Season 3 Ladder Map Changes (Official) - Page 27

On July 26 2011 12:06 Jonny55 wrote:
The real reason they took out Meta was because it had no rocks ... jk jk

You will be missed Meta T_T, good riddence to the other removed maps though

On July 26 2011 11:45 xdeacon wrote:


Just wanted to point out for the thread that while TLPD is totally awesome for what it is, you can't really get meaningful map balance statistics out of it because of the small numbers involved.

Check out the binomial probability distribution: http://en.wikipedia.org/wiki/Binomial_distribution. For a coinflip (i.e. perfectly balanced map 50%), one standard deviation = 1/sqrt(number of games).

Meta 1.1 lists 296 games played. If I flip a coin 296 times one standard deviation will be about 6%. This means TLPDs 53.7% ZvT win ratio could be just as likely representing a sample that is 60% (hugely favored for Zerg) or 48% (slightly favored for Terran) with a 1/3 probability that the margin could be larger.

This is why it is so important to have a humongous sample like blizzard should have (in theory) because the variance shrinks by the square root of total games (1% sigma = 10000 games!). This also has the additional effect of washing out the effect of steep skill differences that appear in tourney matchups that get reflected in TLPD.

I hope this also reinforces the point that just because something 'feels' unbalanced anecdotally (I for one have awful TvZ on Metalopolis!), it really requires an unbiased large sample to determine accurately.

tldr: just because a number sounds convincing, don't believe it!

On July 26 2011 12:01 stormfoxSC wrote:

Uh, TLPD only uses a tiny sample known as professional tournaments. Blizzard has not only that, but also literally every ladder game played by any Masters or Grandmasters around the entire world. It's ridiculously laughable that you'd take the sample size that's only a fraction of what Blizzard uses.

On July 26 2011 10:21 Yoshi Kirishima wrote:
Also strange they removed Gutterhulk from 2v2 pool, I thought it was pretty good. I guess not xD

On July 26 2011 10:21 Yoshi Kirishima wrote:
Oh yeah, about Metal:



The predictability part is SO true. For example, Boxer's 2 rax proxy is very strong on that map without close, because you can scout the overlord if he's close by air and then know he is cross position, but even if he is still close air and simply didn't send his overlord or you didn't see it, then it is still a strong proxy.

On July 26 2011 12:01 stormfoxSC wrote:

Uh, TLPD only uses a tiny sample known as professional tournaments. Blizzard has not only that, but also literally every ladder game played by any Masters or Grandmasters around the entire world. It's ridiculously laughable that you'd take the sample size that's only a fraction of what Blizzard uses.

On July 26 2011 11:43 DeltruS wrote:
I looked through the thread and was shocked that there were no map pics.

Straight from the liquipedia, here you go. (note: might not be up to date)


It looks like blizzard has gone with the philosophy of being able to get 3 bases fairly easily. The fourths are all kinda awkward to get but not nearly as awkward as junk yard, for example.

I really like Antiga_Shipyard. The bases look really dense, so that might favor T/P with tanks/Colo, but the zerg player also has a ton of options to mass expo.

I might even be able to play ladder without having to play a map that I hate.

On July 26 2011 12:18 Toadvine wrote:


The fact that Blizzard's data suggests TDA is somehow Zerg favored is proof enough for me that it's completely worthless.