|
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. |
LMAO this mmr system
I lose more mmr for shit one-sided game than for legit close games lost
The MMR guy at Valve is as clever as the guys fixing the bugs
|
On December 21 2013 22:36 govie wrote:Show nested quote +On December 21 2013 22:16 Targe wrote:On December 21 2013 21:24 govie wrote:On December 21 2013 21:14 Targe wrote:On December 21 2013 20:42 DrPandaPhD wrote:On December 21 2013 12:13 SS.Shark wrote:
This is the worst rating system Volvo could come up with. You lose points no matter what when you lose a game. No matter how much you carry a team. Doesnt matter if you end up 21-0 or 0-21. Doesnt matter if you're support and you finish 5-1 and your carry is 0-5. Same crap. And how is this motivating/rewarding players for their efforts ? You still end up in the same queue with the feeders you had in the previous game because you lost the same amount of points and your MMR is the same. Rdiculous If you are genuinely better than the people you get matched with you will win like 70%ish of the games. Otherwise you are stuck on that rating for a reason.. maybe if youre a lot better, but if youre just better by say 500 mmr's skill worth then this isnt true. not that i agree with first guy yes it is. it says that u will win more then u loose if paired to the same 2 teams again. Winn/loss is easy to calculate, skill however is not. I think u should generally not assume mmr means skillrating. Ofcourse for most there will some sort of correlation between the two, but a higher mmr then your friend doesnt say you are more skilled at all, it says your wins-loss ratio is better then that of your friend within the heropool you and he played. I myself would rate someone that plays random in ranked of higher skill then someone that doesnt if they would have around the same mmr. Thats why mmr is fine if its based on win-loss mostly. Because if the samplesize is big enough, u will see a pretty descent and accurate rating to give everyone a nice game of there skilllevel within the size of there heropool they like to play with. Even if they play random, or only 1 hero. MMR isnt there to give u a rating on how good u are compared to others, it exists to give u the most pleasant dotaexperience Valve can offer you, thats something totally different. youre saying someone whos slightly better can carry his team to a 70% wr every game? right. U should not ask a question which u allready know the answer too. No, doesnt seem likely now does it? If u have a little higher mmr u will never carry a team to 70% without u playing an OP hero and owning with it. 2nd u quote me but ignore my main point. You can also have a higher winrate because u only counterpick heros in AP. Does this imply that the player in question is more skilled? No, i dont think so, it implies he knows something about the heros. Skill, knowledge, execution is hard to measure in the dota2 and still providing a fun gamingexperiences at that. But if he played 100 games in the same setting, he should win more then his friends. As i see it, winrates are the only possibility to ensure a ratingsystem so that everybody gets alot of enjoyment no matter if they only play 1 hero, allways random, random draft, cm or cd.
DotA 2 is roughly 10% mechanics and about 90% knowledge/experience. The guy counterpicking you is just better than you. There might be a handful of match-ups where you literally just get shutout by a counterpick, and most of those involve dumb heroes like OD, Skywrath, etc. who are easy to shutdown with a gank or simply leaving the lane and ganking another once you soak up enough exp.
The only problem with the rating system right now is that some dumbass always ques with his friend who is almost 1k points lower than him. At that point, it's just pure luck where the lower skill friend is, as if he decides to take an important pub position like mid/offlaner, he will just feed relentlessly and lose you the game.
|
On December 22 2013 01:39 superstartran wrote:Show nested quote +On December 21 2013 22:36 govie wrote:On December 21 2013 22:16 Targe wrote:On December 21 2013 21:24 govie wrote:On December 21 2013 21:14 Targe wrote:On December 21 2013 20:42 DrPandaPhD wrote:On December 21 2013 12:13 SS.Shark wrote:
This is the worst rating system Volvo could come up with. You lose points no matter what when you lose a game. No matter how much you carry a team. Doesnt matter if you end up 21-0 or 0-21. Doesnt matter if you're support and you finish 5-1 and your carry is 0-5. Same crap. And how is this motivating/rewarding players for their efforts ? You still end up in the same queue with the feeders you had in the previous game because you lost the same amount of points and your MMR is the same. Rdiculous If you are genuinely better than the people you get matched with you will win like 70%ish of the games. Otherwise you are stuck on that rating for a reason.. maybe if youre a lot better, but if youre just better by say 500 mmr's skill worth then this isnt true. not that i agree with first guy yes it is. it says that u will win more then u loose if paired to the same 2 teams again. Winn/loss is easy to calculate, skill however is not. I think u should generally not assume mmr means skillrating. Ofcourse for most there will some sort of correlation between the two, but a higher mmr then your friend doesnt say you are more skilled at all, it says your wins-loss ratio is better then that of your friend within the heropool you and he played. I myself would rate someone that plays random in ranked of higher skill then someone that doesnt if they would have around the same mmr. Thats why mmr is fine if its based on win-loss mostly. Because if the samplesize is big enough, u will see a pretty descent and accurate rating to give everyone a nice game of there skilllevel within the size of there heropool they like to play with. Even if they play random, or only 1 hero. MMR isnt there to give u a rating on how good u are compared to others, it exists to give u the most pleasant dotaexperience Valve can offer you, thats something totally different. youre saying someone whos slightly better can carry his team to a 70% wr every game? right. U should not ask a question which u allready know the answer too. No, doesnt seem likely now does it? If u have a little higher mmr u will never carry a team to 70% without u playing an OP hero and owning with it. 2nd u quote me but ignore my main point. You can also have a higher winrate because u only counterpick heros in AP. Does this imply that the player in question is more skilled? No, i dont think so, it implies he knows something about the heros. Skill, knowledge, execution is hard to measure in the dota2 and still providing a fun gamingexperiences at that. But if he played 100 games in the same setting, he should win more then his friends. As i see it, winrates are the only possibility to ensure a ratingsystem so that everybody gets alot of enjoyment no matter if they only play 1 hero, allways random, random draft, cm or cd. DotA 2 is roughly 10% mechanics and about 90% knowledge/experience. The guy counterpicking you is just better than you. There might be a handful of match-ups where you literally just get shutout by a counterpick, and most of those involve dumb heroes like OD, Skywrath, etc. who are easy to shutdown with a gank or simply leaving the lane and ganking another once you soak up enough exp.
So you're one of those guys that always wait till creep spawn to counterpick in pubs cause you think it makes you good? Sad
|
On December 22 2013 01:40 Fatalize wrote:Show nested quote +On December 22 2013 01:39 superstartran wrote:On December 21 2013 22:36 govie wrote:On December 21 2013 22:16 Targe wrote:On December 21 2013 21:24 govie wrote:On December 21 2013 21:14 Targe wrote:On December 21 2013 20:42 DrPandaPhD wrote:On December 21 2013 12:13 SS.Shark wrote:
This is the worst rating system Volvo could come up with. You lose points no matter what when you lose a game. No matter how much you carry a team. Doesnt matter if you end up 21-0 or 0-21. Doesnt matter if you're support and you finish 5-1 and your carry is 0-5. Same crap. And how is this motivating/rewarding players for their efforts ? You still end up in the same queue with the feeders you had in the previous game because you lost the same amount of points and your MMR is the same. Rdiculous If you are genuinely better than the people you get matched with you will win like 70%ish of the games. Otherwise you are stuck on that rating for a reason.. maybe if youre a lot better, but if youre just better by say 500 mmr's skill worth then this isnt true. not that i agree with first guy yes it is. it says that u will win more then u loose if paired to the same 2 teams again. Winn/loss is easy to calculate, skill however is not. I think u should generally not assume mmr means skillrating. Ofcourse for most there will some sort of correlation between the two, but a higher mmr then your friend doesnt say you are more skilled at all, it says your wins-loss ratio is better then that of your friend within the heropool you and he played. I myself would rate someone that plays random in ranked of higher skill then someone that doesnt if they would have around the same mmr. Thats why mmr is fine if its based on win-loss mostly. Because if the samplesize is big enough, u will see a pretty descent and accurate rating to give everyone a nice game of there skilllevel within the size of there heropool they like to play with. Even if they play random, or only 1 hero. MMR isnt there to give u a rating on how good u are compared to others, it exists to give u the most pleasant dotaexperience Valve can offer you, thats something totally different. youre saying someone whos slightly better can carry his team to a 70% wr every game? right. U should not ask a question which u allready know the answer too. No, doesnt seem likely now does it? If u have a little higher mmr u will never carry a team to 70% without u playing an OP hero and owning with it. 2nd u quote me but ignore my main point. You can also have a higher winrate because u only counterpick heros in AP. Does this imply that the player in question is more skilled? No, i dont think so, it implies he knows something about the heros. Skill, knowledge, execution is hard to measure in the dota2 and still providing a fun gamingexperiences at that. But if he played 100 games in the same setting, he should win more then his friends. As i see it, winrates are the only possibility to ensure a ratingsystem so that everybody gets alot of enjoyment no matter if they only play 1 hero, allways random, random draft, cm or cd. DotA 2 is roughly 10% mechanics and about 90% knowledge/experience. The guy counterpicking you is just better than you. There might be a handful of match-ups where you literally just get shutout by a counterpick, and most of those involve dumb heroes like OD, Skywrath, etc. who are easy to shutdown with a gank or simply leaving the lane and ganking another once you soak up enough exp. So you're one of those guys that always wait till creep spawn to counterpick in pubs cause you think it makes you good? Sad
Do people complain when pros counter pick each other in drafts? Like I said, 90% of DotA is knowledge of certain situations and how to get around said situations. Just because you got counterpicked doesn't mean you still can't get around it. If the guy is lower skill than you, he will not counter you as hard as he should, so you can go soak exp and simply gank the shit out of everyone on the map (i.e. Nightstalker vs OD). If you're as good as you think you are, you can play around most counter picks.
|
On December 22 2013 01:39 superstartran wrote:Show nested quote +On December 21 2013 22:36 govie wrote:On December 21 2013 22:16 Targe wrote:On December 21 2013 21:24 govie wrote:On December 21 2013 21:14 Targe wrote:On December 21 2013 20:42 DrPandaPhD wrote:On December 21 2013 12:13 SS.Shark wrote:
This is the worst rating system Volvo could come up with. You lose points no matter what when you lose a game. No matter how much you carry a team. Doesnt matter if you end up 21-0 or 0-21. Doesnt matter if you're support and you finish 5-1 and your carry is 0-5. Same crap. And how is this motivating/rewarding players for their efforts ? You still end up in the same queue with the feeders you had in the previous game because you lost the same amount of points and your MMR is the same. Rdiculous If you are genuinely better than the people you get matched with you will win like 70%ish of the games. Otherwise you are stuck on that rating for a reason.. maybe if youre a lot better, but if youre just better by say 500 mmr's skill worth then this isnt true. not that i agree with first guy yes it is. it says that u will win more then u loose if paired to the same 2 teams again. Winn/loss is easy to calculate, skill however is not. I think u should generally not assume mmr means skillrating. Ofcourse for most there will some sort of correlation between the two, but a higher mmr then your friend doesnt say you are more skilled at all, it says your wins-loss ratio is better then that of your friend within the heropool you and he played. I myself would rate someone that plays random in ranked of higher skill then someone that doesnt if they would have around the same mmr. Thats why mmr is fine if its based on win-loss mostly. Because if the samplesize is big enough, u will see a pretty descent and accurate rating to give everyone a nice game of there skilllevel within the size of there heropool they like to play with. Even if they play random, or only 1 hero. MMR isnt there to give u a rating on how good u are compared to others, it exists to give u the most pleasant dotaexperience Valve can offer you, thats something totally different. youre saying someone whos slightly better can carry his team to a 70% wr every game? right. U should not ask a question which u allready know the answer too. No, doesnt seem likely now does it? If u have a little higher mmr u will never carry a team to 70% without u playing an OP hero and owning with it. 2nd u quote me but ignore my main point. You can also have a higher winrate because u only counterpick heros in AP. Does this imply that the player in question is more skilled? No, i dont think so, it implies he knows something about the heros. Skill, knowledge, execution is hard to measure in the dota2 and still providing a fun gamingexperiences at that. But if he played 100 games in the same setting, he should win more then his friends. As i see it, winrates are the only possibility to ensure a ratingsystem so that everybody gets alot of enjoyment no matter if they only play 1 hero, allways random, random draft, cm or cd. DotA 2 is roughly 10% mechanics and about 90% knowledge/experience. The guy counterpicking you is just better than you. There might be a handful of match-ups where you literally just get shutout by a counterpick, and most of those involve dumb heroes like OD, Skywrath, etc. who are easy to shutdown with a gank or simply leaving the lane and ganking another once you soak up enough exp. The only problem with the rating system right now is that some dumbass always ques with his friend who is almost 1k points lower than him. At that point, it's just pure luck where the lower skill friend is, as if he decides to take an important pub position like mid/offlaner, he will just feed relentlessly and lose you the game.
I understand what your trying to say, certainly for soloqeue. Although I like many others dont play like that, either becuz we dont want too or we cant! But i understand that if u have that kind of knowledge u could use it often ingame.
|
On December 22 2013 01:49 govie wrote:Show nested quote +On December 22 2013 01:39 superstartran wrote:On December 21 2013 22:36 govie wrote:On December 21 2013 22:16 Targe wrote:On December 21 2013 21:24 govie wrote:On December 21 2013 21:14 Targe wrote:On December 21 2013 20:42 DrPandaPhD wrote:On December 21 2013 12:13 SS.Shark wrote:
This is the worst rating system Volvo could come up with. You lose points no matter what when you lose a game. No matter how much you carry a team. Doesnt matter if you end up 21-0 or 0-21. Doesnt matter if you're support and you finish 5-1 and your carry is 0-5. Same crap. And how is this motivating/rewarding players for their efforts ? You still end up in the same queue with the feeders you had in the previous game because you lost the same amount of points and your MMR is the same. Rdiculous If you are genuinely better than the people you get matched with you will win like 70%ish of the games. Otherwise you are stuck on that rating for a reason.. maybe if youre a lot better, but if youre just better by say 500 mmr's skill worth then this isnt true. not that i agree with first guy yes it is. it says that u will win more then u loose if paired to the same 2 teams again. Winn/loss is easy to calculate, skill however is not. I think u should generally not assume mmr means skillrating. Ofcourse for most there will some sort of correlation between the two, but a higher mmr then your friend doesnt say you are more skilled at all, it says your wins-loss ratio is better then that of your friend within the heropool you and he played. I myself would rate someone that plays random in ranked of higher skill then someone that doesnt if they would have around the same mmr. Thats why mmr is fine if its based on win-loss mostly. Because if the samplesize is big enough, u will see a pretty descent and accurate rating to give everyone a nice game of there skilllevel within the size of there heropool they like to play with. Even if they play random, or only 1 hero. MMR isnt there to give u a rating on how good u are compared to others, it exists to give u the most pleasant dotaexperience Valve can offer you, thats something totally different. youre saying someone whos slightly better can carry his team to a 70% wr every game? right. U should not ask a question which u allready know the answer too. No, doesnt seem likely now does it? If u have a little higher mmr u will never carry a team to 70% without u playing an OP hero and owning with it. 2nd u quote me but ignore my main point. You can also have a higher winrate because u only counterpick heros in AP. Does this imply that the player in question is more skilled? No, i dont think so, it implies he knows something about the heros. Skill, knowledge, execution is hard to measure in the dota2 and still providing a fun gamingexperiences at that. But if he played 100 games in the same setting, he should win more then his friends. As i see it, winrates are the only possibility to ensure a ratingsystem so that everybody gets alot of enjoyment no matter if they only play 1 hero, allways random, random draft, cm or cd. DotA 2 is roughly 10% mechanics and about 90% knowledge/experience. The guy counterpicking you is just better than you. There might be a handful of match-ups where you literally just get shutout by a counterpick, and most of those involve dumb heroes like OD, Skywrath, etc. who are easy to shutdown with a gank or simply leaving the lane and ganking another once you soak up enough exp. The only problem with the rating system right now is that some dumbass always ques with his friend who is almost 1k points lower than him. At that point, it's just pure luck where the lower skill friend is, as if he decides to take an important pub position like mid/offlaner, he will just feed relentlessly and lose you the game. I understand what your trying to say, certainly for soloqeue. Although I like many others dont play like that, either becuz we dont want too or we cant! But i understand that if u have that kind of knowledge u could use it often ingame.
That is exactly the point. Just because you get counterpicked doesn't mean the game is instantly over. Most pub games are decided by any major mistakes made in the mid game, or if there's a severe mismatch in skill level on a certain important lane like mid or the offlaner vs dual lane.
Also, counter picking is optimal play. Ranked matchmaking encourages this. If you're not playing optimal, you really don't have a right to complain really.
|
Most of the counterpickers I played against weren't even that good and failed most of the time.
|
went 6-6 (2 games were abandons or disconnects) and ended @ 3054, since then lost 7 games in a row because of trolls and facing 2-3 man stacks... i think the games are worse than unranked. gonna take a break from dota (again)
p.s. at least it made me realize that after 230 games, i still dont have a clue about laning, drafting in CM or map awareness.
|
On December 22 2013 04:46 Pik wrote:
went 6-6 (2 games were abandons or disconnects) and ended @ 3054, since then lost 7 games in a row because of trolls and facing 2-3 man stacks... i think the games are worse than unranked. gonna take a break from dota (again)
p.s. at least it made me realize that after 230 games, i still dont have a clue about laning, drafting in CM or map awareness.
At least you can join a very large group of people. Most people, even those that have more than 1000 games played still don't know about that crap. The problem is that you have to make an effort to learn them. Most people don't feel like taking the time to learn something in a game.
|
What a lyfe, terrible rating.
|
Is something wrong with ranked? I won 3 games in a row and I didnt have an ass playing as ally in all 3 games.
|
Northern Ireland22213 Posts
On December 22 2013 08:52 SkelA wrote:
Is something wrong with ranked? I won 3 games in a row and I didnt have an ass playing as ally in all 3 games.
People might actually want to win!
|
On December 22 2013 01:37 Fatalize wrote:
LMAO this mmr system
I lose more mmr for shit one-sided game than for legit close games lost
The MMR guy at Valve is as clever as the guys fixing the bugs
makes sense? if u get stomped means other teams much better than you (theoretically) so the difference needs to grow while very close games means teams are evenly matched, therefore should be even mmrd
|
On December 22 2013 10:00 teddyoojo wrote:Show nested quote +On December 22 2013 01:37 Fatalize wrote:
LMAO this mmr system
I lose more mmr for shit one-sided game than for legit close games lost
The MMR guy at Valve is as clever as the guys fixing the bugs makes sense? if u get stomped means other teams much better than you (theoretically) so the difference needs to grow while very close games means teams are evenly matched, therefore should be even mmrd
Stomping a team you're expected to win against gives you less MMR. You get a large amount of MMR when you beat a team that you're expected to lose to (and they lose more MMR) because your rating should be higher than theirs since you beat them.
The uncertainty also increases when you have an unexpected result.
|
So after the first few ranked matches I have the following question, with respect to some posts above
Is it practically even possible to play a match that you are supposed to win/lose? Most of my matches so far have features very even teams in terms of rating: I checked a few other players. Also after each game, and my rating changed by 23-29 after each game; if there were uneven teams, I would expect some bigger (worse team wins) or smaller (better team wins) margins of rating changes.
I would feel that if every match is balanced (each team having ~50% chance to win based on rating), then after placement matches, either result is equally likely and therefore not surpirising, and therefore
On December 07 2013 12:08 Heyoka wrote:
A surprising match outcome will tend to cause an increase in uncertainty.
(from the op) has a 0% chance of practically happening.
|
It's not going to be like 70% vs 30% rating difference. More like 48% vs 52%. If you win with 48% you get more than when you win with 52%.
|
On December 22 2013 10:00 teddyoojo wrote:Show nested quote +On December 22 2013 01:37 Fatalize wrote:
LMAO this mmr system
I lose more mmr for shit one-sided game than for legit close games lost
The MMR guy at Valve is as clever as the guys fixing the bugs makes sense? if u get stomped means other teams much better than you (theoretically) so the difference needs to grow while very close games means teams are evenly matched, therefore should be even mmrd
If it means the other team is much better thann us, it means Vlave matchmaking is fucking godawful at matching teams. And it is, but it doesnt mean matching you shitly should also induce you losing more MMR
|
No matchmaking can always match 100% equal opponents, even in a 1v1 game like SC2 you have favored/unfavored matchups. Balanced for an algorithm doesn't mean ~50%. It means exactly 50%.
|
United States12280 Posts
On December 23 2013 04:16 Mafe wrote:So after the first few ranked matches I have the following question, with respect to some posts above Is it practically even possible to play a match that you are supposed to win/lose? Most of my matches so far have features very even teams in terms of rating: I checked a few other players. Also after each game, and my rating changed by 23-29 after each game; if there were uneven teams, I would expect some bigger (worse team wins) or smaller (better team wins) margins of rating changes. I would feel that if every match is balanced (each team having ~50% chance to win based on rating), then after placement matches, either result is equally likely and therefore not surpirising, and therefore Show nested quote +On December 07 2013 12:08 Heyoka wrote:
A surprising match outcome will tend to cause an increase in uncertainty.
(from the op) has a 0% chance of practically happening.
Say you're at 3000 rating with 200 uncertainty. That's pretty low, meaning a high confidence that you are where you should be. Statistically, players at 3200 rating will have, say, a 60% chance at beating a 3000. But let's say you go 10-0 against some same-skill players (or slightly higher, maybe +100 rating over you) and after that streak you're at 3200 rating yourself. Uncertainty will increase gradually with each game because you're bucking the system. So now your uncertainty is probably 300 (ranging from 2900-3500 when it used to range from 2800-3200, so it's pretty sure that you're not a 2800-caliber player, but it's not convinced you're not on edge case a 2900 player).
It's not as dramatic as you're thinking, where you're 3000 and you have a 1% chance of playing 4000-rated players and then if you win your uncertainty increases by 1000. It's a lot more gradual and methodical. There's probably also an uncertainty threshold where, if you hit it, you get thrown back into calibration again.
|
RMM is broken.
I lost 8/10 calibration matches. Admittedly in 2 of those I had -ve KDA and lost lane. In the rest... won my lane, tried to help team, they just YOLO'ed and died.
Since then, been getting a lot of TBD's on team going through calibration... Lost 11/12 and my MMR has gone from around 3.5K (solo) to 3.2K
MORAL OF THE STORY: This game is still broken for a Solo player. Get you some friends and play in parties.
THE THING THAT SUCKS: To dig myself out of my current rating is close to impossible. If I keep getting matched with tools I jsut don't see a way out. I remember I BARELY got out of "normal" bracket when the old system was up. Got to high and was on win streak after win streak (p.s. I only achieved this by going solo-queue only.. no parties). Was slowly getting some vh games as a result but went on a BAAAAAAAAD matched run. This run of 10 straight losses set me back in normal... I slowly fought back into high again and then valve scrapped the bracket system... I really didn't mind the old ranking once you got in the right bracket, but.... getting there... jeeez... lots of noobs to tolerate and lucky games to go your way...
SUGGESTED IMPROVEMENT: Only solo players should play solo players. I think part of my problem is I get matched against coherent teams who seem to 5 man and rape... Every time I haven't gone mid I've had a terrible experience in lane with retards who just don't know how to play.. even when you tell them "hey please harass"... nope... "Ok if you wont harass please pull creeps" ... nope... "Please fucking do somethign in lane"... *gets agressive and feeds*... SIGH.. Trying to play as a solo is just too hard when you're facing the prospect of parties. To even the playing ground there should be a solo queue in RMM.
|
|
|
|
|
|
EPT
