|
On March 16 2012 14:00 Mr. Nefarious wrote:
From a logic perspective, it's difficult to glean as much as we'd like from the best of the best. The problem when comparing the highest level players is that they frequently have something they do that is MUCH better then the average player, such as MMA's multitasking or DRG's overall macro and mechanics. Why not try to look at a group of players who are supposedly much closer in "skill"? I decided to look at a group of Code A players that while still are top level players, don't typically have a personal skill far beyond what we see from other high level pros.
More interesting statistics from TLPD:
Zerg:
Lucky
vT: 16-21 (43.24%)
vZ: 7-6 (53.85%)
vP: 13-6 (68.42%)
Terious
All: 16-22 (42.11%)
vT: 8-9 (47.06%)
vZ: 1-4 (20.00%)
vP: 7-9 (43.75%)
YuGioH
All: 37-38 (49.33%)
vT: 17-20 (45.95%)
vZ: 11-12 (47.83%)
vP: 9-6 (60.00%)
July
All: 63-51 (55.26%)
vT: 29-19 (60.42%)
vZ: 7-12 (36.84%)
vP: 27-20 (57.45%)
Terran:
GanZi
All: 45-43 (51.14%)
vT: 21-23 (47.73%)
vZ: 19-8 (70.37%)
vP: 5-12 (29.41%)
Keen
All: 45-42 (51.72%)
vT: 12-18 (40.00%)
vZ: 12-9 (57.14%)
vP: 21-15 (58.33%)
Happy
All: 41-32 (56.16%)
vT: 18-21 (46.15%)
vZ: 15-3 (83.33%)
vP: 8-8 (50.00%)
MKP
1v1 Record:
All: 135-90 (60.00%)
vT: 62-50 (55.36%)
vZ: 39-21 (65.00%)
vP: 34-19 (64.15%)
Ryung
All: 67-52 (56.30%)
vT: 30-19 (61.22%)
vZ: 22-15 (59.46%)
vP: 15-18 (45.45%)
Protoss:
Ace
All: 38-52 (42.22%)
vT: 21-24 (46.67%)
vZ: 13-17 (43.33%)
vP: 4-11 (26.67%)
Alicia
All: 35-38 (47.95%)
vT: 17-21 (44.74%)
vZ: 13-10 (56.52%)
vP: 5-7 (41.67%)
Brown
All: 13-11 (54.17%)
vT: 4-5 (44.44%)
vZ: 5-0 (100.00%)
vP: 4-6 (40.00%)
Inca
All: 55-47 (53.92%)
vT: 21-17 (55.26%)
vZ: 12-22 (35.29%)
vP: 22-8 (73.33%)
JYP
All: 52-51 (50.49%)
vT: 7-26 (21.21%)
vZ: 27-15 (64.29%)
vP: 18-10 (64.29%)
Interestingly enough, the trend continues down through Code A. The "average" professional Z/P players have a hell of a hard time vT with very few even breaking 50%. On the flip side, the "average" professional T players are all well above 50%, with Happy at a simply unbelievable 83% vZ.
Why is MKP in this group of players? He has spent far more time in Code S then he has in Code A and yet you have him in the Code A group?
|
On March 16 2012 13:35 Whatson wrote:Show nested quote +On March 16 2012 12:58 SafeAsCheese wrote:
After ghost changes, MMA is going to lose more. LOL somebody doesn't watch MMA TvZ games. If you skimmed through a couple of his games, you'll see that MMA 99% of the time does not use Ghosts even against a ling-broodlord-infestor composition. A Ghost change will have 0 impact on a player that has never used Ghosts in the past, and has always stuck with mass drops and vikings. Even Idra in his IEM casting of Zenio vs MMA said in the beginning that MMA never uses Ghosts and just out multitasks his opponents lategame. At least try to do some research before making stupid statements like these...
Yes, and eventually that won't work for him anymore, there will be Z's who don't get him get potshots with vikings or get out of position and let 30 marines kill all their brood lords.
MMA almost always attempts to end the game in the mid-game, if he is able.
|
On March 16 2012 14:13 SafeAsCheese wrote:Show nested quote +On March 16 2012 13:35 Whatson wrote:On March 16 2012 12:58 SafeAsCheese wrote:
After ghost changes, MMA is going to lose more. LOL somebody doesn't watch MMA TvZ games. If you skimmed through a couple of his games, you'll see that MMA 99% of the time does not use Ghosts even against a ling-broodlord-infestor composition. A Ghost change will have 0 impact on a player that has never used Ghosts in the past, and has always stuck with mass drops and vikings. Even Idra in his IEM casting of Zenio vs MMA said in the beginning that MMA never uses Ghosts and just out multitasks his opponents lategame. At least try to do some research before making stupid statements like these... Yes, and eventually that won't work for him anymore, there will be Z's who don't get him get potshots with vikings or get out of position and let 30 marines kill all their brood lords. MMA almost always attempts to end the game in the mid-game, if he is able.
Are you kidding me? MMA practically invented macro marine tank medivac AND fast third...
|
On March 16 2012 14:10 MONXY FIST wrote:
Why is MKP in this group of players? He has spent far more time in Code S then he has in Code A and yet you have him in the Code A group?
With him removed, the argument still stands on its own just fine. So many people are arguing the individual player card and its just illogical. If a vast majority of pros have a larger than average win rate in a certain matchup, it is pretty damn obvious that all of them are not especially talented in that matchup, rather that they are somehow favored in that matchup. This can clearly been seen with the "average" Code A Terrans being in the 55-60% WR vs Z and the naturally talented in that matchup ones being 70-80%. Compare with the average Code A zerg being in the 40-50% range and the best of the best sitting at 64%. It even applies to Protoss as well, most of their winrates vs T are abysmal with many of them doing extremely well in the other two matchups with 70%+.
|
whoever said the ghost nerf will affect MMA haha he doesn't even use them...MVP was the Ghost King of Terran.
|
On March 16 2012 14:13 SafeAsCheese wrote:Show nested quote +On March 16 2012 13:35 Whatson wrote:On March 16 2012 12:58 SafeAsCheese wrote:
After ghost changes, MMA is going to lose more. LOL somebody doesn't watch MMA TvZ games. If you skimmed through a couple of his games, you'll see that MMA 99% of the time does not use Ghosts even against a ling-broodlord-infestor composition. A Ghost change will have 0 impact on a player that has never used Ghosts in the past, and has always stuck with mass drops and vikings. Even Idra in his IEM casting of Zenio vs MMA said in the beginning that MMA never uses Ghosts and just out multitasks his opponents lategame. At least try to do some research before making stupid statements like these... Yes, and eventually that won't work for him anymore, there will be Z's who don't get him get potshots with vikings or get out of position and let 30 marines kill all their brood lords. MMA almost always attempts to end the game in the mid-game, if he is able.
You're talking with zero sense and entirely just fanboyism rage. Cover what you said up all you want, you are still wrong
|
|
|
On March 16 2012 11:14 neoghaleon55 wrote:Edit 1: Show nested quote +On March 16 2012 11:24 Mentalizor wrote:
When you "only" have a 50 game sample (DRG's ZvT is 50games) small mistakes, missmicroes can easilly cost 2-5 games. if just 5 games a lost due to mistakes, that will alter your statistics by 10% which is pretty much. Get bigger samples before comparing statistics. They are just not viable to look at. The optimal minimum sample size is 20. Above 20, the n value does not relevantly contribute over all (n-1) to the statistics. I'm sure you remember from your AP stats class and college. The statistics presented in the OP are greater than 20 sample sizes and thus are relevant. Edit 2: Maybe presenting that article wasn't such a good idea as it only confuses people more. Let me try to explain this in easier terms to understand. So how about this...http://en.wikipedia.org/wiki/Standard_deviation lol Statistics is very dependent on standard deviations which accounts for your confidence interval. Standard deviation (SI) uses an (N-1) factor, which contributes less and less as N gets larger. at 20 or above, N-1 is seen as negligible in mathematics terms. I actually use 20 or greater in my research and published works as well... it's quite well known.
Haha, you claim to be a relatively high level statistician and then you say that as long as the sample size is above 20 it is relevant? That is a completely arbitrary number that differs on most textbooks and in most research. For instance, my textbook says that 30 is the minimum sample size to assume a normal distribution. Furthermore, you are ignoring countless factors that influence data, for instance, basic independence, or even..opponent skill level... This is silly.
|
I noticed this months ago but didn't think it was threadworthy. There are usually players with a specialty matchup where they held a winrate over 65%, but that was not true for any zerg in ZvT except DRG when he was new to the scene and was 75%. After becoming more known, even he couldn't stay over 65%.
|
Canada8840 Posts
id say those are the sort of statistice were really looking for, a good terran wins about 60-70% of his ZvT games (MMA is the outlier but outliers are outliers for a reason) it seems in general that everyone is winning about the same amount and pretty mcuh everyones around 60% so id say it shows that ZvT is very slightly imbalanced in the favour of terran
|
On March 16 2012 12:28 schmutttt wrote:Show nested quote +On March 16 2012 11:30 acrimoneyius wrote:
those extra few percents probably account for bunker rushing/2 rax and helion run-bys, which are extremely unforgiving for zerg and even top tier players die to all the time.
Makes you wonder if they'll ever put in a marine range upgrade (but that would possibly screw up TvP) ??? Why would you bring up ZvT being unforgiving then wondering if they will bring in a buff to a unit that obviously does NOT need one?
He's referring to BW, where marines had 4 range before U-238 Shells was researched, giving them the 5 range they have by default in SC2.
|
On March 16 2012 14:30 Demonhunter04 wrote:
I noticed this months ago but didn't think it was threadworthy. There are usually players with a specialty matchup where they held a winrate over 65%, but that was not true for any zerg in ZvT except DRG when he was new to the scene and was 75%. After becoming more known, even he couldn't stay over 65%.
cuz he played more games......the more games one play the more likely their winrate will approach 50% unless you are God/Flash or a Demon like MMA.
|
Idra has been right all along, Zerg just dies to random shit, Terran doesn't.
I've see top level Zergs lose to any number of Terran allins, from bunker rushes to scv pulls to cloaked banshees. Nestea didn't sense a 2-factory hellion build coming and he got wiped with it.
Compare this to what, a single roach-ling-bling allin that requires a greedy terran to be relatively successful (and can still be held)? Mass eco-busts, hoping the Terran doesn't have hellions, bunkers, or tanks? 6-pools that all top level players know how to hold? 1-base Baneling busts that don't work in diamond league?
DRG is noted for his exceptional ZvT, but it's still very easy for him to just lose a game.
Protoss is much the same, lose a whole batch of probes or throw down a wrong FF and you may very well have lost the game.
This isn't true for Terran, it never has been, and goes to explain their almost constant 50%+ winrates against Zerg since the end of Beta.
|
On March 16 2012 14:44 joocybaneling wrote:Show nested quote +On March 16 2012 14:30 Demonhunter04 wrote:
I noticed this months ago but didn't think it was threadworthy. There are usually players with a specialty matchup where they held a winrate over 65%, but that was not true for any zerg in ZvT except DRG when he was new to the scene and was 75%. After becoming more known, even he couldn't stay over 65%. cuz he played more games......the more games one play the more likely their winrate will approach 50% unless you are God/Flash or a Demon like MMA.
Another important thing to note is that he only played in GSTL and was relatively unknown at the time, so players couldn't prepare for him nearly as well.
Why is MMA a "Demon"? Did my name subconsciously influence your thoughts? 
|
So basically, MMA is the outlier and the only one at that. You know about outliers don't you? Minimum sample size for approximating normal distribution is >= 30 btw.
|
Terran having higher win rates at the top compared to the top zergs even tho TLPD says its even is nothing to be surprised about.
Its simply a balance issue, terran is mechanically harder to play so top terrans that are mechanically perfect or close to perfect aren't in the "balanced" are of the game, while Z, being easier to play mechanic wise ( a high master/gm zerg is 95-99% as good at macro/micro as a top one ), is balanced for a top level of play.
I don't know why you would be surprised about it, top terran have always dominated for that very reason, the gold medal ratio is around 60% terran 40% other races, it will be fixed in time when all the pros will have good mechanics and blizzard will be able to nerf terran without killing 90% of the terran pros, terran has been nerfed in each patch and it will continue to be nerfed imo.
|
Katowice25012 Posts
On March 16 2012 11:14 neoghaleon55 wrote:
Edit 1:
The optimal minimum sample size is 20. Above 20, the n value does not relevantly contribute over all (n-1) to the statistics.
I'm sure you remember from your AP stats class and college.
The statistics presented in the OP are greater than 20 sample sizes and thus are relevant.
Edit 2: Maybe presenting that article wasn't such a good idea as it only confuses people more.
Let me try to explain this in easier terms to understand.
So how about this...http://en.wikipedia.org/wiki/Standard_deviation
lol
Statistics is very dependent on standard deviations which accounts for your confidence interval.
Standard deviation (SI) uses an (N-1) factor, which contributes less and less as N gets larger.
at 20 or above, N-1 is seen as negligible in mathematics terms. I actually use 20 or greater in my research and published works as well... it's quite well known.
Yes, 20 is so-called "statistically relevant"...when there are a set number of variables. Those rule-of-thumbs apply for very specific kinds of tests when you're sampling populations in very controlled ways. Looking at MMA's history of 26 games is nowhere near enough to be relevant because there are too many variables within a game, you need to control for opponent, map, style of play at the very least. If you had 26 games of MMA playing DRG on Shakuras with the same openers then yes maybe this would start to apply.
You can't just say you're testing for winrate or balance or something because you're abstracting it in multiple ways, you're several levels above what you're trying to look at.
|
On March 16 2012 11:23 corpuscle wrote:
A winrate of 64% is fucking amazing, though, as Jinro points out. That basically means that statistically you win almost every bo3 against Terran opponents.
Haha, when you think about it this way....
And the fact that DRG also likes a good comeback after being down 0-1, or 0-3.
|
It doesn't surprise me to see MMAs winrate so high vZ and about even in vP and vT.
His style stresses the crap out of his opponents multitask and critical decision making via DROPS and tons of HARASS. Not to say that it is "weak" in vT and vP, T and P have easier ways to control space and stop harass from doing insane amounts of damage (turrets/warpins for drops, siege tanks/templar/collosus for map control strength). MMA will eventually be the standard for Code A in 6 months to a year (at least TvZ). To the other terrans around 60-70% winrated I will say this.
Terran vT and vZ are similar beasts in the curret meta, with more harass than space control in vZ compared to vT. In specifically training either vT or vZ a T gets better at tank positioning, posturing, and attacking multiple locations at once which "bleeds" into the other matchup naturally. Therefore, T prepping hardcore for a season of TvT will have an immediate benefit in their vZ without specific training. There is different decision making and faster/harder micro to be done, but the broad brushstrokes are far less different and therefore T becomes stronger at the MU faster.
Similarly, a Z who practices a meta of ling infestor mass ups in vT AND vP will innately get better at his other MU as well.
I don't want to say that Z players are worse, since many of them have proven to be with the legends of Code S (Nestea, DRG specifically), but their ZvT metagame isn't caught up with the raw mechanics of Marine Tank Medivac compositions to force late-games where zerg has the army composition advantage AND force the terran to trade inefficiently. To boot they are weak in the ZvP mu as well, which shows their relative extinction in Code S this season.
|
The optimal minimum sample size is 20. Above 20, the n value does not relevantly contribute over all (n-1) to the statistics.
I'm sure you remember from your AP stats class and college.
The statistics presented in the OP are greater than 20 sample sizes and thus are relevant.
Edit 2: Maybe presenting that article wasn't such a good idea as it only confuses people more.
Let me try to explain this in easier terms to understand.
So how about this...http://en.wikipedia.org/wiki/Standard_deviation
lol
Statistics is very dependent on standard deviations which accounts for your confidence interval.
Standard deviation (SI) uses an (N-1) factor, which contributes less and less as N gets larger.
at 20 or above, N-1 is seen as negligible in mathematics terms. I actually use 20 or greater in my research and published works as well... it's quite well known.
What in the everlasting f**k are you going on about here? I hope nobody believes it and spreads this nonsense around...
|
|
|
|
|
|
EPT
