^
yeah right, i play solo queue all day yet my rating is lower than a normal bracket player GG WP. Btw, what is that prediction about, how the hell can we use that?
'-DSR rating can now predict the winning team with 65% chance of success' <- talking about this
Don't feel bad Blunt. Look at my last post on P6. I'm a 290 which says I'd be the bottom of normal. But like 97.5% of my games are above normal. Go figure.
You may start a new account if you want accurate reading and still play with your friend on your normal account.
@ blunt
With dsr you can guess the outcome of the match, before it starts by calculating each team average DSR. Higher DSR teams will win 65% of the time. With bullshit algo this would be 50% ( random chance ).
% is only relative to the people that have put their ID in the DSR system (I think), which as you can imagine, is going to include significantly better players on average than the DBR system which had everyone. 0-400 or something is going to be ~70% of the total player base, but no where near that for DSR.
^ Yep, I am working on visualizing the DSR rating for all of the playerbase. I think people will be surprised how about the value of DSR is for the average player.
I notice that players with less than 500 matches played get over 1000 rating ?
players with less than 500 matches should not get more than 1000 rating unless he/she have over 60% win rate
but you already use that game to calculate the dsr so the winning team gets a dsr boost and the losing team loses dsr
so the chance is not 50% to begin with anyway
So if Dendi made a new account and started playing on it, I shouldn't give him more than 1000 points unless he has more than 500 games? What purpose would that serve?
so any player could just make a smurf account to get over 1000 rating then ?
and i did said that if dendi make a smurf and get over 60 % - 70 % win rate in less than 500 matches then he deserved it.
but less than 500 matches played 55 % win rate..I think most people can do that
@MyDotaSkill.com #650 DSR:
I think you have a hard time reading. As i said, 98% solo queuing, still the rating is shit, also Fuzzy wuzzy 900DSR, seems legit.
@mydotaskill, dude, don't even bother with these retards.
If dendi made a new account with 100 matches and it had a low rating, they would tell you your system is shit. If it had a high rating, they would complain about smurfing or something. I'm surprised you are still responding
NOT ANY player can get over 1000 rating by making a new acc
only the good ones that would acquire that anyway.
stacking and getting it with a new acc would also tell how much of a tryharder u r.
try making it in the first 10 games solo queing! that way u should get an appropiate rating (which i bet would resemble the shitty onne u already got)
btw i got 200 lower rating now but it looks like all the top tier also lost the same so its ok xD
My intentional abandon low priority account is rated higher than the account i am posting on by about 100 points. That low priority account has a 43% win rate, vs this account which has a 56% win rate. http://dotabuff.com/players/118502741
Do you think that is dunning kruger?
I am all for a great skill rating system, but I'm not convinced by this one yet. It seems like a decent indicator, but nothing to take too seriously. A skill system should try to reflect where you are ranked on valve's system as closely as possible.
@Concede, those systems are really a thing of the past. Newer, more dynamic systems like these are a lot better. If anyone played HoN, and it was their first game of that type, you might drop down quite a lot early on, and even when you become a lot better, it is quite difficult to move up. More fluid rating systems like these allow you to quickly move up and down to where you should be.
@myskill, does game duration play a factor? Some games are over in 10 minutes, where XPM/GPM, and total K/D/A (not ratio) and hero damage are likely to be very low. Or are these numbers looked at relative to other heroes/team totals? I don't even know if they are variables in your system, but I'm guessing something similar to those are.
The only thing that matters in a dota game is who wins and who loses, at least that is how the valve metrics seem to handle it. Now, winning vs better players is what increases MMR/ELO, so of course once you are highly rated it will balance out to 50%.
100 points may not be a big difference, but it doesn't change the fact that your system rated my account that i intentionally abandoned at least 10 games on in the past 20 games, as well as played in a very silly unskilled manner on, higher than an account where I am, relative to that other account, playing significantly more serious on.
I don't care what my rating is, i'm just saying that its clearly needs improvements.
@ mydotaskill Pretty sure valve doesn't share that information with anyone, or you website wouldn't be needed in the first place :)
@親和性 ' FameGG.com: I'm not trying to intentionally fool a system. I play an LPQ account when i know i dont have time for a full game, or i don't want to care about the outcome of the game.
Why you think you need to white knight for mydotaskill is beyond me. Unless you are just attempting to troll and be an elitist because it makes you feel good. In that case do continue.
This would be so much easier if it was like HON where MMR is public. Honestly, it's too hard to quantify skill into one number which is why imo kda, gpm, xpm, etc, shouldn't matter when it comes to a rating. Just have people compete in matches at different levels. If they can survive in a certain match level, great. If they don't belong, their rating drops, if they are better their rating rises.
Should just matter on win/loss based on the level of competence you're playing with/against.
"It seems like a decent indicator, but nothing to take too seriously."
That its the problem, people IS going to use this to indicate who is better than who, when is clear that lot of really good players are going to have low dsr, but they WILL still be better than those with high dsr. In other words, the ones who actually care with this numbers are going to play solo and all pick and fucking tryhard as shit, so they can say that they are actually the best and that there are numbers suporting them. Simple.
Meanwhile really good players like singsing, dendi, etc .... arent going to give a shit about this and probably they will end up with low dsr.
Dont get me wrong, i am not butthurt or anything, i know i am really bad and i am not going to play solo for 20 match only for this, but in so little time that this is created assholes like benao and terrible appear only to support something that they dont understand, you know why? because their rating is high, so they fell better about themselves.
Personally i think if you are a really good player you will get notice, you dont need simple numbers generate with little to non-information to support you.
Anyway i have to say that the guy behind this system is actually trying, the old one there was a guy who coudnt take one fucking negative critic and he just gave up, so credits to the creator of mydotaskill to keep trying to improve this.
"This is the reason why Valve doesn't like ladders.
Bad players get extremely butthurt, dunning kruger all over the place."
Not only that, good players get extremely arrogant too.
@Dorkly: Its not a "problem" if people take DSR seriously. People can assign value to w/e the hell they like. Yes it would be better if valve just told us our ranking but in that absence sites like this will pop up and players will assign importance to them.
People will discount you, ignore you, or insult you no matter what they decide to use as evidence.
People are going to be assholes in dota (or nice) regardless of how much or little information is available. Thats because people are assholes.
Well mydotaskill.com is officially bonkers. I went from Silver to Bronze and now to gold. But I'm terrible, as I sometimes I don't have the reaction time or hero pool that makes me any good.
Like the op said, it is not accurate AT ALL if you stack with players who aren't on your skill level. In order for it to be accurate, you need to stack with players around your skill level so you face off against other players in your bracket. It then probably takes the average DSR of each team and does some calculations to determine your rating.
Even if it's only a 20 game sample, that is more than ample to determine someone's rating. Even if you go on a huge losing streak, you will probably still stay in your skill bracket and your DSR will obviously drop, but not by much. The only thing wrong with DSR is when you stack with players who aren't on your skill level. It will obviously give you an incorrect DSR if you are actually a higher level player.
You need to clearly state on your website that DSR will not be accurate if you stack with lower level players. If you can find some way to fix this, I feel like it'll be pretty accurate then. However, based on the information you guys have, it's a pretty decent job. Not bad at all... Keep up the good work.
i would totally agree with u if that was the case
but it so happens that i am rly skilled so... next time check the skill of whoever u target before naming him
@vaelditha or smth xd dont quit :p
you're getting better... if u were a carry instead of support from the start u would be owning guys! so transition to a pretty carry player :p try drow/potm/luna
Meh I probably should stop stacking with my friends then
^
yeah right, i play solo queue all day yet my rating is lower than a normal bracket player GG WP. Btw, what is that prediction about, how the hell can we use that?
'-DSR rating can now predict the winning team with 65% chance of success' <- talking about this
it is bugged, mine says 1570..
Im much better than that..
^
lol you're garbage. gtfo of here
Don't feel bad Blunt. Look at my last post on P6. I'm a 290 which says I'd be the bottom of normal. But like 97.5% of my games are above normal. Go figure.
^
i saw that, and i lold hard.
You may start a new account if you want accurate reading and still play with your friend on your normal account.
@ blunt
With dsr you can guess the outcome of the match, before it starts by calculating each team average DSR. Higher DSR teams will win 65% of the time. With bullshit algo this would be 50% ( random chance ).
% is only relative to the people that have put their ID in the DSR system (I think), which as you can imagine, is going to include significantly better players on average than the DBR system which had everyone. 0-400 or something is going to be ~70% of the total player base, but no where near that for DSR.
^ Yep, I am working on visualizing the DSR rating for all of the playerbase. I think people will be surprised how about the value of DSR is for the average player.
@MyDotaSkill.com #650 DSR
I notice that players with less than 500 matches played get over 1000 rating ?
players with less than 500 matches should not get more than 1000 rating unless he/she have over 60% win rate
Why does Blunt think that he would deserve higher rating, blows my mind
but you already use that game to calculate the dsr so the winning team gets a dsr boost and the losing team loses dsr
so the chance is not 50% to begin with anyway
@Lavexis
So if Dendi made a new account and started playing on it, I shouldn't give him more than 1000 points unless he has more than 500 games? What purpose would that serve?
so any player could just make a smurf account to get over 1000 rating then ?
and i did said that if dendi make a smurf and get over 60 % - 70 % win rate in less than 500 matches then he deserved it.
but less than 500 matches played 55 % win rate..I think most people can do that
@MyDotaSkill.com #650 DSR:
I think you have a hard time reading. As i said, 98% solo queuing, still the rating is shit, also Fuzzy wuzzy 900DSR, seems legit.
http://mydotaskill.com/index.php?id=http%3A%2F%2Fdotabuff.com%2Fplayers%2F59625555
I guess 950DSR -> Page 1 player. Good to know!
i went from 118 to 771. I have no idea what that means.
Doesn't show what the ratings mean anymore. "platinum" is just a word with out a percentage.
The idea of starting a new account to try to make this algorithm work is kinda stupid.
@mydotaskill, dude, don't even bother with these retards.
If dendi made a new account with 100 matches and it had a low rating, they would tell you your system is shit. If it had a high rating, they would complain about smurfing or something. I'm surprised you are still responding
@Lavexis,
NOT ANY player can get over 1000 rating by making a new acc
only the good ones that would acquire that anyway.
stacking and getting it with a new acc would also tell how much of a tryharder u r.
try making it in the first 10 games solo queing! that way u should get an appropiate rating (which i bet would resemble the shitty onne u already got)
btw i got 200 lower rating now but it looks like all the top tier also lost the same so its ok xD
oh wait..... this only accounts for the last 20 games? Well that seems completely pointless...
I'm all for a functional MMR/ Skill rating or w/e the community can cook up, but this should be taken with a grain of salt.
This is the reason why Valve doesn't like ladders.
Bad players get extremely butthurt, dunning kruger all over the place.
http://en.wikipedia.org/wiki/Dunning%E2%80%93Kruger_effect
^
Sorta. Some of it is just flaws in the system.
My intentional abandon low priority account is rated higher than the account i am posting on by about 100 points. That low priority account has a 43% win rate, vs this account which has a 56% win rate. http://dotabuff.com/players/118502741
Do you think that is dunning kruger?
I am all for a great skill rating system, but I'm not convinced by this one yet. It seems like a decent indicator, but nothing to take too seriously. A skill system should try to reflect where you are ranked on valve's system as closely as possible.
Winrate has nothing to do with skill. It's a noisy metric and shouldn't be used.
Your rating can change on average by 20 points per game. So 100 points ain't a big difference.
PS: If you check average fresh DSR's between two teams matchmaked by Valve, they are often times 20 points away from each other.
Some real examples from my Excel right now:
match #1:
radiant:1523
dire:1570
match #2:
radiant:1487
dire:1461
DSR seems to be converging towards Valve's internal rating. With a few more signals added, it should become even more accurate.
When you create a smurf your MMR skyrockets, if you play more bad games your rating will go down to where it belongs.
Winrate doesn't really mean anything.
-edit-
You beat me to it. x_x
@Concede, those systems are really a thing of the past. Newer, more dynamic systems like these are a lot better. If anyone played HoN, and it was their first game of that type, you might drop down quite a lot early on, and even when you become a lot better, it is quite difficult to move up. More fluid rating systems like these allow you to quickly move up and down to where you should be.
@myskill, does game duration play a factor? Some games are over in 10 minutes, where XPM/GPM, and total K/D/A (not ratio) and hero damage are likely to be very low. Or are these numbers looked at relative to other heroes/team totals? I don't even know if they are variables in your system, but I'm guessing something similar to those are.
winrate has nothing to do with skill?
The only thing that matters in a dota game is who wins and who loses, at least that is how the valve metrics seem to handle it. Now, winning vs better players is what increases MMR/ELO, so of course once you are highly rated it will balance out to 50%.
100 points may not be a big difference, but it doesn't change the fact that your system rated my account that i intentionally abandoned at least 10 games on in the past 20 games, as well as played in a very silly unskilled manner on, higher than an account where I am, relative to that other account, playing significantly more serious on.
I don't care what my rating is, i'm just saying that its clearly needs improvements.
^ I don't think you understand how much your Valve's Elo rating fluctuates each game.
> Abandones 10 games to intentionally fool the system.
> Complains that system doesn't work properly
My sides.
@ mydotaskill Pretty sure valve doesn't share that information with anyone, or you website wouldn't be needed in the first place :)
@親和性 ' FameGG.com: I'm not trying to intentionally fool a system. I play an LPQ account when i know i dont have time for a full game, or i don't want to care about the outcome of the game.
Why you think you need to white knight for mydotaskill is beyond me. Unless you are just attempting to troll and be an elitist because it makes you feel good. In that case do continue.
1687 wow i didn't expect that...
This would be so much easier if it was like HON where MMR is public. Honestly, it's too hard to quantify skill into one number which is why imo kda, gpm, xpm, etc, shouldn't matter when it comes to a rating. Just have people compete in matches at different levels. If they can survive in a certain match level, great. If they don't belong, their rating drops, if they are better their rating rises.
Should just matter on win/loss based on the level of competence you're playing with/against.
"It seems like a decent indicator, but nothing to take too seriously."
That its the problem, people IS going to use this to indicate who is better than who, when is clear that lot of really good players are going to have low dsr, but they WILL still be better than those with high dsr. In other words, the ones who actually care with this numbers are going to play solo and all pick and fucking tryhard as shit, so they can say that they are actually the best and that there are numbers suporting them. Simple.
Meanwhile really good players like singsing, dendi, etc .... arent going to give a shit about this and probably they will end up with low dsr.
Dont get me wrong, i am not butthurt or anything, i know i am really bad and i am not going to play solo for 20 match only for this, but in so little time that this is created assholes like benao and terrible appear only to support something that they dont understand, you know why? because their rating is high, so they fell better about themselves.
Personally i think if you are a really good player you will get notice, you dont need simple numbers generate with little to non-information to support you.
Anyway i have to say that the guy behind this system is actually trying, the old one there was a guy who coudnt take one fucking negative critic and he just gave up, so credits to the creator of mydotaskill to keep trying to improve this.
"This is the reason why Valve doesn't like ladders.
Bad players get extremely butthurt, dunning kruger all over the place."
Not only that, good players get extremely arrogant too.
can u post my rating please
http://dotabuff.com/players/90586993
@Dorkly: Its not a "problem" if people take DSR seriously. People can assign value to w/e the hell they like. Yes it would be better if valve just told us our ranking but in that absence sites like this will pop up and players will assign importance to them.
People will discount you, ignore you, or insult you no matter what they decide to use as evidence.
People are going to be assholes in dota (or nice) regardless of how much or little information is available. Thats because people are assholes.
@Benao
you kidding me right ?
do you even take math class ? lol
it takes more than 5 wins to get 1 % win rate if you have over 2000 games played
and it only take you at least 2 3 games to get 1 % win rate if you have below 200 matches played
if you notice ppl can make smurf account and winlike 17 out of 30 games and they will get high rating
^ Do you know that you can play on top5 pages after 3-4 games on a new account?
http://mydotaskill.com/index.php?id=114192409
Well ... looks like my skill dropped to 811. I wonder how did I lose over 200 points just over night :(
^
Everybody lost rating, new model.
Ah, thanks for clearing that up, mate. Oh well, back to the bottom of low normal skill :D
Well mydotaskill.com is officially bonkers. I went from Silver to Bronze and now to gold. But I'm terrible, as I sometimes I don't have the reaction time or hero pool that makes me any good.
For fun, I checked one of the players I queue with. I'm a MUCH better player than him, and he'd admit that. It's not even close.
He's a 401. That's the newbiest player I queue with, so that makes me wonder even more then how I'm a 290. Makes me think it's broken.
Do people not read?
edit: @Zaiko, are you blind or stupid? Pick two.
Like the op said, it is not accurate AT ALL if you stack with players who aren't on your skill level. In order for it to be accurate, you need to stack with players around your skill level so you face off against other players in your bracket. It then probably takes the average DSR of each team and does some calculations to determine your rating.
Even if it's only a 20 game sample, that is more than ample to determine someone's rating. Even if you go on a huge losing streak, you will probably still stay in your skill bracket and your DSR will obviously drop, but not by much. The only thing wrong with DSR is when you stack with players who aren't on your skill level. It will obviously give you an incorrect DSR if you are actually a higher level player.
@Zaiko
I've checked your more recent games and your score should stabilize around 900-1000. Your score was lowered by stacking with worse players.
god dammit im getting ignored by the mydotaskill people
You need to clearly state on your website that DSR will not be accurate if you stack with lower level players. If you can find some way to fix this, I feel like it'll be pretty accurate then. However, based on the information you guys have, it's a pretty decent job. Not bad at all... Keep up the good work.
@m god kami sama 2000
Your DSR is: 1538
@dorkly
i would totally agree with u if that was the case
but it so happens that i am rly skilled so... next time check the skill of whoever u target before naming him
@vaelditha or smth xd dont quit :p
you're getting better... if u were a carry instead of support from the start u would be owning guys! so transition to a pretty carry player :p try drow/potm/luna
oh my god is it really that low
i went from 1720 to 1850 and now im down to 1538
how the hell does the latest iteration calculate this stuff
@MyDotaSkill.com #650 DSR
I added you for a talk would be cool if you have time
also im 18-2 in my last 20 games