I don't really see the point in dismissing his skill rating as useless/bad/etc. Sure it will have flaws but what model is perfectly accurate? Since valve doesn't reveal MMR this is a usable alternative, even if it isn't entirely accurate.
If you provide some examples to discredit the system such as
"make a new account
stack with 5 to 30-0 or so score with high kda and gpm
enjoy 1800+"
then at least explain how a good model will avoid such issues.
Valve doesn't provide all the data through their API so there really is a limit of how accurate you can get with the information available. I don't even think you can get a 'correct algorithm' with the data available. Even though it's possible to parse replays and get more information, the benefits aren't that significant and it would expend too many resources.
It's really easy to bash on a system and nitpick some flaws, it's a lot more difficult to provide useful criticism that could improve it.
i've seen couple accounts who are barely in very high after 50 or so games having better ratings than pro players who play all of their games on front page, i think that pretty much seals the deal that valve's rating system is better.
my "30-0" was just an example, you most certainly don't need to have that kind of stats to place good.
whatever im better than all of you bronze shits, get good then start arguing rofl.........
VALVE's internal ranking doesn't stabilize untill you've played at least 50 games. It's obvious you can't really measure anyone's skill level if he or she didn't play enough games.
As far as I've seen so far, the system doesn't really work. I think it's very dependent on winrate. I checked high/very high bracket players with 47,5% winrate and stuff like this and they were at 370 or 450 and these kind of ratings. I know I'm not better than they are but I'm at 800 and I am in high bracket as far as I know.
It's almost impossible to create a fair ladder system since you can't define what skill level your opponents got. It's pretty easy to get high KDA in normal, but if you play very high against good players - its not that easy.
So I'm guessing 493 is like super good yeah? I'm so pro. Nearly 500 is so strong. Better put that in my steam name, make people know exactly what they're going up against when we play. Not fair to just rape everyone.
"It's almost impossible to create a fair ladder system since you can't define what skill level your opponents got. It's pretty easy to get high KDA in normal, but if you play very high against good players - its not that easy. "
This.
Better WinRate or Better KDA doesn't necessarily mean that it's better.
I get 1157 -_-
I have no idea, the site told me 99%. I don't think 99% of dota players have 51% win rate, therefore my terrible dota skills have been picked up by the site despite me having a better than average win rate.
I believe he said it looks at the people you face (dno how he does that), your hero performance compared to other players and your win rate. It is most likely weighted towards people with more matches played but not necessarily.
We can see that this system favorises typical playstyles by just looking at the spread of the Alliance members' DSR. There is no way that Akke/EGM is that much worse than Loda.
"1152" (top 10%). Someone clarify if I'm understanding this properly but I'm thinking it's "Top 10%" of the current database w/c is probably a tentative figure and will change as more data is gathered.
Any words from the site dev?
It's probably rating people either too high or too low atm because of the lack of player data.
not really, russia has more skilled players than euw.
you have qwistas stack and many more playing in russia and europe east but it's understandable, you are probably in a really low tier bracket so you dont' have to worry about facing them
Seems rather biased towards carries and the likes, as most systems have been. I have a 717 rating on this account, and a 993 on my smurf (around 82 games played). I've yet to see someone who plays support as often as I do with an even vaguely good score. Anyone with an example of a proper support with a decent rating?
Guys do you see once you check your `DSR` that`s it. After 10 games or smth is still same lol. Doesn`t change even 1 point like dbr was or smth lol... so guy who is first will 1910 that`s it......... no changes if he loss next 100 :D
I dont get how anyone can take this shit seriously, I linked Terrible's games in which a majority of them are played in normal to high ( 70 games total) and he is ranked higher than most of the top players in the world
Fuck this site, firslt I got 1000 points as others, then I got -1 lol, and now I get "Error: Player number 80952441 game data couldn't be fetched. Please try again soon.".
I don't really see the point in dismissing his skill rating as useless/bad/etc. Sure it will have flaws but what model is perfectly accurate? Since valve doesn't reveal MMR this is a usable alternative, even if it isn't entirely accurate.
If you provide some examples to discredit the system such as
"make a new account
stack with 5 to 30-0 or so score with high kda and gpm
enjoy 1800+"
then at least explain how a good model will avoid such issues.
Valve doesn't provide all the data through their API so there really is a limit of how accurate you can get with the information available. I don't even think you can get a 'correct algorithm' with the data available. Even though it's possible to parse replays and get more information, the benefits aren't that significant and it would expend too many resources.
It's really easy to bash on a system and nitpick some flaws, it's a lot more difficult to provide useful criticism that could improve it.
i've seen couple accounts who are barely in very high after 50 or so games having better ratings than pro players who play all of their games on front page, i think that pretty much seals the deal that valve's rating system is better.
my "30-0" was just an example, you most certainly don't need to have that kind of stats to place good.
whatever im better than all of you bronze shits, get good then start arguing rofl.........
VALVE's internal ranking doesn't stabilize untill you've played at least 50 games. It's obvious you can't really measure anyone's skill level if he or she didn't play enough games.
As far as I've seen so far, the system doesn't really work. I think it's very dependent on winrate. I checked high/very high bracket players with 47,5% winrate and stuff like this and they were at 370 or 450 and these kind of ratings. I know I'm not better than they are but I'm at 800 and I am in high bracket as far as I know.
It's almost impossible to create a fair ladder system since you can't define what skill level your opponents got. It's pretty easy to get high KDA in normal, but if you play very high against good players - its not that easy.
I got 1114.
So I'm guessing 493 is like super good yeah? I'm so pro. Nearly 500 is so strong. Better put that in my steam name, make people know exactly what they're going up against when we play. Not fair to just rape everyone.
tis obviously a win rate ranking first and foremost.
"It's almost impossible to create a fair ladder system since you can't define what skill level your opponents got. It's pretty easy to get high KDA in normal, but if you play very high against good players - its not that easy. "
This.
Better WinRate or Better KDA doesn't necessarily mean that it's better.
I get 1157 -_-
Doubt it has much to do with win rate. My 51% would place me higher than 99% I would think.
try it out, I sincerely doubt it.
Yeah, I just did. 493 rating which is in the top 99%, phew, but I have 51% win.
I didn't choose the support life, the support life chose me.
http://mydotaskill.com/index.php?id=http%3A%2F%2Fdotabuff.com%2Fplayers%2F93256974
winrate scaling so impressive
dsr is as good as valves mm
being in the top 99 is not the same as being above 99%.
your two posts contradict itself.
I have no idea, the site told me 99%. I don't think 99% of dota players have 51% win rate, therefore my terrible dota skills have been picked up by the site despite me having a better than average win rate.
it's unbiased or maybe its because my acc is banned here and it doesnt take statistics past the point i got banned or something idk
my rating is 1523 something "This player has a Pro DSR score (top 3%)." and ive looked at the top 5 and they have bad winrate and kda
compared to mine
whats the formula for this? im calling you out site owner fucking say it
I believe he said it looks at the people you face (dno how he does that), your hero performance compared to other players and your win rate. It is most likely weighted towards people with more matches played but not necessarily.
The system does not seem to work very well for people that generally only play one hero, just look at this guys DSR who had one of the highest DBR http://mydotaskill.com/index.php?id=http%3A%2F%2Fdotabuff.com%2Fplayers%2F86736893
So your hero performance is probably very important.
We can see that this system favorises typical playstyles by just looking at the spread of the Alliance members' DSR. There is no way that Akke/EGM is that much worse than Loda.
http://mydotaskill.com/index.php?id=http%3A%2F%2Fdotabuff.com%2Fplayers%2F150672172
1623 rating, but i do not consider myself that good, compared to many others
http://mydotaskill.com/index.php?id=92561815
"1152" (top 10%). Someone clarify if I'm understanding this properly but I'm thinking it's "Top 10%" of the current database w/c is probably a tentative figure and will change as more data is gathered.
Any words from the site dev?
It's probably rating people either too high or too low atm because of the lack of player data.
1250
obviously that 1 should be 2, fundamentally broken.
http://mydotaskill.com/index.php?id=http%3A%2F%2Fdotabuff.com%2Fplayers%2F129865496
1120,
I agree entirely with xan.
However, I do think there's a massive bias here towards smurfs,
Say all mid and I'll dodge has 1623 rating (higher than EGM and Akka among many other pros) on only 111 games.
I also doubt the server strength is taken into account, I'm sure I'd be 1300+ on Russia but I only queue EU West, the 'hardest' server.
not really, russia has more skilled players than euw.
you have qwistas stack and many more playing in russia and europe east but it's understandable, you are probably in a really low tier bracket so you dont' have to worry about facing them
Seems rather biased towards carries and the likes, as most systems have been. I have a 717 rating on this account, and a 993 on my smurf (around 82 games played). I've yet to see someone who plays support as often as I do with an even vaguely good score. Anyone with an example of a proper support with a decent rating?
Pirate also has a 3 digit rating so I think it is safe to say that hero performance on multiple heroes is extremely important.
lulz. even russians dont want to play with other russians.
@ vivi take your 1 game, 1 loss, hard feed smurf trolling elsewhere.
@NP.Penis Monkey
what the fuck does that have to do with anything you bronze tier?
just noticed you play with havoc badger, dont even bother replying and go read basic english for kids or something ~~~
If you even clicked on the original link, I'm placed in Platinum on this current system.
I know Havoc IRL. So what?
Show me your main or are you too embarrassed of the stats on that too?
fucking trolling scrubs.
Hey filthy smurf, I am here so any issues you can address directly to me.
>not really, russia has more skilled players than euw.
well this is just wrong
>863
I lol'd.
http://dotabuff.com/players/150672172 <- higher dsr than dendi :D
Well, mine and my friends' seem accurate
Guys do you see once you check your `DSR` that`s it. After 10 games or smth is still same lol. Doesn`t change even 1 point like dbr was or smth lol... so guy who is first will 1910 that`s it......... no changes if he loss next 100 :D
probably not enough server power to allow regular upates
The guy's still adding features. I think it will refresh; I think Broodstar is now 3rd place for example.
there is no point adding features when the algorithm is completely wrong.....
from now on u need to be at least jadeite or higher to post in this thread
wtf is jadeite I don't even know this word
some kind of emerald or whatever
I prefer my platinum rank, suck it jadeites
685
I am lost. What is this rating? I think I am really bad.
stop posting, ur not ~~~~jadeite~~~~
Terrible 70 games
Much games
So pro
Waw
I dont get how anyone can take this shit seriously, I linked Terrible's games in which a majority of them are played in normal to high ( 70 games total) and he is ranked higher than most of the top players in the world
Cmon upgrade site so no same rating always..
Fuck this site, firslt I got 1000 points as others, then I got -1 lol, and now I get "Error: Player number 80952441 game data couldn't be fetched. Please try again soon.".
Adu'P Earth Spirit best player in the world you scrubs
330 hours of dota 2 OP
mason is the best player?
what a fucking retarded system
781
started at the bottom
NOW WE HERE
pro rank ;_;
road to jadeite