| Forum Index > BW General |
|
old times sake
165 Posts
On December 21 2009 10:52 SuperJongMan wrote: Do these people even play starcraft? No, they don't. Everyone who disagrees with you is a deluded newbie. They weren't talking about Pro gaming or anything in this thread. It was an optical illusion. edit: Good post BTW. When you disagree with people it's helpful to just flat out insult them in a 1-liner and offer nothing else. You should write a guide on how to contribute to a 46-page thread. | ||
|
SuperJongMan
Jamaica11586 Posts
| ||
|
Xiphos
Canada7507 Posts
| ||
|
Kazius
Israel1456 Posts
On December 21 2009 14:54 Xiphos wrote: Guys, just play better. Zerg did it by sniping HTs left and right. You guys have to understand that HT sniping are HARD! Its very hard to click precisely on the HT templar (its such a small units) without clicking on the ground or another units while macroing, manage your resources AND control your army around. The Zergs have found their 'best' way of beating Protoss, now its Protosses' turn to return the favor by twitching their own way of playing. You didn't watch TL Arena II when Cloud got his HTs sniped by a D player. Dude, you have no idea how easy it is. | ||
|
Un Gato
United States123 Posts
The stats ended... Anyways, wouldn't the standard deviation of a binary data set be essentially meaningless? That is, considering that "win" and "lose" aren't really numerical data values, but rather nominal values coded to numbers, the standard deviation of those values produces a number that doesn't represent anything useful. We could have coded "win" to .1 and "lose" to 0 and still had a standard deviation, but it would also still mean nothing. The standard deviation, as I'm sure you know, is a measure of the average amount a set of numbers vary from the average value (the mean), though it is not that average variation itself. However, when the average value means nothing neither does the SD. As zulu_nation8 pointed out .5921 is not the average value of the wins and losses. In terms of our actual data points, which are "win" and "lose," there is no in-between where anything other than "win" and "lose" can mean anything. In fact, the only reason that happens to have been the percentage of games won is because we coded the nominal values to one and zero so that the sum of the values ended up being the number of wins. If wins had been 1 and losses -1 then we would have avoided the coincidence. In short, the standard deviation should only be used for interval or ratio data, where the distance between values has meaning. The only thing that matters is the percentage of games won, so we should use a one proportion z test: ![[image loading]](http://upload.wikimedia.org/math/6/2/d/62d59e17e2e75373e1b0fd7077ec3622.png) z = (.5921 - .5) / sqrt((.5921 * .4079) / 885) = 5.5752 It is significant, that is clear, but the math in here got pretty off track except for Matrijs on page 33 and Traveler on page 40, both of whom were either ignored or misread. And I'm looking at you when I say that Motbob: Motbob: Heh. That's a fair question but if you look a few pages back a bunch of us did z-tests. Unfortunately it didn't convince anyone You'd done the wrong type of test for the data you had. On page 38 Black Gun had pointed out that the data was not normal, and zulu_nation8 repeated it a page later, but you ignored that. As an econ major that's not so good. Though to be fair Traveler also assumed the data was normally distributed, which ironically would have made his test impossible if it were true. mIniAtURe on 46 was getting at the problem, but didn't quite grasp it. ---------------------------------------------------------------------------------------------------------------------------------------------- Stats aside, as much as I dislike anything that handicaps Zerg considering that I am a Zerg player, I do agree that something needs to change, though I'm not sure what might be changed that wouldn't adversely impact the other matchups. My preferred change would probably have to do with improving corsairs given that corsairs are so rare in anything other than PvZ. Perhaps lower the cost and time of the Disruption Web upgrade? I'd like to see that used more. Though that might just lead to them being used in PvT...maybe a slightly faster attack rate, or larger splash radius? | ||
|
Severedevil
United States4839 Posts
| ||
|
Severedevil
United States4839 Posts
On December 21 2009 17:48 Un Gato wrote: Stats aside, as much as I dislike anything that handicaps Zerg considering that I am a Zerg player, I do agree that something needs to change, though I'm not sure what might be changed that wouldn't adversely impact the other matchups. My preferred change would probably have to do with improving corsairs given that corsairs are so rare in anything other than PvZ. Perhaps lower the cost and time of the Disruption Web upgrade? I'd like to see that used more. Though that might just lead to them being used in PvT...maybe a slightly faster attack rate, or larger splash radius? My personal PvZ frustration is that Zerg can get +2 air carapace with only a lair, whereas Protoss has to buy an expensive and not very useful Fleet Beacon for +2 air weapons. (Which in turn makes muta/scourge very hard to combat with Protoss air, allowing nasty templar snipes/muta backstabs and drastically reducing Protoss harassment.) If Stargate allowed +2 air weapons, and Citadel allowed +2 ground weapons, I'd be happy. (Or improve the Fleet Beacon as a choice - buff scouts, reduce the cost on Fleet Beacon + Scout Speed + D-Web.) | ||
|
Un Gato
United States123 Posts
| ||
|
Un Gato
United States123 Posts
Lower scout build time, perhaps shave the cost. Make this build faster to implement so that a good Zerg can't just hydra his way out of it damage-free. | ||
|
Un Gato
United States123 Posts
On December 21 2009 14:38 SuperJongMan wrote: lol, fuckn deluded newbie. Well, ten thousand posts and a jackass to boot. TeamLiquid has clearly benefited from your presence immensely. | ||
|
Foucault
Sweden2826 Posts
On December 21 2009 14:54 Xiphos wrote: Guys, just play better. Zerg did it by sniping HTs left and right. You guys have to understand that HT sniping are HARD! Its very hard to click precisely on the HT templar (its such a small units) without clicking on the ground or another units while macroing, manage your resources AND control your army around. The Zergs have found their 'best' way of beating Protoss, now its Protosses' turn to return the favor by twitching their own way of playing. LOL it's not hard man, seriously. While everyone can misclick, it's relatively easy to do. | ||
|
AzureEye
United States1360 Posts
| ||
|
FieryBalrog
United States1381 Posts
On December 21 2009 14:54 Xiphos wrote: Guys, just play better. This is about the pro scene. I think they know about that idea already. | ||
|
Foucault
Sweden2826 Posts
On December 21 2009 18:48 FieryBalrog wrote: This is about the pro scene. I think they know about that idea already. Yeah, we're talking about players who practice 8+ hours/day. Shit is still imbalanced and it's not because they play bad or don't practice enough. One thing with the new maps is that alot of them are quite big. This is good for ZvT (secure many expos due to distance), and ZvP (in this matchup zerg is also able to secure a gazillion bases due to map distance). This is also why Protoss owns terran so hard on big maps like Andromeda etc PvZ tends to be more balanced on smaller maps. | ||
| ||
StarCraft 2 StarCraft: Brood War Dota 2 League of Legends Counter-Strike Heroes of the Storm Other Games summit1g7811 FrodaN2267 shahzam518 mouzStarbuck456 B2W.Neo323 KnowMe294 ToD250 Livibee120 QueenE93 elazer88 Mew2King61 ViBE54 Trikslyr52 SteadfastSC30 CosmosSc2  22 |
|
Replay Cast
GSL
Shopify Rebellion Sundays
Spirit vs Mixu
BSL
Ret vs TerrOr
DragOn vs StRyKeR
Replay Cast
Afreeca Starleague
Shine vs Rush
WardiTV Weekly
Monday Night Weeklies
Sparkling Tuna Cup
Afreeca Starleague
Light vs EffOrt
[ Show More ] |
|
|
EPT
