EPTMap stats: testing whether a map is balanced - Page 3
| Forum Index > BW General |
CaucasianAsian
Korea (South)11599 Posts
| ||
|
mitsy
United States1792 Posts
| ||
|
Mortality
United States4790 Posts
In ym mind, these approaches miss something: there is a difference between imblance and "tricky." So when is a map "tricky" versus imbalanced? In my mind, it's "tricky" if there are only a couple of viable approaches to playing the map, which require some minimal degree of skill to pull off. For instance, ZvP on Estrella. Is it imbalanced or is it tricky? From my experiences, nobody I can't beat consistently on land maps ZvP is someone I can't beat consistently ZvP on Estrella. Granted, I have never played someone whose PvZ is equal in skill to my ZvP on Estrella, but it should be telling that it is winnable. By comparison, ZvP on Gorky island takes all my energy. I have never lost on Gorky, but I refuse to ever play it again because I know that even players who I can whomp all over have a good chance against me there. The difference, of course, is the easy to expand to tiny island above the main. P with 2 gases means lots of upgraded corsairs, and by taking other mains around the map, the Protoss can wear me down and keep me from playing a centrally dominated late game as I can on Estrella. | ||
Bill307
Canada9103 Posts
On December 27 2005 07:05 oneofthem wrote: hey, i don't think you can set pwins to .5 except in mirror matchups. to evaluate the empirical fairness of a map you must get the map-neutral winning percentage of each MU. pwin is our hypothesized win ratio, so that is not a problem. When considering every game, the sample size is bigger, and a bigger sample size is always good to make quality interpretations. Actually that it not necessarily true. Having such a large number of lower-level PGT players in our sample makes our analyses less applicable to higher-level PGT players. Ultimately it depends on who your target population is. | ||
|
Mortality
United States4790 Posts
As interesting as it may be to look at the actual statistics, they don't quite tell the full story. There are too many factors that are not quantifiable. | ||
|
Knickknack
United States1187 Posts
My second question is what kind of system do the pro leagues use for picking maps? Do they just have them made and hope for the best? Do they test them? | ||
|
panschk[FP]
Germany148 Posts
| ||
| ||
