General DiscussionPro's are overrated in ranking feature

Are you mad? Sorry, but a 3.7k player with 50% wr and <3.0 kda shouldn't even be in the top 100. You're there because the formula used to calculate hero score is so fucking retarded that didn't put any kind of cap to matches played and it seems that it doesn't scale, so you get to be there just because you spammed the hero for 3k matches, not because you're any good at him.
Don't get me wrong, the system is flawed in several lvls, but your argument has the consistency of water. "It's really a lot easier to win with a team of awesome players than to win with a team of raging and flaming pubs who have no synergy whatsoever. So I don't get why professionals get their extra levels, especially since they don't need that extra fame." I'd say it's easier to win playing against 5 solo q raging idiots with no synergy whatsoever than against a coordinated 5 stack (because, you know, when you're stacking you get matched up against another stack).

typhox isn't 3.7k, there were numerous "high skill" closing into "very high" players as top 1 for most heroes who is now in top 20 and there are many 6k+ players who is missing

Thanks for the feedback, we're considering making small adjustments to the platinum, diamond and professional divisions. We're definitely making some tweaks to the algorithms and criteria that will raise someone to the top of the list, it's a bit too vulnerable to extremes right now.

That said, I've looked at this data set specifically and one of the things we're tweaking because of it is the curve on match volume - your score is a bit too inflated from playing so many matches.

49.98% winrate, 2.47 KDA? Just because you have 3.3k games doesn't mean you should be there...

wouldn't it make sense to make the requirements a bit higher?

lets say you need like 150 matches before you can get into the toplist, wouldn't that remove all those extreme examples for now?

This is going to sound stupid, but with so many people complaining, maybe it'd be a good idea to split the rankings into the divisions?
(aka Bronze, Silver, Gold ect.) each have their ranking?

theres a player stuck in high skill on numerous hero rankings and why is me and my friends smurf not in any of these leaderboards when he picked brood to 7k and i picked templar assassin to 6.2k

even swiftending is missing, its hilarious

@wave link profiles and I can look into it.

@Licecolony we've considered that and may do it later but are focused on tuning what's in front of us first.

http://www.dotabuff.com/players/86751819

@Notsocuteowl:
What you have to do is to represent the "number of matches played" factor by a function that describes a bell curve or something similar. This way you will guarantee that said factor will have low impact below and above certain tressholds (e.g. 150-1000 matches) while still rewarding risk.

@wave you should be higher than quite a few people above you - some of the tweaks we're making will take care of that. Playing more matches will improve the score by a LOT if you can keep up the win rate.

I think in general, the system favours abusers currently, the reason why dotabuff even listed me as Platinum 10 instead of Diamond was because I was stacking with low skill allies. I think most of the people in the ranking are currently doing exactly that.

In addition to that, I'd like to suggest to throw away old matches that are later than - say - 5 months. After several balance patches and meta changes, heroes play differently, if a player has played 2k games with Morphling in 2012 but only one in 2014 he should not get to the first place in the ranking (no concrete example).

Also I want to remind you that for me it's completely impossible to get a winrate higher than 50% due to already being pretty high ranked (it would require me to gain 5000~20000 MMR which is impossible). Except if I abuse the system by continuing playing with lower ranked players or by artificially lowering my MMR with other heroes.
That makes the algorithm flawed as it is punishing people like me for not having a "high winrate" despite it being absolutely impossible to get one no matter how good I play.

@Oleksandr

I stack often and also played in some "fun" tournaments and inhouses, and I can confirm that playing in a stack is A LOT easier than playing solo, ESPECIALLY when it comes to playing carries.

you'd need to stack a lot in unranked to increase your winrate on OD

whats ur rating now typhox

4622 after a loss streak

"your score is a bit too inflated from playing so many matches."

haha rekd

"Also I want to remind you that for me it's completely impossible to get a winrate higher than 50%"

no it isnt because if you were actually good with OD you'd simply win more games than you'd lose

the only reason your statement is somewhat right is because mmr is getting inflated and if the number of games is high enough that effect is really insignificant to be seen in your winrate
if you solo queue and play a shitton of games you're gonna have something really close to 50% unless you get a lot of +15-+17 games when you're 6.3k+ MMR and play against 4ks where you can reach a stable MMR while having greater 50%

@Androgynous

It is. If I win my next 1000 games in a row to get a 60% winrate, I would need to gain 25 000 MMR points, which is impossible to do. Learn logic please, it's not hard. 1 + 1 = 2.

you dont need 60% winrate, you can't even manage 50%.

fyi you can play unranked, and when you get higher in MMR, like 6k+ you tend to lose more points when losing aind win less when you win because of the generally uneven mmr spread. so you wouldn't be gaining 25 per match.

learn logic please, it's not hard.

its always +/-25 if you solo queue unless you're 7k+ and queueing at odd hours to beat on 4ks lol

why do you argue with 3k trash? He won't understand either way.

can you pls check if im underrated on SF, Drow, Mirana and Pugna... and by how much (pugna/mirana) cause its really odd... i am definitely better than that and it makes me want to cry being ranked so low :'(

http://www.dotabuff.com/players/78410961

And another questions... Why can't i see other peoples Top 100?

"And another questions... Why can't i see other peoples Top 100?"

you can. you can't see other peoples ratings outside of that though. so if it's percentile or 101-1000th that's only personal

Yeah? Sure? What tab is it under?

lololol fu wave... you went full retard playing us east and met my stack (ex clan) of noobs :'(
http://www.dotabuff.com/matches/417220689
just noticed

@Androgynous

So, according to your logic, by having a 55% winrate with OD what you are EXPECTING from me, I would be the best player in the world by a wide margin.
If it is really like you say, how come people in the top teams aren't just spamming one single hero and easily win every tournament?
Makes no sense. You're just being illogical.
Despite the fact that I highly doubt that you know how the points for 6k+ players are being distributed.

LOL benao

im trying to remember who brood is but i cant, i think it was oxen

wtf this brings memory, good ol' brood stacking

kinda makes no sense u can play as much as u wish ur favorite hero and im pretty sure that the dudes 2k mmr over u and the pros will play that hero alot better even if they dont rly practice it that much, u can get a higher hero score on the board but its sure as hell that u aint no match for the pros, same can be said for myself. Their game mechanics and concept + execution is so superior that its kinda silly to even consider why u aint higher on boards than these dudes. Think it this way...take a random hero and give it to a 2k guy to play that shit for countless games, if he still sits on that 2k chances r that he is no match for u, in the end u will be better than him even on that hero.

oxen lmfao

cancer in a nutshell

hes funny as hell lol but hes creepy as fuck

so how do i see other peoples top 100? i cant seem to find it (if it even exists as you point out)

btw talking about oxen... im pretty sure i reported him and got him mtued for losing me a game playing like shit and thinking he's the shit

Lol booster, Dubai :D:D:D:D

oxen is trying to redeem himself, give him a new chance LOL
im sure he regrets what he did

@ smaug, you don't need to gain thousands of mmr to have over a 50% win rate on OD you need to win 3-4 games on him (75-100mmr). people would also be willing to except you as one of the best OD players if you either had a win rate substantially over 50% OR had an exceptionally high mmr, even just making it past 5k would make your OD look a lot better.

#4 OD
http://www.dotabuff.com/matches/883555339

By the way what is Smaug's mmr? Just wondering.

@ smaug,

I dont mean to be a dick but you can't even hold onto a 50% WR or even break 2.5kda with over 3k games and you honestly expect to be the in the top 100 of a game with over 60mil monthly players

^ He's actually 4th in that ranking, so your remark is way off.

On the other hand, I do agree, that 50% WR and 2.5 KDA OD seems underwhelming to be placed so high in the top ranking. And, @Smaug, about your statement about saying, it's impossible for you to get higher winrate due to an amount of games you played - it's not actually matters. Your winrate for a last month (~100 games) is also sub-50%.

Who has "a win rate substantially over 50%" with 3k+ games on one hero? Basically if you reach your mmr, your winrate should be ~50%. Unless you start to spam a hero from 2k mmr and reach 6k+, most times after 1k games you should have ~50% winrate.

Edit: ah, found him: http://www.dotabuff.com/players/85732420

@Jason just curious.. what statistical tool are you guys using? R language? have u created a statistical model in doing this?

Haha wtf. This guy just makes me rofl so hard, fucking scrub with 3k games of OD and stagnating on MMR. Jesus christ.

Get a life dude, you clearly need one because you seem autistic as fuck.

I'm pretty sure i can play OD better than you with just 10-20 games of pure OD. Rofl fucking retard.

bla bla bla . nobody gives a sh*t

Edit. Go outside bro holy. 4 real

I'm pretty sure Qojqva and Black plays a way better OD than you.

3k games, 48% winrate, 2 KDR

Skill doesnt come with playtime. Also they understand heroes way faster than you, aswell as understanding the game way superior.

You're criticizing other people's logic but you fail to see how your own is flawed.

The answer is that the system is still only young and an extreme in any of the factors is going to inflate your hero score.

You are inflated because you have a huge number of games.

The people above you are inflated (questionable) because they have high % win rate and high KDA.

No matter what you think, you are not better than qojqva and black at OD.

So, even if the system is flawed now, it is correctly representing that those players are better than you.

The question is which factor (kda/winrate/games played) should be weighted the most, and it's undeniable that a good kda and win rate > games played. EVEN IF they are stacking, it's unreasonable for a system to take this into account accurately.

are the ppl flaming op for real.
this metric is so dumb why are getting pros getting extra points for being pros when only games played in ranked and normal matchmaking are considered?
look at blacks od matches, 15 ranked and 38 unranked = 53 total

http://www.dotabuff.com/heroes/outworld-devourer/players
dotabuff uses these 53 matches to access him.
say black is 6k
both black and some other 6k pubstar performs equally well on the hero but black is given more points simply by virtue of having esports history
why?
either dotabuff remove this retarded "professional" division or include their stats from competitive games.

why do people even care so much?

because this rating system is quite the trainwreck

I don't care about it at all but how is the fuck is this guy in top 100 rubick and I'm not? http://www.dotabuff.com/players/136186119

Just check out one game of Black's OD, and then one of mine and you see that there is a huge difference between players. Someone who doesn't even know how Astral Imprisonment or Sanity's Eclipse work mechanically-wise, can't last hit with OD under tower and buys items that make no sense is supposed to be higher ranked than me?

Of course you can do that just because he plays better overall, but then I can just question the use of the rankings altogether, as those people aren't good on OD at all, they are just good at the game, and everyone who copies them will undoubtably fall, because their item builds are shit, their skill builds are shit and the way they play the hero are shit.

Proof?

http://www.dotabuff.com/players/21289303/matches?date=&faction=&hero=outworld-devourer&lobby_type=&game_mode=&region=&duration=

Check out black's history. No constant item decisions/placement, meaning that he has actually no clue what items to buy on the hero and just wildly guesses (as you can see he doesn't have a preferred inventory slot for either item, which means that he doesn't have experience). Lots of last hits, low amount of teamfight participation costing him most of his games (as OD is a better midgamer than lategamer), from his last 5 ranked matches with him, he lost 4.

It's great to have a good KDA, but it's not hard if you're just spending the entire game farming in the jungle instead of taking the risk to teamfight.

My own games are not exactly overwhelming, I am having lots of bad games, many of them lost (or won) by the first second by people raging, flaming and ragequitting or by a single bad decision, I am greatly unhappy with the way I'm playing, etc. But the difference is, I am having an idea of what I am doing, what the hero is capable of, what effects different item- and skill-builds have. Maybe my map awareness isn't as great as his, or my ability to last hit, but I understand this hero in every aspect and don't need to buy random items. People who are looking at my games can actually learn about the hero, learn tricks and techniques, how to blink dagger out of Disruptor Ult, how to win lane against Razor, how to farm when you're behind, etc. I doubt anyone who looks at black's games can actually learn anything.

@dLIGHT

Good Winrate causes good KDA, so it's the same stat, and with a certain number of matches played, the only way to increase your winrate is by abusing.
I'm currently sitting between 4500 and 5300 MMR and I don't think I could get higher unless I start playing different heroes (preferable strong or versatile ones).

I have the #2 Phantom Lancer (TAKI) in my Friend List just to confirm everything. Neither his high winrate nor his high KDA are a result of being so much better. They are a result of stacking with friends.

I stacked the last few weeks and got a huge increase in KDA on OD as well. So what gives.