? I don't see any problems, some people's stacks are performing better than each of the individuals, and some are performing worse - what's wrong here? | ||
|
Plansix
United States60190 Posts
On December 27 2013 23:43 sCCrooked wrote: I know people find it hard to accept that the company that makes their game could possibly make mistakes and implement things badly, but its true. Valve isn't as good at implementing proper matchmaking systems as you think they are. Taking a test where the criteria for figuring out your score is poorly figured out will always have large numbers of people getting incorrect ratings. Also if your statement had any truth to it, we would not see such radical differences in peoples' party ratings vs their "so-called single" ratings. Clearly, that must be it. MMR is messed up and you are way better at Dota than your rank. You and others who have been misplaced in a bracket filled with scrubs have been wronged. The injustice of it all is to much and Valve should issue a formal apology for the harm they have done. | ||
|
MotherOfRunes
Germany2862 Posts
On December 28 2013 01:12 Plansix wrote: Clearly, that must be it. MMR is messed up and you are way better at Dota than your rank. You and others who have been misplaced in a bracket filled with scrubs have been wronged. The injustice of it all is to much and Valve should issue a formal apology for the harm they have done. there are misplaced people no doubt, for example people who dont use smurfs when teaching ultra hard noob friends for months....believe me your MMR will suffer und u end up somewhere u shouldnt be. the thing is though prolly 99% of people are in the right place, and there are way to many cry babies compared to the 1 % who maybe is misplaced. i bet if i would replay stalk the most of those cry babies here i would have a good laugh and witness people who just are in the pool where they belong :D | ||
|
Laurens
Belgium4558 Posts
On December 28 2013 00:01 Erasme wrote: So you want to force a meta, where mirana is forced to play carry or mid else the player will lose/gain less elo ? + Show Spoiler + There are smarter ways to deal with this. For instance they could compare the KDA you got on a certain hero with the KDA that other people get on THAT hero. That stops everyone from picking gankers. Granted people would still attempt to killsteal and perhaps focus too much on ganking. And I'm sure there can be cases of doing a lot for your team, yet having low KDA. You shouldn't be punished for that. It's a very difficult matter, and perhaps part of the reason why visible MMR has been delayed this much. As it is right now though (only win/loss matters) I'm not very happy with the system, but meh. e: misread post, nevermind xD | ||
| ||
