|
On May 03 2011 03:16 joyeaux wrote:Show nested quote +On May 03 2011 01:49 Cloak wrote:
Sample size is smaller for Korea but deviation from average is ridiculously huge. If you do a chi squared, I'm sure it would reject the null at 95% confidence, probably even 99% confidence. 700 games is pretty damn sizeable. PvT and PvZ are looking like absolute garbage right now. That would be relevant if we were 1. looking at a random sample of a population and 2. drawing conclusions about the whole population. These statistics measure a non-random section of the population (results of specific tourneys and league) and from those statistics you aren't trying to draw conclusions about the whole population, but about the ephemeral subject of "game balance." TL;DR statistics do not work that way.
Is it not random? It's all Korean tournament games right? Not just the ones where P loses. Doesn't need to be the whole population, just needs to be "a" population.
On May 03 2011 05:14 Silent331 wrote:
If you think win rates=blanace than you are assuming the following
-sample size is big enough to show the average with no deviation
-all players are playing the best that their race can possibly be played
-all races are proportionally represented in sample size
-all games were played on a perfectly balanced map where every race has an equal chance to win in every spawn position combination
-there has been no game development so nothing can happen that is unexpected
if you think this shows balance than be ready to defend that you believe all the above are true.
Actually you just assume that it's all statistically average. Players don't need to play their best all the time, but you assume that proportionally there's players for all races that screw up equally. If the numbers are large enough, you don't need 33/33/33, you just need the smallest portion to be statistically significant. Map balance is definitely a confounding factor for race balance, but that doesn't make imbalance any better or more justified. Regarding game development, you just look at the whole curve.
|
The graphs are pretty awesome.
One thing that I would see added though. Please add the releases of patches in the graph. I'm really curious how those correspond to the changes in the graph.
|
This is amazing! :D I like the visual representation! Beautiful graphs showing interesting trends in both the foreign and korean scene!
Great job!
I will be interested to see how this continues to change! :D I don't suppose we could have a sample size amount under each month?  that would make me super happy!
|
if you made the graph for tournament wins, it will show mostly protoss this month and last. so stop complaining
|
On May 03 2011 16:11 gNs.I-Jasa wrote:
if you made the graph for tournament wins, it will show mostly protoss this month and last. so stop complaining
Nobody is complaining except zergs and terrans who are trying to justify their perception of race balance which has been totally disproved by actual data. Perhaps you need to stop being so defensive.
|
it would be nice if you could post which tournaments are included!
especially if you talk about korea, and we see protoss getting destroyed in the last month, I wonder if this only contains this GSL season, or if there are more korean tournaments in which P didn't perform well.
|
Seeing community collected data like this makes me really kind of hope to see Blizzard release some kind of data set/charts/graphs. You know they have data on everything and I think they could put a lot of perceived issues in the community to rest if we had something more tangible to make our observations from the just personal experiences.
|
why did you exclude mirror matchups? in BoX tournament format in particular, winrates dont always exactly match success in the form of advancement in the tourneys - therefore, a high amount of mirrors in the higher stages of the tournament, after suitably considering the difference in the amount of people playing the races, does say something about the strenght of a race. obviously, this is a whole different and much more intricate story than looking at winrates only, but imho it would be worth it.
i appreciate your work though, its very nice and well-done in itself, i just think it could be extended to tell even more about the development of balance.
balance is not exactly equal to winrate! 
|
|
|
On May 03 2011 16:21 Black Gun wrote:why did you exclude mirror matchups? in BoX tournament format in particular, winrates dont always exactly match success in the form of advancement in the tourneys - therefore, a high amount of mirrors in the higher stages of the tournament, after suitably considering the difference in the amount of people playing the races, does say something about the strenght of a race. obviously, this is a whole different and much more intricate story than looking at winrates only, but imho it would be worth it. i appreciate your work though, its very nice and well-done in itself, i just think it could be extended to tell even more about the development of balance. balance is not exactly equal to winrate!
If you want to see the graph with mirrors included that is very quick to do. In fact the hardest part of making the chart was excluding them. Set analysis expressions are hard...
|
On May 03 2011 01:42 loveeholicce wrote:Show nested quote +On May 03 2011 01:15 Moriwo wrote:
I think these stats show a lot to prove that map changes and race development are making the real changes to the "balance" of the game than actual balance changes (mostly looking at the international scene).
I would imagine that the decline in overall Terran win percentages is mostly due to the introduction of bigger maps and the removal of small maps (i.e. Steppes).
Protoss had a pretty big jump in win percentage with the big Patch 1.2, but I think that's a case of correlation vs. causation. Around that time, Protosses began getting upgrades a lot faster (double forge, etc.), which really proved to be hugely effective in the PvT matchup. Also around that time began the development of the 3 gate expansion and the Void Ray/Colossus deathball in PvZ. As Protoss strategy became more fleshed out, they began winning more despite relatively small changes.
It's also nice to note how these days Zerg is getting much better in the ZvP matchup. Indeed, the infestor buff helped Zerg out a lot, but Zergs also began developing new strategies such as fast burrow, more ling/baneling styles, and more effective use of their "tier 3" units, so much so that in fact the win rates slightly favor Zerg. It was just a few months ago that Zergs everywhere were complaining about ZvP imbalance, and nowadays seem to be doing fine in the matchup.
TvZ really surprised me, although if I were to guess at why the matchup favors Terran so much, I'd probably attribute it to the massive Terran presence in Europe. European Terrans are quite strong, whereas there just aren't that many European Zergs. That's just my guess, I really don't know for sure.
Overall, this just kind of proves that the game needs time to develop and grow before we should really be calling for massive balance changes. Just my take on ZvP trends: Since beta it was obvious Zerg had the greatest production potential with spawn larva. The only difference was that unlike the other two races, Zerg couldn't make workers and attacking units at the same time (well it technically could, but since the units were weaker it just wouldn't work because you needed to establish a stronger economy). However, once Zerg got an economy going it was accepted that the Protoss would get rolled. So basically Protoss play centered around the idea of not letting a Zerg get his economy up uncontested, as they would be sure to get outmacroed in the late game. This pretty much defined the dynamic of the matchup. Zerg would drone for as long as possible, and then make units in anticipation of a push. They'd crush that push and win. Meanwhile Protoss would try to work on various timing rushes and high army pressure builds to prevent this. Recently though, it dawned on Protoss that Zerg lategame was actually quite terrible, and that Zerg couldn't beat a 200 / 200 army. On top of that Zerg couldn't saturate more than ~3 1/2 bases because ot would mean making too many workers. So for the last few months Protoss have more or less always had their core army sitting at home, turtling 3 bases, and maxing out before they pushed. Zerg would wait and build an army in anticipation of a push that never came, and then get completely rolled by a maxed out army that their units simply can't stand up to. Hence we see from ~January till now Zergs doing horribly, especially vs Protoss, because their style was still to wait for a Protoss to move out, which isn't happening until 200 / 200. Trends are now improving with perhaps the infestor buff (although I still maintain they suck in ZvP) but more notably a greater Zerg emphasis on aggression, drops, attacks, etc that try to stop Protoss from building a 200 / 200 army uncontested, cause you're gonna lose every time they do that =_=
See, I disagree that a 200 Zerg can't beat a 200 Protoss. Zergs lose to the deathball because they populate their armies with a mass of garbage low tier units. I've seen Zerg deathballs full of only high-tier units, broods, infestors, ultras, absolutely destroy Protoss deathballs. Fungal is so god damn good when you have enough of them to kill a stalker. And sticking NP on half or all of the colossi is just cool as hell.
Infestors are the future, baby. And broods. And ultras. Newsflash: colossi AoE sucks if it's only hitting one fatass ultralisk and a couple lings.
Let me not misrepresent myself, though: I still think toss is too strong, in a lot of ways. But I think maybe Zerg actually does have the tools to overcome.
|
On May 03 2011 16:11 gNs.I-Jasa wrote:
if you made the graph for tournament wins, it will show mostly protoss this month and last. so stop complaining
Eh, the big things Protosses have won in the last couple of months are + Show Spoiler +GSL Mar Code S, Copenhagen, IEM, Dreamhack and MLG Dallas (and 3 of those were by a single player, MC, who didn't even compete in the other ones and could very well have won those as well) the IPL, Assembly, GSL Mar Code A, GSL World Championship and GSL May Code A (all Protosses have been knocked out).
If MVP went on a tour and cleaned out a bunch of foreign tournaments while MC stayed in Korea, you'd be able to say the same thing about Terran. I think that the majority of the "Protoss OP" perception originated from a combination of MC stomping over everyone and the trouble Zergs have been having in late-game ZvP. Meanwhile Terrans have been flying under the radar and dominating in the GSL WC and most recent GSL, and are presumably happy that Protoss now have most of the attention.
(I may have missed some things but the point remains)
|
From these statistics it seems as if zerg players should be complaining more about terrans then protoss.
|
United States13896 Posts
On May 03 2011 05:45 Sporadic44 wrote:
correct me if im wrong but the OP made these graphs himself based off of data posted on liquipedia?
it would be cool if we could get more information like this on the site. not necessarily line graphs but sets of observable data. for example if you click on metalopolis you could pull up some data on wins/losses regarding matches. played on that map. or other information such as average match length. just some examples. i have limited knowledge on statistics, entry level college math, and a scientific understanding of analyzing and displaying data through biology labs. but i feel this could be something worth at least trying to work on. if anyone else likes the idea send a pm my way.
Metalopolis - original ladder version
Metalopolis GomTV - ramp blockers
Metalopolis 1.1 - ramp blockers + no close spawn
Metalopolis DH - ramp blockers + cross position only
|
Sc2Statistics have been doing some good work! Keep it up!
Curious, how have you been pulling the TLPD information? I been working with TLPD.I data myself, but the lack of an API means I have not been able do more interesting number crunching.
|
United States13896 Posts
On May 03 2011 06:38 TheBanana wrote:Show nested quote +On May 03 2011 06:25 Yoshi Kirishima wrote:Just data, remember not to take this and make a quick assumption or anything ^_^ The number of players and tournaments in Korea is much lower, so therefore these numbers are more volatile. How do you know this? It's a bit ignorant of you to say that, unless you also are very close with the Korean scene. Just because we don't get too many news or posts here on TL or in the western world in general, it doesn't mean they don't have many tournaments or players. One way of knowing this is that the Koreans players have stated this numerous of times in interviews. There are very few tournaments besides GSL in KR.
There simply aren't many (maybe 1 or 2 events a month outside of the GSL that carry any decently large prize purse and can attract progamers to play in them. Sure there may be some lower level community based tournaments going on in pc baangs that we don't know about but they aren't putting forth top tournament cash for the best players to vie over.
|
On May 03 2011 01:42 loveeholicce wrote:Show nested quote +On May 03 2011 01:15 Moriwo wrote:
I think these stats show a lot to prove that map changes and race development are making the real changes to the "balance" of the game than actual balance changes (mostly looking at the international scene).
I would imagine that the decline in overall Terran win percentages is mostly due to the introduction of bigger maps and the removal of small maps (i.e. Steppes).
Protoss had a pretty big jump in win percentage with the big Patch 1.2, but I think that's a case of correlation vs. causation. Around that time, Protosses began getting upgrades a lot faster (double forge, etc.), which really proved to be hugely effective in the PvT matchup. Also around that time began the development of the 3 gate expansion and the Void Ray/Colossus deathball in PvZ. As Protoss strategy became more fleshed out, they began winning more despite relatively small changes.
It's also nice to note how these days Zerg is getting much better in the ZvP matchup. Indeed, the infestor buff helped Zerg out a lot, but Zergs also began developing new strategies such as fast burrow, more ling/baneling styles, and more effective use of their "tier 3" units, so much so that in fact the win rates slightly favor Zerg. It was just a few months ago that Zergs everywhere were complaining about ZvP imbalance, and nowadays seem to be doing fine in the matchup.
TvZ really surprised me, although if I were to guess at why the matchup favors Terran so much, I'd probably attribute it to the massive Terran presence in Europe. European Terrans are quite strong, whereas there just aren't that many European Zergs. That's just my guess, I really don't know for sure.
Overall, this just kind of proves that the game needs time to develop and grow before we should really be calling for massive balance changes. Just my take on ZvP trends: Since beta it was obvious Zerg had the greatest production potential with spawn larva. The only difference was that unlike the other two races, Zerg couldn't make workers and attacking units at the same time (well it technically could, but since the units were weaker it just wouldn't work because you needed to establish a stronger economy). However, once Zerg got an economy going it was accepted that the Protoss would get rolled. So basically Protoss play centered around the idea of not letting a Zerg get his economy up uncontested, as they would be sure to get outmacroed in the late game. This pretty much defined the dynamic of the matchup. Zerg would drone for as long as possible, and then make units in anticipation of a push. They'd crush that push and win. Meanwhile Protoss would try to work on various timing rushes and high army pressure builds to prevent this. Recently though, it dawned on Protoss that Zerg lategame was actually quite terrible, and that Zerg couldn't beat a 200 / 200 army. On top of that Zerg couldn't saturate more than ~3 1/2 bases because ot would mean making too many workers. So for the last few months Protoss have more or less always had their core army sitting at home, turtling 3 bases, and maxing out before they pushed. Zerg would wait and build an army in anticipation of a push that never came, and then get completely rolled by a maxed out army that their units simply can't stand up to. Hence we see from ~January till now Zergs doing horribly, especially vs Protoss, because their style was still to wait for a Protoss to move out, which isn't happening until 200 / 200. Trends are now improving with perhaps the infestor buff (although I still maintain they suck in ZvP) but more notably a greater Zerg emphasis on aggression, drops, attacks, etc that try to stop Protoss from building a 200 / 200 army uncontested, cause you're gonna lose every time they do that =_=
In the past 2 months zergs have more tournaments wins against protoss then protoss have against zerg so i dont see how you can say from january till now zergs have been doing horribly you could say from january until febuary zergs have been doing horribly but i would stack that up to a meta game shift over imbalance, and now with zergs having a higher win rate in zvp i would stack that up to another metagame shift.
Also i was looking at statistics for NASL and zerg has almost a 70% win rate against protoss, i know that is mostly insignificant because NASL hasnt been on for a very long period of time , but i still found it interesting.
|
On May 02 2011 23:50 Zaros wrote:
Nice to know the PvZ troubles i have been having is echoed in korea. Very interesting results.
Well, you're not a korean, so you shouldn't be having any troubles.
Edit: Also, even in Korea, protoss isn't actually doing poorly. The race is winning tournaments a lot. If it's just one tournament, it doesn't necessarily make the race balanced (Z did win GSL 1 during 1.00), but as far as I know, Protoss is doing just fine.
Correct me if I'm wrong, and Protoss isn't winning any korean tournaments.
|
On May 03 2011 16:38 Primadog wrote:
Sc2Statistics have been doing some good work! Keep it up!
Curious, how have you been pulling the TLPD information? I been working with TLPD.I data myself, but the lack of an API means I have not been able do more interesting number crunching.
I'm scraping the data from the html XML tables. The program I'm using is called Qlikview, and I'm using it both for the scraping and the aggregation/visualization.
I would love some API's though!
|
On May 03 2011 16:34 Severian wrote:Show nested quote +On May 03 2011 16:11 gNs.I-Jasa wrote:
if you made the graph for tournament wins, it will show mostly protoss this month and last. so stop complaining Eh, the big things Protosses have won in the last couple of months are + Show Spoiler +GSL Mar Code S, Copenhagen, IEM, Dreamhack and MLG Dallas (and 3 of those were by a single player, MC, who didn't even compete in the other ones and could very well have won those as well) . Meanwhile, Zergs/Terrans have won + Show Spoiler +the IPL, Assembly, GSL Mar Code A, GSL World Championship and GSL May Code A (all Protosses have been knocked out). If MVP went on a tour and cleaned out a bunch of foreign tournaments while MC stayed in Korea, you'd be able to say the same thing about Terran. I think that the majority of the "Protoss OP" perception originated from a combination of MC stomping over everyone and the trouble Zergs have been having in late-game ZvP. Meanwhile Terrans have been flying under the radar and dominating in the GSL WC and most recent GSL, and are presumably happy that Protoss now have most of the attention. (I may have missed some things but the point remains)
Yea i have no idea why zergs have decided to put all their attention (and hatred) towards protoss while terran have been secretely doing better against zerg this whole time, just goes to show that peoples perception of what is a problem and what actually might be a problem according to statistics can be completely different. Like i said if you look in other leagues such as NASL you can see zergs having a 70% win rate against protoss, wich although may be insignificant should that trend continue maybe protoss players may have something to complain about(although overall i think zerg is by far the whiniest race, they complain about anything and everything they see one finals and suddenly however someone won those finals against the zerg, that strategy/strategies is suddenly the most OP shit in the world)
Hey zergs newsflash your doing fine at the highest level ZVP so stop complaining maybe try something besides roach hydra corrupter, alot of zergs i see complaining are still going roach hydra corrupter and then wondering why they are losing =/ Infestors broodlords and ultra's are where its at. Im sick and tired of hearing zergs complain about every little advantage protoss has instead of using their own unique advantages to win, each race has certain advantages thats just how it works.
|
|
|
|
|
|
EPT
