EPTWhy not make more dramas and upboat more just to get le karma XDXD and let TL do their thing.
What you don't know about statistics - Page 7
| Forum Index > SC2 General |
|
Nokshalees
United States120 Posts
Why not make more dramas and upboat more just to get le karma XDXD and let TL do their thing. | ||
|
See.Blue
United States2673 Posts
| ||
|
Destroyr
Germany299 Posts
| ||
|
taddel
Germany13 Posts
To the people who think it's common sense what he wrote I have to say that there are plenty of posts on team liquid which make exactly the mistakes he mentioned. | ||
|
Azarkon
United States21060 Posts
On May 04 2012 09:00 Cyberonic wrote: Drabzalver - Fallacy: winrates indicate balance Yes and no, for the most part, they do not indicate balance, rather, they indicate balance shifts. It's so trivial to see that it boggles me that people don't figure this out on their own. Assume that race X was actually underpowered last month and is balanced this month. This means that the X players who actually qualified last month and stayed in tournaments are actually better than the Y and Z players, therefore, now that it gets balanced, as they are overal better, they start smashing Y and Z because they are better, thereby suddenly making the graph appear as X has been 'overbuffed' or whatever else while simply the few X players that were around in the scene were better. This will continue on until the mediocre Y and Z players who were more so carried by their race than the X players get weeded out. ... The only way to find out absolute balance is to get a random pool of pros, force them to play random, and have a round robin tournament to ensure that everyone plays the same amount of games. Very unfeasible to get enough games with that for a reasonable sample size. It's worse than this. Your definition of balance shifts is that it varies with the distribution of better players in a month's tournaments. But it's not just players that are shifting, it's their meta-game. Say Naniwa figures out a new mothership rush style tomorrow that Zerg has no answer for. The win rate for P vs. Z during that month is going to be lopsided in favor of Protoss because all the Protoss pros are going to switch to it and crush their Zerg opponents. But then a few weeks later, Stephano figures out a three hatch mass ultra style that beats the crap out of the mothership rush. Then the pendulum swings right back and those Protoss who practiced their arses off on how to rush motherships is going to get creamed until they readjust. Because of this, getting a random pool of pros, forcing them to play random, and having a round robin tournament to ensure that everyone plays the same amount of games is still isn't going to give you absolute balance - because the pros are going to play each race the way the meta-game tells them to play, which gives you the balance of the current meta-game but not the balance of the game. A lot of obvious imbalances are detectable through observing lopsided match-ups and thinking through the game logic to see that the lopsidedness is due to game design and not player skill, but there is no way to figure out the absolute balance of the game without solving the game, and thereby knowing the optimal strategies in each match-up. | ||
|
Chargelot
2275 Posts
| ||
|
upperbound
United States2300 Posts
| ||
|
CluEleSs_UK
United Kingdom583 Posts
Good post, well written, statistics very rarely have useful connotations in the world of Starcraft, and are reserved only for people with bogus arguments they're trying to defend. | ||
| ||
