|
On June 07 2011 23:48 awu25 wrote:Show nested quote +On June 07 2011 23:41 Yaotzin wrote:On June 07 2011 23:37 Lonyo wrote:
MC lost to Alicia in a PvP in GS ST
In GSTL before that he went 3-1
He lost to Polt in GSL April, PvT, but won a PvZ and beat a terran in Up/Down matches.
He won the GSL before that.
Protoss may be doing worse against Zerg overall, but MC's performance isn't bad because of balance changes. He lost a PvT 2-0 and a PvP 2-0 and that's why he went out of the two recent GSLs. Yeah I know, I suspect he's just spreading himself thin with so many trips abroad for tournaments. Anyway it's about Protoss overall, they aren't doing well, not even the good ones.
That's definitely not true. You rarely see an even distribution of a races in the opening round of a tournament. It would be a very intensive study to perform but I can guarantee that it wouldn't be close to cancelling out when you start looking at professionals vs amateurs in early rounds.
Do you have any data on the races of amateurs in competitions? No but I'm saying you can't just say the skill level thing will cancel out. Like in MLG, you have tons of open bracket amateurs who are playing against professionals. For every professional T who played an amateur Z, you won't see an equal amount of professional Z who played an amateur T. I don't expect anyone to take these factors into account since it would be too much work and the boundaries are vague
Fair point. Though if the racial distribution of amateurs entering competitions is equal then that's moot. Need moar data :>
|
On June 07 2011 23:43 Olinim wrote:Show nested quote +On June 07 2011 23:41 Alpina wrote:On June 07 2011 23:31 awu25 wrote:
I think many people still need to take a statistics course. These matches weren't all played under the same circumstances (different maps, different players, different skill levels) so to say that the game is balanced/imbalanced based on these statistics is wrong. Exaclty. Those statistic does not mean much at all. Just take a GSL final game nestea vs. inca, and inca DT rushed nestea every single game. And all that horrible play comes into statistics saying "toss is underperforming", while it does not saying anything about game balance. Then when do any statistics matter if you can just brush it off so easily? Surely MC's wins shouldn't count for protoss, idra just roach ling all inned him every game.
they matter when you cant brush them off so easily.
even if we could be certain that these results were of all equal value a 51% win ratio instead of 50% doesnt mean anything.
these stats are super interesting but this is not going to be used as data for balance changes
|
On June 07 2011 23:24 aurum510 wrote:
How can IdrA cry about ZvP... It's the most imbalanced matchup and it's in favor of Zerg...
Not if protoss players are "competent" whatever the fuck that means.
Also, LOL at september 2010, 42% win for zerg. I think blizzard would say that the game was atleast "cool" and "awesome" back then.
|
Using the data for the last month and a null-hypothesis that all matchups are balanced (meaning a 50% winrate for all 3 races), you can estimate the expected standard deviation to be about 1.5%. This means that all data for last month is within a two sigma band around the expected mean, which is pretty decent and there are no statistically significant outliers.
|
On June 07 2011 23:50 DannyJ wrote:
I'd love to know what stats blizzard themselves study.
The overall trends as opposed to exclusively pro play. Also, they look at 2v2s, 3v3s and 4v4s.
|
Protoss seems to do the least QQing about their race, yet seems to be doing the worst overall.
Interesting.
Are these the statistics all across the board? Across all servers and all leagues?
|
On June 07 2011 23:49 Chaosvuistje wrote:
I would love to see statistics of this in relation to game duration, not just win/loss.
Yes, these statistics clearly show that overall, zerg has about the same chances to win as protoss. However, none of these statistics show if these wins come from the early game, midgame or the lategame.
If we could see winpercentages per matchup overall , y-axis winpercentage, x-axis game duration, we could deduct if lategame protoss is imbalanced, or if lategame TvZ is terribly lopsided in favour of Zerg. Until then these numbers are a good commodity, but nothing conclusion can be deducted from them balance wise.
Don't think game length is recorded anywhere so alas.
An unequal winrate at various times would not, however, mean that one race is imba at that point. For example, let's say current metagame dictates incredibly greedy early game play. Under this metagame, the large majority of their losses will be early game, with many economic wins lategame. Yet the metagame could simply be flawed - perhaps this race would actually be better off with a middling approach throughout the game, which could produce a 50% winrate at all times.
That's not even getting into whether having 50% at all times is even desirable.
|
On June 07 2011 23:53 Yaotzin wrote:Show nested quote +On June 07 2011 23:48 awu25 wrote:On June 07 2011 23:41 Yaotzin wrote:On June 07 2011 23:37 Lonyo wrote:
MC lost to Alicia in a PvP in GS ST
In GSTL before that he went 3-1
He lost to Polt in GSL April, PvT, but won a PvZ and beat a terran in Up/Down matches.
He won the GSL before that.
Protoss may be doing worse against Zerg overall, but MC's performance isn't bad because of balance changes. He lost a PvT 2-0 and a PvP 2-0 and that's why he went out of the two recent GSLs. Yeah I know, I suspect he's just spreading himself thin with so many trips abroad for tournaments. Anyway it's about Protoss overall, they aren't doing well, not even the good ones.
That's definitely not true. You rarely see an even distribution of a races in the opening round of a tournament. It would be a very intensive study to perform but I can guarantee that it wouldn't be close to cancelling out when you start looking at professionals vs amateurs in early rounds.
Do you have any data on the races of amateurs in competitions? No but I'm saying you can't just say the skill level thing will cancel out. Like in MLG, you have tons of open bracket amateurs who are playing against professionals. For every professional T who played an amateur Z, you won't see an equal amount of professional Z who played an amateur T. I don't expect anyone to take these factors into account since it would be too much work and the boundaries are vague Fair point. Though if the racial distribution of amateurs entering competitions is equal then that's moot. Need moar data :>
Not just equal racial distribution though. You could have a situation where 80% of amateur Ts play professional Zs but only 40% of amateur Zs played professional Ts, which doesn't cancel out. The percentages would have to be equal
|
Keep the good work.
Personally I like the pooled data including both foreign and Korean games. Larger sample is always better.
But you could also include separated data as well to see if there is a significant difference as there is a difference in the two populations.
An interesting stat to also look at would be the racial percentages in the first round of all the major tournaments. Would allow for some analysis of the percentages of races played.
|
On June 08 2011 00:00 awu25 wrote:Show nested quote +On June 07 2011 23:53 Yaotzin wrote:On June 07 2011 23:48 awu25 wrote:On June 07 2011 23:41 Yaotzin wrote:On June 07 2011 23:37 Lonyo wrote:
MC lost to Alicia in a PvP in GS ST
In GSTL before that he went 3-1
He lost to Polt in GSL April, PvT, but won a PvZ and beat a terran in Up/Down matches.
He won the GSL before that.
Protoss may be doing worse against Zerg overall, but MC's performance isn't bad because of balance changes. He lost a PvT 2-0 and a PvP 2-0 and that's why he went out of the two recent GSLs. Yeah I know, I suspect he's just spreading himself thin with so many trips abroad for tournaments. Anyway it's about Protoss overall, they aren't doing well, not even the good ones.
That's definitely not true. You rarely see an even distribution of a races in the opening round of a tournament. It would be a very intensive study to perform but I can guarantee that it wouldn't be close to cancelling out when you start looking at professionals vs amateurs in early rounds.
Do you have any data on the races of amateurs in competitions? No but I'm saying you can't just say the skill level thing will cancel out. Like in MLG, you have tons of open bracket amateurs who are playing against professionals. For every professional T who played an amateur Z, you won't see an equal amount of professional Z who played an amateur T. I don't expect anyone to take these factors into account since it would be too much work and the boundaries are vague Fair point. Though if the racial distribution of amateurs entering competitions is equal then that's moot. Need moar data :> Not just equal racial distribution though. You could have a situation where 80% of amateur Ts play professional Zs but only 40% of amateur Zs played professional Ts, which doesn't cancel out. The percentages would have to be equal
Hum, I guess you'd need equal distribution of pros as well, then it would equal out with sufficient sample size.
Anyway it's mostly the trends that are interesting and they're almost certainly less affected by such things.
|
On June 07 2011 23:12 CursedFeanor wrote:
great data! it seems no one can really complain about balance for the time being. that's great, good job Blizzard!
This implies that those who complain about balance care about statistics.
|
guess that % drop in PvZ players used to 4 gating half or more of the games and the new roach ling timing on zerg part
|
4gate died like months ago. Got a small revival with the nexus cancel version but that's disappearing now too.
|
On June 08 2011 00:14 Yaotzin wrote:
4gate died like months ago. Got a small revival with the nexus cancel version but that's disappearing now too.
And MC basically solved the roach/ling timing attack.
|
I'm interested to know if data like this exists within separate leagues?
|
On June 07 2011 23:13 Zaphid wrote:Show nested quote +On June 07 2011 23:06 Yaotzin wrote:
Please please separate Korean data from International. Even though the number of games is sadly small, Korea is still a very good indicator of future international trends - as we've now seen with ZvP. I believe Korean trends are more pronounced by whoever wins the GSL, as usual more games means better statistics. Interesting how in the last 1-2 months the zerg crying was turned on it's head, is it time for other races to start whining ? :D
D'oh, I was still hoping we'd get the 2 armor 1 supply roaches back :p
But seriously, I'm pretty sure all races have been accused of being imba from the very beginning. These accusations going in all directions might even be a good indicator of balance.
|
On June 07 2011 23:18 Day[9] wrote:
Fantastic graphs! I have a really stupid request if possible lol:
I'm used to Zerg = Red, Terran = Blue, Protoss = Green from years of surfing Teamliquid. Would it be possible to do some color swapouts?
Regardless, I LOVE seeing data like this :D
Especially interesting considering your discussion with Idra!
|
On June 07 2011 23:09 Mailing wrote:
Not really. Korean protoss are falling apart, to the point where only Alicia, MC, Ace, Squirtle, and maybe Huk are having any form of consistent results, namely because they are actually really good players who belong in Code S >>
I would suspect the rest, like inca/anypro/tester/hongun/sangho to fall out of code S soon.
Only Alicia and MC I would say show consistensy. Like 10 protoss players fails so badly in GSL super tournament, either they're no good or something is wrong with the race.
Either fact is that only 1 protoss is left in the round of 16.
|
On June 08 2011 00:19 eYeball wrote:Show nested quote +On June 07 2011 23:09 Mailing wrote:
Not really. Korean protoss are falling apart, to the point where only Alicia, MC, Ace, Squirtle, and maybe Huk are having any form of consistent results, namely because they are actually really good players who belong in Code S >>
I would suspect the rest, like inca/anypro/tester/hongun/sangho to fall out of code S soon. Only Alicia and MC I would say show consistensy. Like 10 protoss players fails so badly in GSL super tournament, either they're no good or something is wrong with the race. Either fact is that only 1 protoss is left in the round of 16.
and which other protosses would you say really stood up there? AnyPro? Inca? only toss i feel bad for is San but he lost in PvP
|
It's a concern that there are so few notable Protoss players though. Just as it was a concern how few notable Zergs there were after release.
Even if one of or two players can make Protoss competitive at the top, something has to be done to get more middle of the road pros for variety's sake.
(assuming the horrible number of tosses continues, might just be a metagame thing)
|
|
|
|
|
|
EPT
