EPTTvZ Winrates with Mass Widow Mine - Page 7
| Forum Index > SC2 General |
|
YourAdHere
United States216 Posts
| ||
|
dsjoerg
United States384 Posts
On April 08 2013 12:47 petered wrote: A much better metric would be either percentage of army resources in mines Great idea, thanks! ![[image loading]](http://i.imgur.com/8Y75XGl.png) So here we are looking only at Master level TvZ games that made it to at least the 15:00 mark. And then we measure the resource value of widow mines in their army at that time, compared to the total resource value of their Active Army. Mostly this data reinforces things that "everybody knows" about widow mines -- they don't suck, they're not necessarily OP, T loves em vs Z, especially at the high level. That 10-15% row is interesting and we should keep an eye on it as more data rolls in. If I were a betting man (and I am!), I'd bet on that row remaining the highest winrate, but for it to move closer to the rest. | ||
|
plogamer
Canada3132 Posts
On April 08 2013 14:03 Orek wrote: I just hope OP dsjoerg or any other people reading this thread is not discouraged to make interesting threads in future due to negative feedback. Some people don't realize that 1 liner negativity discrediting OP's work does harm to the community by turning away potential posters doing interesting researches. Constructive criticism is fine, though. Natural selection, baby. (I just hope this 1 liner will turn you off from coddling the community) On April 08 2013 14:10 dsjoerg wrote: Great idea, thanks! So here we are looking only at Master level TvZ games that made it to at least the 15:00 mark. And then we measure the resource value of widow mines in their army at that time, compared to the total resource value of their Active Army. Mostly this data reinforces things that "everybody knows" about widow mines -- they don't suck, they're not necessarily OP, T loves em vs Z, especially at the high level. That 10-15% row is interesting and we should keep an eye on it as more data rolls in. If I were a betting man (and I am!), I'd bet on that row remaining the highest winrate, but for it to move closer to the rest. How does this show that widow mines don't suck? And what's with the evasive-as-a-slippery-eel language? "They're not necessarily OP". No, your results don't show that widow mines are not necessarily OP either. Why? Read the damn thread. What you just showed was an exercise in wasting everypony's time. | ||
|
tenklavir
Slovakia116 Posts
Any of you geniuses arguing for the validity of this kind of nonsense want to guess what that means? Edit: hint - it means exactly what this entire thread means. Nothing. The problem with threads of this kind like this is exactly reflected in BronzeKnee's comments of "people can draw their own conclusions" etc. (I'm paraphrasing, before you jump at my throat saying TAHTS NOT WUT I SED). You cannot draw any reasonable conclusion from any of this, and it seems that most lack the mathematical background (or common sense) to understand why. It just starts a whine-fest shitstorm that accomplishes nothing but getting people pissed off over meaningless data. | ||
|
aksfjh
United States4853 Posts
On April 08 2013 14:22 tenklavir wrote: A quick basic linear regression analysis of Master TvZ Widow Mine % of Active Army @ 15:00, testing Win % as a function of WM%, yields an Rsq = 0.097. Any of you geniuses arguing for the validity of this kind of nonsense want to guess what that means? Zerg don't brush their teeth regularly? | ||
|
phodacbiet
United States1739 Posts
| ||
|
Entirety
1423 Posts
On April 08 2013 14:22 tenklavir wrote: A quick basic linear regression analysis of Master TvZ Widow Mine % of Active Army @ 15:00, testing Win % as a function of WM%, yields an Rsq = 0.097. Any of you geniuses arguing for the validity of this kind of nonsense want to guess what that means? Merely 9.7% of the variation in winrate can be attributed to a variation in the % of resources devoted to Widow Mines. In other words, the correlation is lousy. | ||
|
tenklavir
Slovakia116 Posts
On April 08 2013 14:28 Entirety wrote: Merely 9.7% of the variation in winrate can be attributed to a variation in the % of resources devoted to Widow Mines. In other words, the correlation is lousy. Textbook answer! A+  | ||
Meatex
Australia285 Posts
Think about hellions produced when asking if a terran will mech or go bio after - most common is 6 into bio but sometimes terrans would go 8 or 10 but aren't committing to factory play | ||
|
spalding
95 Posts
On April 08 2013 14:03 Orek wrote: I just hope OP dsjoerg or any other people reading this thread is not discouraged to make interesting threads in future due to negative feedback. Some people don't realize that 1 liner negativity discrediting OP's work does harm to the community by turning away potential posters doing interesting researches. Constructive criticism is fine, though. sometimes blunt responses get the point across very well though and from reading this thread it's clear that a lot of people have no clue whatsoever about statistics and if you open a thread to represent this data, one may expect some evidence in the form of math (hint : variance is a huge concern with such a limited sample size and so is selection bias) combined with logic to back up your thoughts/conclusion. | ||
|
Orek
1665 Posts
On April 08 2013 14:10 plogamer wrote: Natural selection, baby. (I just hope this 1 liner will turn you off from coddling the community) Very convincing when the word comes from a guy whose last thread was closed. I guess you are right. Some bad threads deserve to die. Coddling does harm, too it seems. Maybe OPs have to try harder to bring better contents. | ||
|
dsjoerg
United States384 Posts
On April 08 2013 14:10 plogamer wrote: How does this show that widow mines don't suck? If widow mines sucked badly enough, then win% would decrease as WM% increased. They're not so bad that you lose by using them. Even in large quantities like 15% of army strength. No, your results don't show that widow mines are not necessarily OP either. I thought of "OP" as some unit that, if you got a lot of it, you would have an EZ win. By that definition, you'd expect people with armies that were 25 and 30% widow mine to have higher win% than they do. But clearly not everyone thinks of "OP" this way. | ||
|
plogamer
Canada3132 Posts
On April 08 2013 14:35 Orek wrote: Very convincing when the word comes from a guy whose last thread was closed. I guess you are right. Some bad threads deserve to die. Coddling does harm, too it seems. Maybe OPs have to try harder to bring better contents. Yep, my thread got closed so I must be defensive and blame everyone but my hilariously lousy post. Or I can be petty and sling personal insults. Nah, I have more fun things to do than go through someone's post history. How lame! | ||
|
Myrddraal
Australia937 Posts
On April 08 2013 12:16 BronzeKnee wrote: The fact the Mutalisk guy is the OP doesn't change anything regarding the conclusions of what was said or done. I could be the Mutalisk guy and the OP, still nothing would have changed. And I didn't backtrack to attack his first post, re-read the whole chain. Seriously your posting frustrates me to no end, stop trying to change the focus when you are blatantly wrong, I said nothing about the conclusions, I said you look like a fool for attacking someone who was making reasonable arguments. I read the whole exchange, and what it boiled down to was you were trying to dispute arguments that you claimed opterown made, which he never did, and when he shut you down you instead attacked his first post saying that he was trying to discredit the OP and that he should be more supportive like he was of "Mutalisk guy", which is itself is stupid because this kind of criticism is actually the most beneficial discussion that the OP could hope for, because if his statistics and conclusion are in fact solid then they only become more credible if they can stand up to such criticism. Discussion is good, but when you get over defensive and start reading into things too much you kind of ruin any chance for a decent discussion. I could be the Mutalisk guy and the OP Actually considering how defensive you are being over every little criticism of the OP it wouldn't surprise me if you were. | ||
|
Drowsy
United States4876 Posts
| ||
|
Novacute
Australia313 Posts
| ||
|
tenklavir
Slovakia116 Posts
On April 08 2013 14:42 dsjoerg wrote: If widow mines sucked badly enough, then win% would decrease as WM% increased. They're not so bad that you lose by using them. Even in large quantities like 15% of army strength. But that is exactly what happens (win % decreases as WM% increase) using the data that you just posted. The admittedly lousy correlation (Rsq = 0.097) between Win % and WM% is negative (Win % = 60.29 - 0.1429(WM%)). Does this mean we now conclude that the higher % of your army makeup being widow mines means your will lose? What if I left out the Rsq and never mentioned that the correlation was terrible? Would you have believed me? Does it even make sense? Of course not, and this is exactly what I'm talking about. This entire discussion is worthless and people need to start understanding why you cannot draw any reasonable conclusion from such meaningless data. | ||
|
Weryeery
288 Posts
| ||
|
phodacbiet
United States1739 Posts
On April 08 2013 14:10 dsjoerg wrote: Great idea, thanks! So here we are looking only at Master level TvZ games that made it to at least the 15:00 mark. And then we measure the resource value of widow mines in their army at that time, compared to the total resource value of their Active Army. Mostly this data reinforces things that "everybody knows" about widow mines -- they don't suck, they're not necessarily OP, T loves em vs Z, especially at the high level. That 10-15% row is interesting and we should keep an eye on it as more data rolls in. If I were a betting man (and I am!), I'd bet on that row remaining the highest winrate, but for it to move closer to the rest. Umm.. this doesnt prove widow mines OP or not OP. It just shows that terrans are winning in general (which is fine). You cant just pick a unit at random and say its the cause of increase win rates... | ||
|
Novacute
Australia313 Posts
On April 08 2013 14:59 phodacbiet wrote: Umm.. this doesnt prove widow mines OP or not OP. It just shows that terrans are winning in general (which is fine). You cant just pick a unit at random and say its the cause of increase win rates... Is OP going to have a heart attack when he does this exact thing with Marines? | ||
| ||
