Don't hold too much stock in the dotabuff rankings, pros tend to get ranked highly even if they haven't ever played the hero in a competitive game, even if they don't have that gr8 of a KDA, win rate or even a super high mmr (compared to the other players on the list). there are also a few people near the tops of those lists who have <50% win rates and bad mmrs, but 4000 games played with one hero.
Professional division have a huge, huge bonus in terms of score, and theres nothing to be done here.
Yeah it's extremely biased towards players who joined ti/major competitions
http://www.dotabuff.com/heroes/bloodseeker/players
Padrinho has similar win rate, lower kda and much lower matches played than me, yet he has 175 higher hero scores than me.
I wonder why the need for a professional division when games played during competitions are not counted in dotabuff stats (kda, winrate etc)? Professionals who are good at certain heroes should do well in public games hence getting a higher rank even at diamond 10 anyway?
Riwu ^ do you have any tips for my BS builds/Items etc. I'm close to 4k and my seeker last week was around 42% overall, now its like 53% with a 72% this week with around 40 games.
If you see this could you give my any tips on how you deal with a storm (i tend to get burst down easily especially once he gets orchid) I generally go jungle/mid and rush a blademail although i sometimes build euls/dagon but not so much atm.
Btw, did we had any reason why the professional bracket is still there ?
That would already fix half of the problem of the rank, without a lot of work.
Ofc the part about people playing 4k ranked game on one hero would be an other problem, but we don't need to solve every probleme at once
thats not the biggest problem!
The problem is they don't only count ranked games, that includes pro bracket and wrong calculation of division by stacking in unranked and playing too many LP games (me -> down to platinum 5?)
the fact that you can loose points for winning a game is pretty bad too (if you stacked with low skill friends). its really just a gimmick so people on the list can use to say "look top 100 heroName", there is nothing wrong with dotabuff having an inaccurate top 100 list as opposed to no lists . Although it would really be nice to see what the best players are doing with specific heroes so that people can know who to watch and what builds to use to improve them selves.
Since they DotaBuff tracks MMR now, more weight should be given to a players MMR and what there winrate is with a hero. It wouldn't be hard to tell if a player was playing in a lower bracket and ignore those games either. if for example a 6k SF player ends up in a high skill, none ranked game it should be easy enough to make that game not count towards his SF ranking, since he was clearly stacking.
The problem with win rate is people can make a new account and spam that hero with their stack in unranked and win most if not all the time. I believe maybe the first 30 or so games should be void of ranking for new players, so players needs to have probably 500 games or so to even be considered in the calculations or so to reduce the number of smurf.
The highest ranking in a pub is ranked games, we can't get to professional division, but maybe there should be a system where we play in a highly competitive games and perform consistently well like win we can get to another division call 'Specialist' which is higher than diamond where we specialise in that hero.
But that's my guess.
There are lots of flaws with the system. Divisions not being locked to the games the hero is being played in, system factoring games multiple patches/years and skill brackets ago with the same importance as recent games, professional tier being completely OP and automatically puts you into the top 10 with the hero, KDA and Winrate being factored in while simple things like Deaths per game or Game Lengths are ignored massively skewing the data, pro's being top #1 when they play the hero less than once a month, unranked games having the same weight as ranked ones, etc.
Tracking MMR makes the listings just more useless. Instead of hero specialists you just get a 1:1 copy of the leaderboards. Oh yea I forgot, Aui is specialist on all heroes, so why even bother having a ranking?
People are claiming that in a correct system I wouldn't be in the top for OD. Correct is: In a correct system I would most likely be the only one in that ranking since there is just noone else who plays OD in high ranked games for a significant amount of games.
The problem with Winrate is that it is highly inaccurate. I played OD on this account for 3 years. To increase my KDA on him by 0.01 points I need to stomp 20 games in a row! 1 win increases my winrate by 0.015%. If the winrate would only consider the stats from the last 3 months, it would be a much better reflection of your overall winrate than the idea of treating games you played 3 years ago in a completely different hero balance on a completely different system as a completely different player just like your most recent game.
That being said, Winrate in itself is a pretty bad statistic. I have a 2k solo MMR smurf (gifted from a friend), I can play 100 games Phoenix or OD on the account before reaching my true MMR, which means I could easily get 80% or 90% winrate on a 4.5k MMR account. Even if I then stay at that MMR for another 100 games (200 in total), I would still have a pretty comfortable 60~70% winrate.
Having a high winrate means you're playing in games that are below your skill level. It doesn't mean you're good. It just means you are playing games against noobs. Having a low winrate doesn't mean you're bad. It just means you're playing against enemies that are stronger than you. That doesn't EXCLUDE that you're bad nor good, it just can not give any information about it. The reason for that is that the Matchmaking Algorithm tries to match you against stronger enemies (or with weaker allies) the more you win and with stronger allies or weaker enemies the more you lose. Having a 50% winrate doesn't mean you're bad or good, it just means the Matchmaking system is working as intended.
Winrate could be an indicator of your skill if you got matched with and against random players every game, with no bias. As long as that doesn't happen, it is useless as a skill indicator. Just like KDA which strongly depends on your winrate.
oh look. Smaug showed up to type walls of texts because playing 2 heroes is all that matters. I wonder what his lawyer is doing.
But yea Benao is right. Top list should really only count ranked games that are very high (doubt it can count based on mmr.) And it should also limit the effect that playing thousands of games has. Should be a bell curve of effect on the ranking with about 250 games at the top.
oh look, Concede the hater is still not banned from this platform :)
Judging from your match history I guess you are just really really bad at dota and have no clue what you are talking about.
Dunning Kruger effect.
Like most 3k's.
@smaug you are mid a 4k player with a 49% win rate and a 2.5kda on OD and you should be happy that you are in the top 100 at all, not ranting about how you should be higher in the rankings.
you can write as many walls of text as you want, but the fact remains that you loose with the hero more then you win, and you are not even loosing to the best players.
I don't think your a bad player or even a bad OD player, but you really need to keep your ego in check.
49.99% winrate is not "lose more than you win" especially when it's about 70% winrate during my last few months.
I am a 5k MMR player, I don't boost my MMR playing overpowered heroes unlike most of you do, in fact I play a very unique style. I have improved by humongous amounts in the last year, but you can't see it because you are too much fixated on winrate. Truth is, my MMR is still 5k just like it was 1 year ago which is exactly 50% winrate. Truth is as well: I defeated most 6k and 7k players including people like this Vroksnak and other people from the leaderboards.
Fact is, I never have said ANYTHING related to my ego, I simply said logical facts that you didn't tackle my arguments at all and are just trying to discredit me for whatever reason. I bet you didn't even read them.
That's just pure bullshitting you're doing right there. I have NEVER mentioned that I was good or in a place that I didn't deserve or that I did. In fact, my posting was COMPLETELY 100% neutral just stating logical connections and it should be in your own interest to take it serious in case you actually want to have a good system and not one that is horribly flawed.
Fact remains, your winrate has nothing to do with your skill. I can play ranked and keep at 50% winrate like forever, I can play unranked and win 70% of my games with ease. Doesn't need more proof to see that it obviously makes a huge difference which Matchmaking you're choosing.
And @Concede,
The Dunning–Kruger effect is a cognitive bias wherein unskilled individuals suffer from illusory superiority, mistakenly assessing their ability to be much higher than is accurate.
In other words, you thinking you can criticize players on 5k MMR while you yourself don't even have 4k.
Yea, you still don't seem to know what it means. I literally never post, write or state anything regarding my skill level. I never claim to be good, average, great or bad.
You however, have claimed that you are going to take legal action against forum posters, along with a litany of other absolutely ridiculous statements in the posts you have made on multiple forums.
You are a joke, and it has very little to do with dota.
do you have nothing better to do than act condescending to everyone? seriously don't, please, dota is supposed to be happy land of love and happiness
"49.99% winrate is not "lose more than you win"'
"I am a 5k MMR player, I don't boost my MMR playing overpowered heroes unlike most of you do, in fact I play a very unique style. I have improved by humongous amounts in the last year"
"Truth is as well: I defeated most 6k and 7k players including people like this Vroksnak and other people from the leaderboards."
"Fact is, I never have said ANYTHING related to my ego" "I have NEVER mentioned that I was good"
"I can play unranked and win 70% of my games with ease."
http://www.dotabuff.com/players/86717663/trends?date=3month&hero=&skill_bracket=&lobby_type=normal_matchmaking&game_mode=®ion=&faction=&metric=win
More like 52% over the last three months in normal matchmaking, not that there is anything wrong with that.
I used to care cause I thought it would be cool. Then I saw a bunch of my 3.8K friends on the leaderboards and just didn't care.
You can be #50th axe and then queue in a stack and be #290th the system is ass and they won't rework it so why bother.
Looks like this Concede guy is just trolling and not putting any content. Too bad this forum doesn't have an ignore function.
Whatever, guess you just can't have a serious discussion on this forum.
" I can play unranked and win 70% of my games with ease."
Yet you had 51-52% winrate the past 3 months.
"Whatever, guess you just can't have a serious discussion on this forum."
Stop bullshitting and people might take you seriously.
^ It's ridiculous how as soon as I post something everyone starts bashing on my person instead of dealing with my points. I mean, my arguments have absolutely nothing to do with myself, I was just trying to improve the system to be more accurate, something YOU would profit from while I wouldn't profit from (at least according to your narrow logic).
But instead you are just hating on my person. I don't understand this childish behaviour.
Not to defecate allover your meal but even my unranked winrates are astronomically high on some heroes that I don't even know how to play or frequently play. I am not in anyway disagreeing that you've improved your performance over the last few months though.
(watched that game you asked me to, you didn't do too bad but I would have gone phoenix vs sniper instead)
That's nowhere near 70%, stop bullshitting already.
76% winrate and 67% are nowhere near 70%. Yes, of course. Guess someone needs some math classes. :>
i don't get why everyone is hating on smaug. just give him a fucking break I can understand what he is trying to say and it makes sense. I also admire his dedication to mastering such unique heroes.
Where are you guys getting your info from! Also you're all using different data sets that you're manipulating in odd ways. Just using the scenarios tab I can see he's consistently been around 65% this month for both ranked and unranked
http://www.dotabuff.com/players/86717663/scenarios
Also using his bracket + ranked + patch gives similar results. He does in fact currently play ranked at about a 65% win rate hero depending.
I think smaug has a fetish. But i can't blame him, OD, phoenix and wyvern are nice looking heroes.
^
The only ones that win are smaug's lawyers with their exorbitant rates. Did you know that most lawyers charge for their time in 6 minute intervals?
I like how you're completely dodging the topic. I think we all got it that you want to keep the system as is. It's fine, I like to stay at the top rankings for Phoenix and OD :)
If you read read my first post, you would see that I don't want to keep the system as it is now as I stated it should only be ranked games counted, and for the record my first point I made was about your star legal defense team that draws up lawsuits against people you don't like on forums. You don't seem to read other peoples posts and you end up posting nonsense very often.
@havocbadger: A month of at 63% winrate for him, when he has such an enormous number of games on the hero far below that, just indicated he is having an upswing month and will inevitably (statistically speaking) have a lower winrate month.
"my first point I made was about your star legal defense"
So you're posting something COMPLETELY off topic relating to something that happened A FEW YEARS ago for absolutely no other purpose than to bash me. And then you're complaining that I wouldn't reply to that.
"and will inevitably (statistically speaking) have a lower winrate month."
This is completely destroying your own point where you tried to argue that winrate would not be statistical, but instead have a meaning. Also I just was on my longest winning streak ever within those 3 years of Dota 2 so you're posting nonsense.
Win rate does factor in, the reason being no matter how good you think you are, if you have less than 50% WR for example, it shows that you have below 50% chance(success) of winning. Because spamming a hero and playing non stop doesn't show how good you are. Every hero has a weakness and if you pick ur hero against counters and lose, that is your fault for being arrogant. Thus you can't say win rate is unnecessary.
In my opinion, there are many factors that weigh in, for example: a carry, your win rate (ability to win by picking that hero), Kda+HD at periodic interval, Last hit and GPM at periodic interval for carry.
If you're a support, the win rate, the assists, wards, number of deaths, etc etc at periodic interval.
Or something similar
Win rate weighs in, but it's hard to consider so many factor on so many roles. Every role is different hence no algorithm will work perfectly, only ones that tailor towards that role.
All these include level of difficulty, and maybe the duration like measure within the year, and use a 365 or 56 week cycle to measure performance. The number of games play is a prerequisite for this cycle to work, eligibility.
So if I have a history of maybe 50 games min as prerequisite over 2 years, and have played only 15 games in the last 56 weeks, then that should be taken into consideration and leave the others out.
It'll be hard to do.
Example, the data measures all games for hero score and remove the old games for more accurate fluctuations
I'll write more if you are interested I'm just using a style from a sports game to measure stats
(Holy crap some of you people dislike each other).
Difficulty is that each hero respectively does their own thing which YH touched on by pointing out the need for a different rating for carries and supports. But there are several other roles aside from that as well. For instance, lets say you're playing a tank (viz. Bristle, Axe). Depending on the game, reflecting the amount of damage you took divided by the number of deaths might more adequately represent how well you performed in the role of providing a cover (meat-shield) for your team to play around.
But the same won't apply for a hero more suited towards, say, a pushing role (Prophet being the example that comes to mind). There you might want to look at damage dealt to towers and other structures as opposed to the actual enemy team. Obviously not discount it entirely, but apply a negative modifier on it? Statistically, there are probably a ton of clever mathematical tricks you could do to try and evaluate skill. For example, just apply Cohen's D to two different measurements where appropriate to compare whether or not a particular result is statistically significant.
On a side note, win rate should apply, but I would recommend a slight alteration to the system there as well. We (that is, Dotabuff) currently calculates the relative effectiveness of hero's against each other. So, lets say you win a game as Huskar against Viper. Who basically rapes you cuz that's what his kit does to you. A win against a Viper should count more than if you win, let's say a Skywrath Mage. On a side note, I just looked at Huskar's page and it doesn't have Viperino on it. Shows you want I know. In any case, my point is overall the same.
idk, it's almost worth a topic in of itself if there would be legit serious discussion by people who knew statistics well.
The system is ridiculously inaccurate. For example it attests me a KDA of 2.50 with OD, while even if I count the games from the last 12 months (!!) I still got way higher KDA (2.7), not to mention the last 3 months (60% WR and 3.1 KDA in 107 matches) or the last month (75% winrate and 3.8 KDA in 55 matches). It still heavily punishes me for games that I lost 2 years ago. Similar to my Phoenix.
Apart from that I have proven that I can inflate my winrate (and thus my KDA) very easily by just playing unranked games. In fact, I just broke my record for the longest winning streak of the 3 years I am playing. After my last post in this thread I won 14 games with OD and lost 0.
"just indicated he is having an upswing month and will inevitably (statistically speaking) have a lower winrate month."
And this is just plain bullshit, I never had this Winrate over this long time EVER before, there is no statistic for that, so you can't speak statistically. You're inventing things.
Por favor, se conecte para postar comentários.
Hi,
I recently look at top 100 Dotabuff with Anti Mage, and I saw Eg.Sumail on it who is directly below me. He has 44% win rate with KDA of 2.69 with 67 games played. I wonder how someone like this gets to top 100? It makes me wonder how the system work. For example Captain America who is rank 53 or so plays 96 games with 72% win rate and nearly 6 kda gets similar hero score to him? Like how much hero score do you get per professional game win? It seems to be so bias and segmented.
Thanks