Yeah, go head. The problem is the more matches you play with certain heroes, your WR on them tend to decrease. One of the heroes of mine was used to be 90% after 30 games. But then when I play more with it, it gets closer to 50..
Its one factor. But obviously the quality of your opponents should be a big factor, else I for example could jump onto a 2k account with 0 games of TA played and win 100 matches to 4.5k straight with 100% win rate.
What I don't understand is why pros get a separate, weighted division, since their stats are from pubs, which are in identical conditions to majority of diamond 10 players. Also dotabuff's definition of professional is quite dubious at best.
well... thats another way to play drow... he sure got balls. ive considered his skill build but no balls xD
http://www.dotabuff.com/players/1275454 he is ranked #2
and he also builds diffu... guess ill try cuirass some times too...
@Zenoth I think that is perfectly true, but only for a smurf account where your MMR/ranked MMR is not yet consistent
CASE 1: If you can increase your MMR from 2k to 4k using TA and any other hero (with a little less efficiency, ~3.5-4k), your rating will tend to decrease as you go higher MMR no matter which hero you played. However, this is a good indicator of a smurf account.
CASE 2: If you can increase your MMR from 2k to 4k using TA but cannot do so with other heroes then correct, your TA rating will tend to decrease. And this is a good indicator that you ARE really good with TA because she was able to increase your MMR drastically.
But if your MMR remains constant (+/- 300) while you keep on playing a combination of heroes including TA. Is it indicative that the more games you play with TA, the more probable that your rating on her will drop?
IMHO, assuming both cases above are true, your TA rating will also remain consistent (at best), a slight drop off (at worst).
@awakened.. yours is a different point from zenoth's and it could be law of averages or it could be any other factor i.e. (1) like feeling too confident with the hero that even if it's not suited for the lineup anymore you still pick it, (2) or the hero was nerfed and such.. etc, (3) or you played the hero during the earlier percentile of your games where your MMR is not yet consistent and then you just stopped playing the hero when you knew that enemies are not noob anymore.. etc etc etc
but only way to find out is to just play more matches.. which is fair at the least because.. the # of matches is part of the Dotabuff's calculation... as to how it should be weighted compared to other factors is a different conversation altogether
SCORE RANK = DIVISION BONUS (DIFFICULTY) X WIN RATE (SUCCESS) X MATCHES PLAYED (RISK AND EFFORT) X KDA RATIO (IMPACT)
Doesn't matter if it is ranked or unranked, solo or party. The easist way to get good rank is to 5man-stack in unranked and try to finish the game with 0 deaths, KDA/WR/Matchesplayed are the only subjects that matter
i'm not sure what you are talking about... but the point i was trying to illustrate is that i am not sure why you are complaining about being the 57th "only", no offense to you but if I were playing on your account for those games, for example, I would get much better stats than you. The quality of opponents is simply different.
@zenoth i was initially complaining because I never thought that MMR/Division and # of matches were included as factors. I was only looking @ KDA and Win Rate... then I read how it's actually calculated and everything made sense to me. Read thread and you'll understand the flow from "why only 57???" to.. "oh ok i get it get it"..
If you are really a 6K player and you..
- Start @ 2K MMR, Play BEST Lanaya = MMR will go up = better HERO RATING at first but it will plateau til it goes down. Why? Because opponents become better
- Start @ 2K Play BEST ANY HERO = MMR will go up = better stats at first but it will plateau til it goes down. Why? Because opponents become better
Doesn't matter which hero you play. You are 6k player, you'll definitely get better stats at first until it plateaus. But that is because you went from 6k MMR games to 2k MMR games. Your point makes sense here.
But in my case, I am a 4k MMR player..
- play @ 4K MMR, play BEST Razor = MMR same
- play @ 4K MMR, play BEST other heroes = MMR same
Again, still curious as to how it works as it seems a high number in either of the varibles results in you ending up on the toplist. While the list is orgnaized in a way that is pretty easy to see and find players who play a LOT of one hero it doesn't separate between who is actually good and who's just there because their numbers came up.
E.G. 5K MMR players stacks with 2K's in unranked games and plays TA and stomps. Does the system...
Punish him for those games? (Difficulty)
Or reward him for stats (winrate/KDA)
as you have mentioned before the "skill" bracket is unreliable so how is difficulty defined as you have neither access to MMR or or skill brackets except in some matches. Are they guesses? It seems that having the rank "professional" gets you in with ease as well but what do you class as "professional" standing in for 1 match at some joindota tournament? While a good player take http://www.dotabuff.com/players/89654154 I don't recall that being a professional player, nor anyone in the semi pro arena a lot of whom are up there under the "professional" status.
Does a good player lose rank on a hero if they play with lower skilled players and perform well?
As a woods tb player, i couldn't watch the rank of tb, without commenting the formula:
Score Calculations ScoreRank = Division BonusDifficulty x Win RateSuccess x Matches PlayedRisk and Effort x KDA RatioImpact
And especially the kda part:
If we remove the suicide that i did (more than 130, but let's remove only 130)
then my kda as terrorblade, would jump from 3.2 to 3.98, which will increase it by 25%!
I don't know how less is the weight of kda. But if it was a weight of "1" (and other higher) it would then give me a sweet 6800 hero score instead of 5505 (over 7k if we consider i suicided 150 times). -which would place me top1 ^^-
I'm also pretty doubtfull about the league part, how do you choose if one has his hero value in professional or not? If it's based on official matches, then professional 1, can be under diamond 10 in terms of skill, but still offer a better ratio for hero score.
I didn't like the old list because it didn't make any sense. All you needed was having high MMR and an old enough account and if you had even slightly above %60 WR on any hero you would get in. But this one feels even worse.
If you can make it region-based it would be great.
Also high KDA players are usually the first ones to stay at fountain & take low risks so they dont mess with their precious stats, they shouldnt have a big impact on the score, winrate, tower dmg/hero dmg/heal should be more relevant.
I just wanted to let you know that we're seeing all of your valuable feedback and tuning the hero rankings accordingly.
Specifically, we'll be focusing on three areas over the next few days:
- Removal of excluded players (who have been flagged for exploiting) - bad oversight and is top priority.
- Improved detection of low-quality matches
- Improved score calculation at both extremes of match volume (35 or 3500)
We'll be continually improving this for a while. Keep the helpful feedback coming. Thanks!
what do you count doing about suicide?
I mean, not suiciding (in case of tb/lycan woods) is generally a bad move, and here the good move is punished
And it's not only those two heroes:
We can actually extend it to almost all woods hero
•At least 10 sec back to base + at least 20 sec to reg is far inferior to suicide.
at lvl 5 you die for 20 secondes, so assuming you don't lose gold, it's great.
The only problem when you want suicide at lvl like 4-5 or more, is that usually it's 3rd rune (and one of the most important one of the game actually), that you could have defended, or some enemy could have wandered there, or you cann't make a goldless suicide etc....
We built the rankings to accommodate all players, not just those who play ranked. We can and will do a better job of qualifying the people in the top 100, and we will continually improve our calculations for match quality (including favoring ranked where it makes sense). This is my top priority right now.
Benao: short mute for your harassing comment toward another player, not your valid criticism towards me.
Ok what is this division ranking? Or better how is it calculated. I really don't get it, sry if it got answered before. It seems to have no influence on the ranking if i'm not mistaken.
Matches played affect the ranking i'm ok with that, but it looks like if you simply spam a hero but are mediocre with him you get a higher ranking just cause of games played. At least i get the feeling if i look at some guys in the higher tops.
But anyway, thx for this i'm even more happy that i tried bristle out after his release^^ ( Don't think i deserve this spot though)
Excuse me sir,
SCORE RANK=DIVISION BONUS DIFFICULTY X WIN RATE SUCCESS X MATCHES PLAYED RISK AND EFFORT X KDA RATIO IMPACT
how to check the division bonus difficulty?
uhuh.. if it's just the # of matches then I'm fine with that! I can play him some more *yahoo*..
and yes, upon checking it could really be just the # of matches
Yeah, go head. The problem is the more matches you play with certain heroes, your WR on them tend to decrease. One of the heroes of mine was used to be 90% after 30 games. But then when I play more with it, it gets closer to 50..
Its one factor. But obviously the quality of your opponents should be a big factor, else I for example could jump onto a 2k account with 0 games of TA played and win 100 matches to 4.5k straight with 100% win rate.
What I don't understand is why pros get a separate, weighted division, since their stats are from pubs, which are in identical conditions to majority of diamond 10 players. Also dotabuff's definition of professional is quite dubious at best.
also the weightage of number of games played should have a cap like 100
i hope they are just telling us how good of a pro they are instead of what you think it is :)
well... thats another way to play drow... he sure got balls. ive considered his skill build but no balls xD
http://www.dotabuff.com/players/1275454 he is ranked #2
and he also builds diffu... guess ill try cuirass some times too...
i dont like how u need 500+ games
please take away unranked games and change it to 300 games already
+1 ^
this will only benefit ppl who 5 stack in unranked
no offense to satellizer but hes ranked #3 with luna when hes 5 stacking in brazil ? pls and that guy from australia is literally #1 right now
lol thats bad xD ^
I HAVE VERY GOOD SEX SKILLS + WINDRANGER
You are qualified but not in the top 100. Subscribe to Plus to calculate your ranking.
T_T! thanks dotabuff you make me feel gr8
It doesnt show my ranking on certain hero even though i have played at least 1 game in last 30 days, please fix it
@Zenoth I think that is perfectly true, but only for a smurf account where your MMR/ranked MMR is not yet consistent
CASE 1: If you can increase your MMR from 2k to 4k using TA and any other hero (with a little less efficiency, ~3.5-4k), your rating will tend to decrease as you go higher MMR no matter which hero you played. However, this is a good indicator of a smurf account.
CASE 2: If you can increase your MMR from 2k to 4k using TA but cannot do so with other heroes then correct, your TA rating will tend to decrease. And this is a good indicator that you ARE really good with TA because she was able to increase your MMR drastically.
But if your MMR remains constant (+/- 300) while you keep on playing a combination of heroes including TA. Is it indicative that the more games you play with TA, the more probable that your rating on her will drop?
IMHO, assuming both cases above are true, your TA rating will also remain consistent (at best), a slight drop off (at worst).
@awakened.. yours is a different point from zenoth's and it could be law of averages or it could be any other factor i.e. (1) like feeling too confident with the hero that even if it's not suited for the lineup anymore you still pick it, (2) or the hero was nerfed and such.. etc, (3) or you played the hero during the earlier percentile of your games where your MMR is not yet consistent and then you just stopped playing the hero when you knew that enemies are not noob anymore.. etc etc etc
but only way to find out is to just play more matches.. which is fair at the least because.. the # of matches is part of the Dotabuff's calculation... as to how it should be weighted compared to other factors is a different conversation altogether
SCORE RANK = DIVISION BONUS (DIFFICULTY) X WIN RATE (SUCCESS) X MATCHES PLAYED (RISK AND EFFORT) X KDA RATIO (IMPACT)
Doesn't matter if it is ranked or unranked, solo or party. The easist way to get good rank is to 5man-stack in unranked and try to finish the game with 0 deaths, KDA/WR/Matchesplayed are the only subjects that matter
It do have flaws, like every other ranking system in the game including MMR.
some how managed a 4th position with bb, even with shit stats LOL
@BlackXargon
i'm not sure what you are talking about... but the point i was trying to illustrate is that i am not sure why you are complaining about being the 57th "only", no offense to you but if I were playing on your account for those games, for example, I would get much better stats than you. The quality of opponents is simply different.
Good job boys, this site is getting better and better every day
@zenoth i was initially complaining because I never thought that MMR/Division and # of matches were included as factors. I was only looking @ KDA and Win Rate... then I read how it's actually calculated and everything made sense to me. Read thread and you'll understand the flow from "why only 57???" to.. "oh ok i get it get it"..
If you are really a 6K player and you..
- Start @ 2K MMR, Play BEST Lanaya = MMR will go up = better HERO RATING at first but it will plateau til it goes down. Why? Because opponents become better
- Start @ 2K Play BEST ANY HERO = MMR will go up = better stats at first but it will plateau til it goes down. Why? Because opponents become better
Doesn't matter which hero you play. You are 6k player, you'll definitely get better stats at first until it plateaus. But that is because you went from 6k MMR games to 2k MMR games. Your point makes sense here.
But in my case, I am a 4k MMR player..
- play @ 4K MMR, play BEST Razor = MMR same
- play @ 4K MMR, play BEST other heroes = MMR same
is my HERO RATING going to increase?
well wave is talking important things about 500 matches and only ranked games should count in top list!
im on none of the lists from what ive seen
then youve got a 4700 mmr OD player with 3000 games being the 3rd best in the world
and 5000 mmr ta players in front of dfist and wagamama
qojqva is 7th on kunkka with 47% win rate on the hero
unranked games and party count as well so 5 man unranked stackers on australia and south africa will have the highest scores
and this is why valve's mmr system is hands down the best skill measuring tool, everything else ive seen is so inaccurate it's hilarious
dear juice gracing us with his presence from pd forums to rant about the rankings
you're still the most well known with mmr experiments its k
Does the plus feature also apply to mobile ware?
Again, still curious as to how it works as it seems a high number in either of the varibles results in you ending up on the toplist. While the list is orgnaized in a way that is pretty easy to see and find players who play a LOT of one hero it doesn't separate between who is actually good and who's just there because their numbers came up.
E.G. 5K MMR players stacks with 2K's in unranked games and plays TA and stomps. Does the system...
Punish him for those games? (Difficulty)
Or reward him for stats (winrate/KDA)
as you have mentioned before the "skill" bracket is unreliable so how is difficulty defined as you have neither access to MMR or or skill brackets except in some matches. Are they guesses? It seems that having the rank "professional" gets you in with ease as well but what do you class as "professional" standing in for 1 match at some joindota tournament? While a good player take http://www.dotabuff.com/players/89654154 I don't recall that being a professional player, nor anyone in the semi pro arena a lot of whom are up there under the "professional" status.
Does a good player lose rank on a hero if they play with lower skilled players and perform well?
Cool update if possible, made the top 100 list to top 200 list like dota 2 leaderboards!
As a woods tb player, i couldn't watch the rank of tb, without commenting the formula:
Score Calculations ScoreRank = Division BonusDifficulty x Win RateSuccess x Matches PlayedRisk and Effort x KDA RatioImpact
And especially the kda part:
If we remove the suicide that i did (more than 130, but let's remove only 130)
then my kda as terrorblade, would jump from 3.2 to 3.98, which will increase it by 25%!
I don't know how less is the weight of kda. But if it was a weight of "1" (and other higher) it would then give me a sweet 6800 hero score instead of 5505 (over 7k if we consider i suicided 150 times). -which would place me top1 ^^-
I'm also pretty doubtfull about the league part, how do you choose if one has his hero value in professional or not? If it's based on official matches, then professional 1, can be under diamond 10 in terms of skill, but still offer a better ratio for hero score.
I think KDA should be replaced with something that contains Healing / Tower damage / Hero damage.
If you have 10k+ tower damage you have quite an impact, even with negative kda. Your team gets plenty of gold because of you.
then full rat furions would be placed higher than good old utility furions
These full rat furions get higher kda.
I didn't like the old list because it didn't make any sense. All you needed was having high MMR and an old enough account and if you had even slightly above %60 WR on any hero you would get in. But this one feels even worse.
so it isnt about how many matches played
http://www.dotabuff.com/players/117782763
this guy is #2 on brood with literally zero ranked matches played
much amaze
If you can make it region-based it would be great.
Also high KDA players are usually the first ones to stay at fountain & take low risks so they dont mess with their precious stats, they shouldnt have a big impact on the score, winrate, tower dmg/hero dmg/heal should be more relevant.
Hi everyone,
I just wanted to let you know that we're seeing all of your valuable feedback and tuning the hero rankings accordingly.
Specifically, we'll be focusing on three areas over the next few days:
- Removal of excluded players (who have been flagged for exploiting) - bad oversight and is top priority.
- Improved detection of low-quality matches
- Improved score calculation at both extremes of match volume (35 or 3500)
We'll be continually improving this for a while. Keep the helpful feedback coming. Thanks!
^
Solo players being judged the same as stacked players
Un ranked and ranked judge the same.
I realize there are logistic issues there
Interesting feature. Keep up the good work.
@Concede both good points, we'll be improving in this area. It's a part of detecting the quality of a match.
yeah but jason... thats a huge defect wtf. thats like the basics of ranking the top100
what do you count doing about suicide?
I mean, not suiciding (in case of tb/lycan woods) is generally a bad move, and here the good move is punished
And it's not only those two heroes:
We can actually extend it to almost all woods hero
•At least 10 sec back to base + at least 20 sec to reg is far inferior to suicide.
at lvl 5 you die for 20 secondes, so assuming you don't lose gold, it's great.
The only problem when you want suicide at lvl like 4-5 or more, is that usually it's 3rd rune (and one of the most important one of the game actually), that you could have defended, or some enemy could have wandered there, or you cann't make a goldless suicide etc....
Dying to neutrals or Roshan isn't always intentional so I think it should count really.
We built the rankings to accommodate all players, not just those who play ranked. We can and will do a better job of qualifying the people in the top 100, and we will continually improve our calculations for match quality (including favoring ranked where it makes sense). This is my top priority right now.
Benao: short mute for your harassing comment toward another player, not your valid criticism towards me.
Re: suicides, KDA makes up only a tiny portion of your score. An occasional intentional suicide wont have any noticeable effect.
Ok what is this division ranking? Or better how is it calculated. I really don't get it, sry if it got answered before. It seems to have no influence on the ranking if i'm not mistaken.
Matches played affect the ranking i'm ok with that, but it looks like if you simply spam a hero but are mediocre with him you get a higher ranking just cause of games played. At least i get the feeling if i look at some guys in the higher tops.
But anyway, thx for this i'm even more happy that i tried bristle out after his release^^ ( Don't think i deserve this spot though)
Excuse me sir,
SCORE RANK=DIVISION BONUS DIFFICULTY X WIN RATE SUCCESS X MATCHES PLAYED RISK AND EFFORT X KDA RATIO IMPACT
how to check the division bonus difficulty?
This system is great! Well done Dotabuff guys!
I'm 46th percentile on Lifestealer. 6.6 kda and 70% wr. Means I am playing against bad players which is also helpful information.
I've been a plus member for awhile and improvements like this will keep me doing so.
Thanks guys!
#43 wisp!
bought + just to check my sf ranking cause i was mad at not being top 100 xD
turns out im 123rd ... gg
>30
>unranked
>party
none of these three should matter