|
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. |
On January 21 2014 09:44 TrainSamurai wrote:Show nested quote +On January 21 2014 06:46 SeakayKu wrote:On January 21 2014 06:32 sCCrooked wrote:On January 21 2014 06:17 SeakayKu wrote:
4. elo hell exists?
for a 5k player, it takes less than 1 month to get from 3k to 4.5k+, so no
for players who are not good enough to out-class his/her current skill plateau, unfortunately yes
If you really read all of the thread and came away with this as a conclusion, you might want to go re-read or look more thoroughly. Even at 5k there are highly incompetent players and The 5 and even 6k players that get an alt ID that has been secured as "2k-4k" have a hell of a time getting it up to where they know they should be. Hence, the cry of "rating doesn't work well" since even pros can't seem to change the ELOs of accounts the system thinks should stay in the middle. eh? who has that problem? i know at least 1 player who proved that he does not have that problem, which was mention somewhere between pages 85~95 where did you get that player who is 5k+ but could not get alt account up to 4.5k+? of all 112 pages i read, i could not find that He's talking about http://www.playdota.com/forums/showthread.php?t=1398477. AFAIK he's still stuck in shit creek. You can move from 2k to 4k without system dectecting easily just by inhousing a lot for example. anyone can move from 2k easily... say they moved from 2k to 3k using this method. Are we supposed to just not care because that is still in the trenches? Juice hasn't even managed a 1k rise yet (he hasn't even lost a single game)
Bullshit test which doesnt say much. People dont get born with 5k rating and have to work themselves up the ratingsystem. They start out as noobs and get better slowly because the opposition is also better. U cant really test that on a smale scale (or not as simple as this test). I.e. think its logical that usain bolt would win a gold medal on the paralympics too. He is a capt. obvious.
|
So far all Juice has proved is that there's no retroactive smurf detection.
The only real situation in which this is valid is if you have a player who significantly improves from playing off-radar matches. Considering that Dota's MM is fragmented into Solo Ranked, Party Ranked and Unranked in addition to inhouses/scrims, that's not a totally unrealistic situation. A 2k MMR difference is pretty unlikely though.
I'm interested in seeing what happens when he gets to 4.5k and up. I'm betting he will slow right down once he gets much within 1k MMR of his real rating. In fact I would bet that he gives up before he gets there just because of how long it will take when he can't 50-0 every game.
|
On January 21 2014 11:57 govie wrote:Show nested quote +On January 21 2014 09:44 TrainSamurai wrote:On January 21 2014 06:46 SeakayKu wrote:On January 21 2014 06:32 sCCrooked wrote:On January 21 2014 06:17 SeakayKu wrote:
4. elo hell exists?
for a 5k player, it takes less than 1 month to get from 3k to 4.5k+, so no
for players who are not good enough to out-class his/her current skill plateau, unfortunately yes
If you really read all of the thread and came away with this as a conclusion, you might want to go re-read or look more thoroughly. Even at 5k there are highly incompetent players and The 5 and even 6k players that get an alt ID that has been secured as "2k-4k" have a hell of a time getting it up to where they know they should be. Hence, the cry of "rating doesn't work well" since even pros can't seem to change the ELOs of accounts the system thinks should stay in the middle. eh? who has that problem? i know at least 1 player who proved that he does not have that problem, which was mention somewhere between pages 85~95 where did you get that player who is 5k+ but could not get alt account up to 4.5k+? of all 112 pages i read, i could not find that He's talking about http://www.playdota.com/forums/showthread.php?t=1398477. AFAIK he's still stuck in shit creek. You can move from 2k to 4k without system dectecting easily just by inhousing a lot for example. anyone can move from 2k easily... say they moved from 2k to 3k using this method. Are we supposed to just not care because that is still in the trenches? Juice hasn't even managed a 1k rise yet (he hasn't even lost a single game) Bullshit test which doesnt say much. People dont get born with 5k rating and have to work themselves up the ratingsystem. They start out as noobs and get better slowly because the opposition is also better. U cant really test that on a smale scale (or not as simple as this test). I.e. think its logical that usain bolt would win a gold medal on the paralympics too. He is a capt. obvious.
Wait? Logic? Now you can't be serious.
You mean that rating that I've managed to get after at least 150 games of Dota is why they have put that stipulation there and why they made me play those games? You mean that a person is always going to end up wherever it is that they belong? Blasphemy. Clearly I'm a 5k player and just stuck with huge noobs here at 3k.
Seriously it's like people think that just playing the game should make them swing up by 1k points every time they start it up. Actual improvement is slow. That's how you know it's real improvement. You don't suddenly wake up one day twice as good as you were, you put the effort in and eventually you get to your pinnacle.
And you do this by playing. A lot.
|
On January 21 2014 11:57 govie wrote:Show nested quote +On January 21 2014 09:44 TrainSamurai wrote:On January 21 2014 06:46 SeakayKu wrote:On January 21 2014 06:32 sCCrooked wrote:On January 21 2014 06:17 SeakayKu wrote:
4. elo hell exists?
for a 5k player, it takes less than 1 month to get from 3k to 4.5k+, so no
for players who are not good enough to out-class his/her current skill plateau, unfortunately yes
If you really read all of the thread and came away with this as a conclusion, you might want to go re-read or look more thoroughly. Even at 5k there are highly incompetent players and The 5 and even 6k players that get an alt ID that has been secured as "2k-4k" have a hell of a time getting it up to where they know they should be. Hence, the cry of "rating doesn't work well" since even pros can't seem to change the ELOs of accounts the system thinks should stay in the middle. eh? who has that problem? i know at least 1 player who proved that he does not have that problem, which was mention somewhere between pages 85~95 where did you get that player who is 5k+ but could not get alt account up to 4.5k+? of all 112 pages i read, i could not find that He's talking about http://www.playdota.com/forums/showthread.php?t=1398477. AFAIK he's still stuck in shit creek. You can move from 2k to 4k without system dectecting easily just by inhousing a lot for example. anyone can move from 2k easily... say they moved from 2k to 3k using this method. Are we supposed to just not care because that is still in the trenches? Juice hasn't even managed a 1k rise yet (he hasn't even lost a single game) Bullshit test which doesnt say much. People dont get born with 5k rating and have to work themselves up the ratingsystem. They start out as noobs and get better slowly because the opposition is also better. U cant really test that on a smale scale (or not as simple as this test). I.e. think its logical that usain bolt would win a gold medal on the paralympics too. He is a capt. obvious.
First of all this ladder system isn't new, this is pretty much gosugamer's ranking system and we've been bitching about it despite it only applying to pro teams. Even gg has acknowledged thier ladder sucks but ranking teams isn't easy and we just have to live with it.
What Juice has shown that uncertainty was a very poorly implemented. If you have a large pool of 3k players will any outcome be suprising? No of course not.
I think a ladder should place people where they belong, now given that skill=! MMR but I think we can all agree that there is some relationship between the two.
Right now climbing the ladder is just a shit experience, say someone falls 500MMR for whatever stupid reason, insomnia/stress over exams playing dota to relax what ever the reason. Climbing back up would be a pain in the ass. I'm not talking about ELO hell here since I believe the climb is possible. But should the climb be such a pain? Sure he wins on his way up but would he enjoy being in 3k when he is really 3.5k? According to juice there is a visible difference between the 2 levels so I would say the 3.5k player would be raging most of the time.
Proposal:
Fix the implmentation of "uncertainty". Right now it should only change at the two tails of the MMR bell curve.
A lot of us has had many problems with sc2 but no one complained about climbing its ladder. In sc2 if you're on a win streak it progressively puts you up against harder opponents giving you a chance to accelerate your MMR gain, vice versa if you're up too high. In the above example you would make up 500 MMR easily, you would still be unhappy about the games but at least you won't have to put up with it for long.
I think thats a better way of putting people where they "belong". It doesn't change the status quo but it helps people who believe they're in ELO hell and are actually climbing up.
I've been watching juice's game and some of them were only even because he did a calculated dive into 5 people to get some kills. That is something no one in 4k has the guts to do so I doubt you would have the same results as him even if you're higher than his opponents. You can't just be better you gotta be significantly better to climb the solo queue by yourself.
|
You understand that uncertainty works both ways, right?
Uncertainty is totally useless for the purposes of, say, buffering yourself against bad days and loss streaks. While you are on that loss streak, your uncertainty would be increasing as well, so you would drop even further and faster.
By definition, reversing an MMR drop requires you to do the opposite of what got you there. It's a closed system; for every rise, someone else has to fall. There is no way around that, other than inflation systems like SC2's battle points.
Uncertainty is only useful in situations where a player's skill has changed dramatically due to something the system could not track, like bulk inhouse games or account trading. These are rare cases, and the MMR will eventually correct anyway.
|
On January 21 2014 15:18 Belisarius wrote:
You understand that uncertainty works both ways, right?
Uncertainty is totally useless for the purposes of, say, buffering yourself against bad days and loss streaks. While you are on that loss streak, your uncertainty would be increasing as well, so you would drop even further and faster.
By definition, reversing an MMR drop requires you to do the opposite of what got you there. It's a closed system; for every rise, someone else has to fall. There is no way around that, other than inflation systems like SC2's battle points.
Uncertainty is only useful in situations where a player's skill has changed dramatically due to something the system could not track, like bulk inhouse games or account trading. These are rare cases, and the MMR will eventually correct anyway.
Yes but should that really be the point of a ladder system? Shouldn't the ladder instead just quickly get you back to where you belong if you start playing at optimal condition? This isn't about rising your MMR but the speed of getting to where you should be if you decide to start playing serious. We could be saying how you shouldn't be doing dumb things to lower your MMR but thats a strawman really if we create a system that tries to measure skill the only thing that matters is your current skill level. If I was playing dota during a bad week I should be able to quickly get back to where I was b4 that week if that is where I really belong.
I don't really think its fair to say skill doesn't change, when we look at sc2 moving from bronze to diamond was a pretty easy process and they had a system in place that didn't require us to put millions of hours into achieving that and it worked out fine. We didn't have millions of people in diamond who didn't belong( you might consider them bad but they were generally better than bronze). The only difference is that in sc2 you could move quickly.
By definition, reversing an MMR drop requires you to do the opposite of what got you there
No there is a difference, you won't free fall, just because you're not playing optimal doesn't mean you gone full retarded. If under the sc2 system and lose 1k you would have a much easier time getting on a winstreak and getting that bonus MMR. Going up you would face much tougher opponents so its not like it is a one way ticket up.
|
is there a new scale of what percentile you belong?
|
I dont get the complaining here. The only point worth complaining about is the huge variance in one team going for a 2000mmr difference between the worst and the best player in a team.
I didnt play dota for 2 months until the ranking system came out, i started out at 3.4k rating and now i m up to 4.3k rating that is pretty consistant. The only problem i had was arround the 3.5k rating when sometimes the games would completely spiral out of control in the early game were a bad player from one team was feeding the good player of the other team and therefore the good player would wreck the entire game. When i was playing 3.5k rating i sometimes had players with a 2.5k rating and players with 4.5k rating in my team.
When the bad player didnt feed we wouldwin the game, but sometimes it gets impossible when the 2.5k rating player picks a hardcarry and then goes on a solo mission against 5 players at a critical point in the game.
I had that shit happen many many times and we lost won games because of that. But if you are a better player than your mmr then you eventually overcome this and go up the ratings.
You cannot expect to win 20 games in a row. I also had games were we had 2 supports in our team that didnt buy a single ward the whole game and didnt upgrade the courier, and i m not talking about a 15min game i m talking about 40mins no courier and wards.
But this doesnt happen every game and then you can win the games that are fairly equal and then you will go up in the rating. So stop whining and play.
|
Woah, this thread is going on FOREVER.
None of you will enjoy the game until your levels become meaningful, right?
|
On January 21 2014 18:30 makmeatt wrote:
Woah, this thread is going on FOREVER.
None of you will enjoy the game until your levels become meaningful, right?
i think people want a visible ladder to truly compare who has the bigger D
honestly, just play the game, i went on a 12 game losing streak in solo q, but you just keep playing and go for the win.
|
On January 21 2014 08:46 sCCrooked wrote:Show nested quote +On January 21 2014 08:27 Erasme wrote:
Actually i do understand what you're saying and if i can carry 4.5k, I will easily win 99% of my games lower 1v9. Also you have a poor understanding of the matchmaking system. Unless you believe that, overnight, a 2k player can become a 5k player. Since you have roughly 200games to place yourself, if you're really good, you will not end up at 2k. That's not at all what I was saying. It appears you don't understand that I wasn't in any way suggesting that an overnight MMR change of 3k would happen nor would most people interpret it as such. I was not aware the MMR balanced out over 200 games, as I have only played perhaps a quarter of that in total ranked games so far. I guess that design was so that wild swings do not occur often, but all I have been doing since I got placed is steadily rise up. After 50 games, its only +200-300 or so, but I could see how after 150 more, I might have over 1,000 or 2,000 difference made. Perhaps it just FEELS so ridiculously long because 200 games is a lot of time spent playing for the casuals. A 5k player who only plays 15 games a week probably will have to wait a long time to see where they actually are if they got placed at 3.5k for example.
You don't understand. A 5k player can play 1game a week, if they have the level, they will be placed around 5k. The 200games are the unranked games you have to play to reach lvl 13. In your imaginary world where 6k players get placed at 2k for some unknown reasons, the ranked matchmaking needs to change. Luckily, in reality, it doesn't exist.
|
On January 21 2014 13:15 TrainSamurai wrote:Show nested quote +On January 21 2014 11:57 govie wrote:On January 21 2014 09:44 TrainSamurai wrote:On January 21 2014 06:46 SeakayKu wrote:On January 21 2014 06:32 sCCrooked wrote:On January 21 2014 06:17 SeakayKu wrote:
4. elo hell exists?
for a 5k player, it takes less than 1 month to get from 3k to 4.5k+, so no
for players who are not good enough to out-class his/her current skill plateau, unfortunately yes
If you really read all of the thread and came away with this as a conclusion, you might want to go re-read or look more thoroughly. Even at 5k there are highly incompetent players and The 5 and even 6k players that get an alt ID that has been secured as "2k-4k" have a hell of a time getting it up to where they know they should be. Hence, the cry of "rating doesn't work well" since even pros can't seem to change the ELOs of accounts the system thinks should stay in the middle. eh? who has that problem? i know at least 1 player who proved that he does not have that problem, which was mention somewhere between pages 85~95 where did you get that player who is 5k+ but could not get alt account up to 4.5k+? of all 112 pages i read, i could not find that He's talking about http://www.playdota.com/forums/showthread.php?t=1398477. AFAIK he's still stuck in shit creek. You can move from 2k to 4k without system dectecting easily just by inhousing a lot for example. anyone can move from 2k easily... say they moved from 2k to 3k using this method. Are we supposed to just not care because that is still in the trenches? Juice hasn't even managed a 1k rise yet (he hasn't even lost a single game) Bullshit test which doesnt say much. People dont get born with 5k rating and have to work themselves up the ratingsystem. They start out as noobs and get better slowly because the opposition is also better. U cant really test that on a smale scale (or not as simple as this test). I.e. think its logical that usain bolt would win a gold medal on the paralympics too. He is a capt. obvious. First of all this ladder system isn't new, this is pretty much gosugamer's ranking system and we've been bitching about it despite it only applying to pro teams. Even gg has acknowledged thier ladder sucks but ranking teams isn't easy and we just have to live with it. What Juice has shown that uncertainty was a very poorly implemented. If you have a large pool of 3k players will any outcome be suprising? No of course not. I think a ladder should place people where they belong, now given that skill=! MMR but I think we can all agree that there is some relationship between the two. Right now climbing the ladder is just a shit experience, say someone falls 500MMR for whatever stupid reason, insomnia/stress over exams playing dota to relax what ever the reason. Climbing back up would be a pain in the ass. I'm not talking about ELO hell here since I believe the climb is possible. But should the climb be such a pain? Sure he wins on his way up but would he enjoy being in 3k when he is really 3.5k? According to juice there is a visible difference between the 2 levels so I would say the 3.5k player would be raging most of the time.
This thread would actually get somewhere if we would define ELOhell first, as most individuals define it differently. I would define ELOhell like you are playing atleast slightly (not much, just slightly) better then all the other players, but your positive gameimpact will allways be less, then the negative game impact from one of the other players.
To a certain degree this definition could make sence and elohell could exist. Your positive game impact would allmost have no influence to the outcome of a game and your personal mmr. The only way to get out of this ELOhell would a the road of an unlimited amount of games to make "the negative game impact players turn into insignificant static in the data". Good luck with finding the time to play so many games before you die of a stroke by pure stress and agony alone, haha It would basically mean that most players will be stuck after valve gives you your mmr range.
Because valve mmr system doesnt start of with alot of indidivual data, this definition of ELOhell could seem possible ( i think). If that capt. obvious could test this instead of that 5k player in 2k range, it would seem more meaningfull.
|
On January 21 2014 19:47 govie wrote:Show nested quote +On January 21 2014 13:15 TrainSamurai wrote:On January 21 2014 11:57 govie wrote:On January 21 2014 09:44 TrainSamurai wrote:On January 21 2014 06:46 SeakayKu wrote:On January 21 2014 06:32 sCCrooked wrote:On January 21 2014 06:17 SeakayKu wrote:
4. elo hell exists?
for a 5k player, it takes less than 1 month to get from 3k to 4.5k+, so no
for players who are not good enough to out-class his/her current skill plateau, unfortunately yes
If you really read all of the thread and came away with this as a conclusion, you might want to go re-read or look more thoroughly. Even at 5k there are highly incompetent players and The 5 and even 6k players that get an alt ID that has been secured as "2k-4k" have a hell of a time getting it up to where they know they should be. Hence, the cry of "rating doesn't work well" since even pros can't seem to change the ELOs of accounts the system thinks should stay in the middle. eh? who has that problem? i know at least 1 player who proved that he does not have that problem, which was mention somewhere between pages 85~95 where did you get that player who is 5k+ but could not get alt account up to 4.5k+? of all 112 pages i read, i could not find that He's talking about http://www.playdota.com/forums/showthread.php?t=1398477. AFAIK he's still stuck in shit creek. You can move from 2k to 4k without system dectecting easily just by inhousing a lot for example. anyone can move from 2k easily... say they moved from 2k to 3k using this method. Are we supposed to just not care because that is still in the trenches? Juice hasn't even managed a 1k rise yet (he hasn't even lost a single game) Bullshit test which doesnt say much. People dont get born with 5k rating and have to work themselves up the ratingsystem. They start out as noobs and get better slowly because the opposition is also better. U cant really test that on a smale scale (or not as simple as this test). I.e. think its logical that usain bolt would win a gold medal on the paralympics too. He is a capt. obvious. First of all this ladder system isn't new, this is pretty much gosugamer's ranking system and we've been bitching about it despite it only applying to pro teams. Even gg has acknowledged thier ladder sucks but ranking teams isn't easy and we just have to live with it. What Juice has shown that uncertainty was a very poorly implemented. If you have a large pool of 3k players will any outcome be suprising? No of course not. I think a ladder should place people where they belong, now given that skill=! MMR but I think we can all agree that there is some relationship between the two. Right now climbing the ladder is just a shit experience, say someone falls 500MMR for whatever stupid reason, insomnia/stress over exams playing dota to relax what ever the reason. Climbing back up would be a pain in the ass. I'm not talking about ELO hell here since I believe the climb is possible. But should the climb be such a pain? Sure he wins on his way up but would he enjoy being in 3k when he is really 3.5k? According to juice there is a visible difference between the 2 levels so I would say the 3.5k player would be raging most of the time. This thread would actually get somewhere if we would define ELOhell first, as most individuals define it differently. I would define ELOhell like you are playing atleast slightly (not much, just slightly) better then all the other players, but your positive gameimpact will allways be less, then the negative game impact from one of the other players. To a certain degree this definition could make sence and elohell could exist. Your positive game impact would allmost have no influence to the outcome of a game and your personal mmr. The only way to get out of this ELOhell would a the road of an unlimited amount of games to make "the negative game impact players turn into insignificant static in the data". Good luck with finding the time to play so many games before you die of a stroke by pure stress and agony alone, haha It would basically mean that most players will be stuck after valve gives you your mmr range. Because valve mmr system doesnt start of with alot of indidivual data, this definition of ELOhell could seem possible ( i think). If that capt. obvious could test this instead of that 5k player in 2k range, it would seem more meaningfull.
That's not the definition of Elo Hell. Anyone that complains about Elo Hell bitches about their teammates being terrible, not "slightly worse". The whole concept of all players being so bad the game ends up as a coin toss, since your contribution ends up being meanings when your teammates always end up thowing the game just doesn't work in the definition you describe. That's why a 5k grabbing a 2k account makes sense. Some people even say not even a pro could easily leave Elo Hell.
This whole slightly better thing seems like a compromise when people realise how Elo Hell makes no sense, so you would grab at a similiar concept, that still allows you to believe you must be better than your teammates, but isn't as easy to refute.
Just google "What is Elo Hell" or something like that. You won't find anything similar to your definition.
|
On January 21 2014 21:12 SKC wrote:Show nested quote +On January 21 2014 19:47 govie wrote:On January 21 2014 13:15 TrainSamurai wrote:On January 21 2014 11:57 govie wrote:On January 21 2014 09:44 TrainSamurai wrote:On January 21 2014 06:46 SeakayKu wrote:On January 21 2014 06:32 sCCrooked wrote:On January 21 2014 06:17 SeakayKu wrote:
4. elo hell exists?
for a 5k player, it takes less than 1 month to get from 3k to 4.5k+, so no
for players who are not good enough to out-class his/her current skill plateau, unfortunately yes
If you really read all of the thread and came away with this as a conclusion, you might want to go re-read or look more thoroughly. Even at 5k there are highly incompetent players and The 5 and even 6k players that get an alt ID that has been secured as "2k-4k" have a hell of a time getting it up to where they know they should be. Hence, the cry of "rating doesn't work well" since even pros can't seem to change the ELOs of accounts the system thinks should stay in the middle. eh? who has that problem? i know at least 1 player who proved that he does not have that problem, which was mention somewhere between pages 85~95 where did you get that player who is 5k+ but could not get alt account up to 4.5k+? of all 112 pages i read, i could not find that He's talking about http://www.playdota.com/forums/showthread.php?t=1398477. AFAIK he's still stuck in shit creek. You can move from 2k to 4k without system dectecting easily just by inhousing a lot for example. anyone can move from 2k easily... say they moved from 2k to 3k using this method. Are we supposed to just not care because that is still in the trenches? Juice hasn't even managed a 1k rise yet (he hasn't even lost a single game) Bullshit test which doesnt say much. People dont get born with 5k rating and have to work themselves up the ratingsystem. They start out as noobs and get better slowly because the opposition is also better. U cant really test that on a smale scale (or not as simple as this test). I.e. think its logical that usain bolt would win a gold medal on the paralympics too. He is a capt. obvious. First of all this ladder system isn't new, this is pretty much gosugamer's ranking system and we've been bitching about it despite it only applying to pro teams. Even gg has acknowledged thier ladder sucks but ranking teams isn't easy and we just have to live with it. What Juice has shown that uncertainty was a very poorly implemented. If you have a large pool of 3k players will any outcome be suprising? No of course not. I think a ladder should place people where they belong, now given that skill=! MMR but I think we can all agree that there is some relationship between the two. Right now climbing the ladder is just a shit experience, say someone falls 500MMR for whatever stupid reason, insomnia/stress over exams playing dota to relax what ever the reason. Climbing back up would be a pain in the ass. I'm not talking about ELO hell here since I believe the climb is possible. But should the climb be such a pain? Sure he wins on his way up but would he enjoy being in 3k when he is really 3.5k? According to juice there is a visible difference between the 2 levels so I would say the 3.5k player would be raging most of the time. This thread would actually get somewhere if we would define ELOhell first, as most individuals define it differently. I would define ELOhell like you are playing atleast slightly (not much, just slightly) better then all the other players, but your positive gameimpact will allways be less, then the negative game impact from one of the other players. To a certain degree this definition could make sence and elohell could exist. Your positive game impact would allmost have no influence to the outcome of a game and your personal mmr. The only way to get out of this ELOhell would a the road of an unlimited amount of games to make "the negative game impact players turn into insignificant static in the data". Good luck with finding the time to play so many games before you die of a stroke by pure stress and agony alone, haha It would basically mean that most players will be stuck after valve gives you your mmr range. Because valve mmr system doesnt start of with alot of indidivual data, this definition of ELOhell could seem possible ( i think). If that capt. obvious could test this instead of that 5k player in 2k range, it would seem more meaningfull. That's not the definition of Elo Hell. Anyone that complains about Elo Hell bitches about their teammates being terrible, not "slightly worse". The whole concept of all players being so bad the game ends up as a coin toss, since your contribution ends up being meanings when your teammates always end up thowing the game just doesn't work in the definition you describe.
This is exactly what i wrote..... , your positive impact would not alter the outcome of most games. Therefore a change in mmr gained by gametime would be minimal even when your playing better then most players in your range. Only thing is that 5k player trapped in 2k range isnt a normal occurence. Dont argue just because u like it
|
On January 21 2014 22:10 govie wrote:Show nested quote +On January 21 2014 21:12 SKC wrote:On January 21 2014 19:47 govie wrote:On January 21 2014 13:15 TrainSamurai wrote:On January 21 2014 11:57 govie wrote:On January 21 2014 09:44 TrainSamurai wrote:On January 21 2014 06:46 SeakayKu wrote:On January 21 2014 06:32 sCCrooked wrote:On January 21 2014 06:17 SeakayKu wrote:
4. elo hell exists?
for a 5k player, it takes less than 1 month to get from 3k to 4.5k+, so no
for players who are not good enough to out-class his/her current skill plateau, unfortunately yes
If you really read all of the thread and came away with this as a conclusion, you might want to go re-read or look more thoroughly. Even at 5k there are highly incompetent players and The 5 and even 6k players that get an alt ID that has been secured as "2k-4k" have a hell of a time getting it up to where they know they should be. Hence, the cry of "rating doesn't work well" since even pros can't seem to change the ELOs of accounts the system thinks should stay in the middle. eh? who has that problem? i know at least 1 player who proved that he does not have that problem, which was mention somewhere between pages 85~95 where did you get that player who is 5k+ but could not get alt account up to 4.5k+? of all 112 pages i read, i could not find that He's talking about http://www.playdota.com/forums/showthread.php?t=1398477. AFAIK he's still stuck in shit creek. You can move from 2k to 4k without system dectecting easily just by inhousing a lot for example. anyone can move from 2k easily... say they moved from 2k to 3k using this method. Are we supposed to just not care because that is still in the trenches? Juice hasn't even managed a 1k rise yet (he hasn't even lost a single game) Bullshit test which doesnt say much. People dont get born with 5k rating and have to work themselves up the ratingsystem. They start out as noobs and get better slowly because the opposition is also better. U cant really test that on a smale scale (or not as simple as this test). I.e. think its logical that usain bolt would win a gold medal on the paralympics too. He is a capt. obvious. First of all this ladder system isn't new, this is pretty much gosugamer's ranking system and we've been bitching about it despite it only applying to pro teams. Even gg has acknowledged thier ladder sucks but ranking teams isn't easy and we just have to live with it. What Juice has shown that uncertainty was a very poorly implemented. If you have a large pool of 3k players will any outcome be suprising? No of course not. I think a ladder should place people where they belong, now given that skill=! MMR but I think we can all agree that there is some relationship between the two. Right now climbing the ladder is just a shit experience, say someone falls 500MMR for whatever stupid reason, insomnia/stress over exams playing dota to relax what ever the reason. Climbing back up would be a pain in the ass. I'm not talking about ELO hell here since I believe the climb is possible. But should the climb be such a pain? Sure he wins on his way up but would he enjoy being in 3k when he is really 3.5k? According to juice there is a visible difference between the 2 levels so I would say the 3.5k player would be raging most of the time. This thread would actually get somewhere if we would define ELOhell first, as most individuals define it differently. I would define ELOhell like you are playing atleast slightly (not much, just slightly) better then all the other players, but your positive gameimpact will allways be less, then the negative game impact from one of the other players. To a certain degree this definition could make sence and elohell could exist. Your positive game impact would allmost have no influence to the outcome of a game and your personal mmr. The only way to get out of this ELOhell would a the road of an unlimited amount of games to make "the negative game impact players turn into insignificant static in the data". Good luck with finding the time to play so many games before you die of a stroke by pure stress and agony alone, haha It would basically mean that most players will be stuck after valve gives you your mmr range. Because valve mmr system doesnt start of with alot of indidivual data, this definition of ELOhell could seem possible ( i think). If that capt. obvious could test this instead of that 5k player in 2k range, it would seem more meaningfull. That's not the definition of Elo Hell. Anyone that complains about Elo Hell bitches about their teammates being terrible, not "slightly worse". The whole concept of all players being so bad the game ends up as a coin toss, since your contribution ends up being meanings when your teammates always end up thowing the game just doesn't work in the definition you describe. This is exactly what i wrote..... , your positive impact would not alter the outcome of most games. Therefore a change in mmr gained by gametime would be minimal even when your playing better then most players in your range. Only thing is that 5k player trapped in 2k range isnt a normal occurence. Dont argue just because u like it
That's not exactly what you wrote. There is a pretty big diference between terrible and slightly worse teammates. And the concept that your team is so bad that no matter what you do you will lose. It's completelly diferent from you being so slightly better than your team that the diference takes too long to adjust.
|
On January 21 2014 15:35 TrainSamurai wrote:Show nested quote +On January 21 2014 15:18 Belisarius wrote:
You understand that uncertainty works both ways, right?
Uncertainty is totally useless for the purposes of, say, buffering yourself against bad days and loss streaks. While you are on that loss streak, your uncertainty would be increasing as well, so you would drop even further and faster.
By definition, reversing an MMR drop requires you to do the opposite of what got you there. It's a closed system; for every rise, someone else has to fall. There is no way around that, other than inflation systems like SC2's battle points.
Uncertainty is only useful in situations where a player's skill has changed dramatically due to something the system could not track, like bulk inhouse games or account trading. These are rare cases, and the MMR will eventually correct anyway. Yes but should that really be the point of a ladder system? Shouldn't the ladder instead just quickly get you back to where you belong if you start playing at optimal condition? This isn't about rising your MMR but the speed of getting to where you should be if you decide to start playing serious. We could be saying how you shouldn't be doing dumb things to lower your MMR but thats a strawman really if we create a system that tries to measure skill the only thing that matters is your current skill level. If I was playing dota during a bad week I should be able to quickly get back to where I was b4 that week if that is where I really belong. I don't really think its fair to say skill doesn't change, when we look at sc2 moving from bronze to diamond was a pretty easy process and they had a system in place that didn't require us to put millions of hours into achieving that and it worked out fine. We didn't have millions of people in diamond who didn't belong( you might consider them bad but they were generally better than bronze). The only difference is that in sc2 you could move quickly. Show nested quote +By definition, reversing an MMR drop requires you to do the opposite of what got you there No there is a difference, you won't free fall, just because you're not playing optimal doesn't mean you gone full retarded. If under the sc2 system and lose 1k you would have a much easier time getting on a winstreak and getting that bonus MMR. Going up you would face much tougher opponents so its not like it is a one way ticket up.
Sry to say but using "strawman" because you read this on reddit doesnt make your statement any better.
You realize that you have to play 150games to acutally start the calibration right? That means all that 150 games will be taken into account once you start the calibration, which means that this isnt just about 10games it is about 160games at least.
Now if you refer to the sc2 system, take a bronze account that made 150games in bronze and then try laddering it up to diamond. It takes 50-100games to ladder this account into diamond, 50-100 straight wins. I did this once because i was curious how long it would take to get a account stuck in bronze to diamond.
There is a huge difference between a new sc2 account and an account that already made a shitload of games and is stuck in a league. So you are comparing two things that cannot be compared.
|
On January 21 2014 22:19 SKC wrote:Show nested quote +On January 21 2014 22:10 govie wrote:On January 21 2014 21:12 SKC wrote:On January 21 2014 19:47 govie wrote:On January 21 2014 13:15 TrainSamurai wrote:On January 21 2014 11:57 govie wrote:On January 21 2014 09:44 TrainSamurai wrote:On January 21 2014 06:46 SeakayKu wrote:On January 21 2014 06:32 sCCrooked wrote:On January 21 2014 06:17 SeakayKu wrote:
4. elo hell exists?
for a 5k player, it takes less than 1 month to get from 3k to 4.5k+, so no
for players who are not good enough to out-class his/her current skill plateau, unfortunately yes
If you really read all of the thread and came away with this as a conclusion, you might want to go re-read or look more thoroughly. Even at 5k there are highly incompetent players and The 5 and even 6k players that get an alt ID that has been secured as "2k-4k" have a hell of a time getting it up to where they know they should be. Hence, the cry of "rating doesn't work well" since even pros can't seem to change the ELOs of accounts the system thinks should stay in the middle. eh? who has that problem? i know at least 1 player who proved that he does not have that problem, which was mention somewhere between pages 85~95 where did you get that player who is 5k+ but could not get alt account up to 4.5k+? of all 112 pages i read, i could not find that He's talking about http://www.playdota.com/forums/showthread.php?t=1398477. AFAIK he's still stuck in shit creek. You can move from 2k to 4k without system dectecting easily just by inhousing a lot for example. anyone can move from 2k easily... say they moved from 2k to 3k using this method. Are we supposed to just not care because that is still in the trenches? Juice hasn't even managed a 1k rise yet (he hasn't even lost a single game) Bullshit test which doesnt say much. People dont get born with 5k rating and have to work themselves up the ratingsystem. They start out as noobs and get better slowly because the opposition is also better. U cant really test that on a smale scale (or not as simple as this test). I.e. think its logical that usain bolt would win a gold medal on the paralympics too. He is a capt. obvious. First of all this ladder system isn't new, this is pretty much gosugamer's ranking system and we've been bitching about it despite it only applying to pro teams. Even gg has acknowledged thier ladder sucks but ranking teams isn't easy and we just have to live with it. What Juice has shown that uncertainty was a very poorly implemented. If you have a large pool of 3k players will any outcome be suprising? No of course not. I think a ladder should place people where they belong, now given that skill=! MMR but I think we can all agree that there is some relationship between the two. Right now climbing the ladder is just a shit experience, say someone falls 500MMR for whatever stupid reason, insomnia/stress over exams playing dota to relax what ever the reason. Climbing back up would be a pain in the ass. I'm not talking about ELO hell here since I believe the climb is possible. But should the climb be such a pain? Sure he wins on his way up but would he enjoy being in 3k when he is really 3.5k? According to juice there is a visible difference between the 2 levels so I would say the 3.5k player would be raging most of the time. This thread would actually get somewhere if we would define ELOhell first, as most individuals define it differently. I would define ELOhell like you are playing atleast slightly (not much, just slightly) better then all the other players, but your positive gameimpact will allways be less, then the negative game impact from one of the other players. To a certain degree this definition could make sence and elohell could exist. Your positive game impact would allmost have no influence to the outcome of a game and your personal mmr. The only way to get out of this ELOhell would a the road of an unlimited amount of games to make "the negative game impact players turn into insignificant static in the data". Good luck with finding the time to play so many games before you die of a stroke by pure stress and agony alone, haha It would basically mean that most players will be stuck after valve gives you your mmr range. Because valve mmr system doesnt start of with alot of indidivual data, this definition of ELOhell could seem possible ( i think). If that capt. obvious could test this instead of that 5k player in 2k range, it would seem more meaningfull. That's not the definition of Elo Hell. Anyone that complains about Elo Hell bitches about their teammates being terrible, not "slightly worse". The whole concept of all players being so bad the game ends up as a coin toss, since your contribution ends up being meanings when your teammates always end up thowing the game just doesn't work in the definition you describe. This is exactly what i wrote..... , your positive impact would not alter the outcome of most games. Therefore a change in mmr gained by gametime would be minimal even when your playing better then most players in your range. Only thing is that 5k player trapped in 2k range isnt a normal occurence. Dont argue just because u like it That's not exactly what you wrote. There is a pretty big diference between terrible and slightly worse teammates. And the concept that your team is so bad that no matter what you do you will lose. It's completelly diferent from you being so slightly better than your team that the diference takes too long to adjust.
My point is that its highly unlikely to get 5k player in a 2k range, ever. He is testing a situation that will never occur therefore his test will not prove a thing. I can test that ducks can dive 50meter below the surface, would these results have any meaning as ducks never dive to 50m below the surface anyway
|
hm... another summary of this thread from a different perspective
2k
didn't hear much from this group...
i suppose these are the happy casual players and small portion of them dont even check forums
they just click and play, who has time to read nowadays?
3k
complaining about bad players who do not know how to play the game ruins a match
the observation so far is that no 3k individual can carry a game over that negative-carry-bad player
given 50% chance that the worst player could be on any team
being in 3k makes you stay in 3k unless you are able to carry a game over that negative-carry player
4k
players here seem to have at least balance the negative impact of that bad player in 3k group
a 4k player can get out of 3k group, proved in many instances in this thread
many 4k players on a bad day will fall back to 3.5k because anyone can easily be on 8 game losing streak
edit#
when that happens, if the 4k player can dig themselves out of 3k group, 4k player will rebound after losing streak
otherwise, 4k player will hover around 3k group or stick around 3.5k if player cannot consistently delivery 4k quality
5k
i do not see many players here argue or complain
6k+
win = small mmr gain; lose = huge mmr loss
well, my thoughts are the following:
1. the purpose of valve should be trying to get everyone to play as many games as possible
if 6k+ are not playing because huge mmr loss, then valve should make a very heavy mmr decay on 6k+ players
not playing = huge mmr loss anyway, playing to maintain huge mmr makes much more sense
high mmr players constantly playing ranked games is good for dota 2 scene anyway
2. there is no change needed to accommodate 3k players frustration
it is a simple indication that 4k players can dig themselves out of 3k situation
and if a player cannot dig themselves out of 3k situation, then that player should stay in 3k
the only way getting out of 3k is to play at 4k quality, not at 4k quantity
3. due to the urgency of getting out of 3k, players tend to pick high influencing heros
i think valve should implement some mmr +/- adjustment factor depending on the hero a player picks
for example
crystal maiden, venge spirit hero picks receives 5 mmr bonus regardless win/loss
hard carries such as antimage pick recieves no bonus for winning, but additional penalty of 2 mmr for losing
|
Where do I see how many ranked games me or my friend have played?
|
Juice lost his first game at 3.6k :O
I predicted 4k, there goes my rare
|
|
|
|
|
|
EPT
