| Forum Index > Dota 2 General |
IRC chatter should remain in IRC - http://webchat.quakenet.org/?channels=tl.dota2 Posts that relate to topics with their own thread, such as in News, Tournaments or Strategy should go in those threads. | ||
|
Dead9
United States4725 Posts
there's a huge, huge gap in skill level between 1700 and 1900 it's like high masters versus high diamond in terms of skill level, or B+ versus C+; you just can't make up for the gap in mechanics, positioning, map awareness etc by simply having a slightly better idea of what you're doing mmr is a fairly accurate to keep track of someone's skill level in their average game, though it does result in a lot of flaming and arguing which kind of sucks | ||
|
BurningSera
Ireland19621 Posts
On January 23 2013 18:08 Itsmedudeman wrote: I can speak from personal experience because I thought in a similar fashion to you. I am a much better player now because I know that getting to a high rating as support is many things more than simply relying on your carries to farm and carry the game for you and that being your goal. All I'm saying is that warding consistently does not make you better at support than someone who doesn't. I never say warding consistently, what i meant was providing critical visions at different phase of the game is the basic of all basics in playing support. You see most people drop a ward on the map and say that's the support. | ||
|
Yoshi-
Germany10227 Posts
On January 23 2013 18:27 lazyitachi wrote: That post confirms that he has no understanding of statistical sampling. Considering that number of plays and rating level seems quite well represented, don't think anyone can actually argue much about the validity of the result. Are you talking to me? If so explain me how an extremely "biased" sample size of 24k could represent over 3 million active players. | ||
|
BurningSera
Ireland19621 Posts
On January 23 2013 18:28 Dead9 wrote: a 1900 player would be better than a 1700 player at supporting off raw mechanics alone 95% of the time there's a huge, huge gap in skill level between 1700 and 1900 it's like high masters versus high diamond in terms of skill level, or B+ versus C+; you just can't make up for the gap in mechanics, positioning, map awareness etc by simply having a slightly better idea of what you're doing mmr is a fairly accurate to keep track of someone's skill level in their average game, though it does result in a lot of flaming and arguing which kind of sucks If only dota is not a team game. No it doesnt work like that. | ||
|
Gheizen64
Italy2077 Posts
| ||
|
Dead9
United States4725 Posts
i dunno how else to put this, the skill gap between 1700 and 1900 in hon is huge seriously massively huge like five 1900 players would crush five 1700 players if the 1700 team picked all 10 heroes 1900 would be like upper-middle of "very high" bracket, 1700 would be like upper mid of "high" bracket there's not much you can do to make up for it | ||
|
nojok
France15846 Posts
| ||
|
Dead9
United States4725 Posts
edit: basically, mmr might not be perfect, but a player at the 90th percentile is going to be better than a player at the 70th percentile pretty much always edit2: if you're wondering why a 1900 player is pretty much always better than a 1700 player in hon, here's a graph by shinbi that might explain a bit. it's a little outdated but accurate enough anyway http://i.imgur.com/NDkxH.png | ||
|
lazyitachi
1043 Posts
On January 23 2013 18:31 Yoshi- wrote: Are you talking to me? If so explain me how an extremely "biased" sample size of 24k could represent over 3 million active players. It was addressed to someone else but seem apt in your case too. The same reason that election polling works well. I don't see the pollers go to everyone's house and get their opinion. As long as the subset is representative of the population. Given the distribution of players, seems that there is quite a good representative at each level. Biased would be all the tight people who don't want their DBR revealed create multiple accounts to vote for private rating. Even with margin of error, the result is too obvious. So... if you are unhappy I guess... write to Dotabuff? | ||
|
Yoshi-
Germany10227 Posts
On January 23 2013 18:51 lazyitachi wrote: It was addressed to someone else but seem apt in your case too. The same reason that election polling works well. I don't see the pollers go to everyone's house and get their opinion. As long as the subset is representative of the population. Given the distribution of players, seems that there is quite a good representative at each level. Biased would be all the tight people who don't want their DBR revealed create multiple accounts to vote for private rating. Even with margin of error, the result is too obvious. So... if you are unhappy I guess... write to Dotabuff? But the subset is not a representative of the population, that why your logic fails miserable. If we look at the stats dotabuff gives us, they say that 50% of the players are before 1200 and the rest after 1200. But the poll result states that over 20k of the voters have an higher rating than 1200. That are like 80% of the total. Arguably the median is not exactly at 1200, but even than the amount of voters with a high rating are much higher than the ones with a low. And then we still have the point that someone who uses Dotabuff most probably have a higher interest that it adds DBR, than someone who never/rarely visit it. | ||
|
Itsmedudeman
United States19229 Posts
| ||
|
Gheizen64
Italy2077 Posts
| ||
|
PrinceXizor
United States17713 Posts
| ||
|
SirMilford
Australia1269 Posts
Also this will be interesting for me because i always wonder if games are fair in the australia server. Sometimes they seem to be very biased maybe due to low player pool at certain times of night. | ||
|
Mithhaike
Singapore2759 Posts
On January 23 2013 18:46 Dead9 wrote: that's why mmr measures your average level of play edit: basically, mmr might not be perfect, but a player at the 90th percentile is going to be better than a player at the 70th percentile pretty much always edit2: if you're wondering why a 1900 player is pretty much always better than a 1700 player in hon, here's a graph by shinbi that might explain a bit. it's a little outdated but accurate enough anyway http://i.imgur.com/NDkxH.png Burningsera he speaks the truth. The difference between 1900 & 1700 is huge in HoN. Chances are much greater that the 1900 player is better in all aspect of the game than the 1700 player regardless of what he plays, carry/mid/support. Unless we talk about outliers who only play 1hero | ||
|
babishh
Canada965 Posts
oh well, shall the flamefest begin. woops i thought i posted on dbr thread, seeing all you discussing about it.. | ||
|
duoform
Spain5180 Posts
| ||
Firebolt145
Lalalaland34505 Posts
On January 23 2013 21:15 duoform wrote: Valve is not allowing people to see their stats because it increases flame and whatnot but they are allowing DOTABUFF to do it? DOTABUFF said the launch is on Friday. I don't get it... either you hide it everywhere or you better fucking build a proper MMR. What's the deal with Valve lately? Seriously? Dotabuff is making their own MMR system which does not piggyback Valve's MMR system at all. The two are not connected in any way. | ||
|
cilinder007
Slovenia7251 Posts
| ||
|
duoform
Spain5180 Posts
| ||
| ||
StarCraft 2 StarCraft: Brood War Britney Dota 2  43478Calm  20262Shuttle 3087Horang2 1088Hyuk 907firebathero  285Rush 275Larva 197actioN 133Barracks 85[ Show more ] Counter-Strike Heroes of the Storm Other Games |
|
Replay Cast
Rogue vs Cure
Big Brain Bouts
Garitos vs ArT
Lambo vs Percival
TriGGeR vs ByuN
Sparkling Tuna Cup
OSC
Shopify Rebellion Sundays
Mixu vs TBD
Spirit vs TBD
Clem vs TBD
Patches Events
Afreeca Starleague
Soma vs Shuttle
BeSt vs Rush
WardiTV Weekly
Afreeca Starleague
Leta vs ggaemo
Jaedong vs Queen
GSL
[ Show More ] |
|
|
EPT
