For the past nearly 5 months, we've held 2 Playhem Daily tournaments every day on EU and NA. Our brackets fill to an average of 150-220 players each time and match verification is done by the players uploading replays and those replays have just been sitting on servers collecting dust - until now. Thanks in part to the wonderful work done by TL user Dakota_Fanning in creating Sc2Gears I've taken an in-depth look at how our map pool has been influencing tournament outcomes and balance in general.
Over 31,500 replays (so that's what it looks like) of mostly masters, grand masters, and professional gamers. I consider these to be a great representation of real-world statistics as the entry is free and no qualifications to sign up so a very wide range of players enter, unlike data from TLPD or the like which reflect mostly the very top tier. I present you my findings:
MLG Shattered Temple
Replays: 7486 Average Game Length: 10:49
TvP - 54% PvZ - 45% ZvT - 43%
MLG metalopolis
Replays: 6548 Average Game Length: 10:47
TvP - 54% PvZ - 46% ZvT - 49%
ESV Cloud Kingdom
Replays: 3040 Average Game Length: 11:36
TvP - 45% PvZ - 52% ZvT - 51%
MLG Antiga Shipyard
Replays: 2750 Average Game Length: 11:55
TvP - 59% PvZ - 50% ZvT - 52%
MLG Shakuras Plateau
Replays: 2425 Average Game Length: 11:46
TvP - 50% PvZ - 53% ZvT - 48%
GSL Dualsight
Replays: 2251 Average Game Length: 10:53
TvP - 57% PvZ - 44% ZvT - 50%
Nexus First FE - 35% win rate Forge Fast Expand - 27% win rate Pool, double hatch, double gas, roach - 61% win rate
Tal'Darim Altar LE
Replays: 1458 Average Game Length: 10:58
TvP - 51% PvZ - 49% ZvT - 51%
Misc. Stats:
Top 5 players by win-rate (with more than 50 games played) Maynard (T) 87% - 169W 24L viOLet[EU] (Z) 85% - 91W 16L Campanella (T/Z) 85% - 69W 12L uGpSwip (T) 84% - 93W 16L BLY (Z) 84% - 70W 13L
Highest winrate opening build order for each race across all maps/matchups [Terran] Supply Depot, Barracks, Refinery, Supply Depot, Factory, Supply Depot - 588 occurences / 60% winrate [Protoss] Pylon, Gateway, Gateway, Pylon - 116 occurences / 62% winrate [Zerg] Extractor, Spawning Pool, Hatchery - 95 occurences / 67% winrate
Honorable mention - the most played opening build order overall also has one of the consistently worst win rates. [Protoss] Pylon, Gateway, Assimilator, Pylon, Cybernetics Core, Assimilator - 6289 occurences / 46% winrate
Conclusion As tournament organizers, map makers, and a community as a whole we can have a huge impact on the balance of this game that goes almost as far as Blizzard's balance patches. As the game continues to develop and move forward I think it's important that some of these older maps are discontinued and replaced with better options (Dual Sight, Shattered, Metalopolis) or modified to compensate for the differences in win rates that are seen.
Tal'Darim Altar LE gets a special shout out for being possibly the most balanced map currently in our pool, it scored nearly 50% win rates for every match up and overall race win rates.
As a result of these numbers we've significantly changed our map pool the last few weeks and are going to continue to change it to include new maps and see how they stack up to the rest.
Finally, a word cloud including all chat from the replays. 33,076 GGs.
These are by no means perfect statistics, they are very rough data that I just wanted to share. There are surely many techniques and filters that could be applied to get much more accurate information out of this but I'll leave that to map makers and Blizzard. I think even with just the basic level of sorting I applied to the data, it can provide useful insight into map balance and overall race interactions.
Very very interesting information here. It's interesting to see that maps that people feel are favoured to once race v another (ZvX on Metal, TvZ on Antiga) are in fact rather balanced. Tal darim as the most balanced map is also quite interesting.
The results of Dualsight aren't really surprising I think. Despite the general consensus that it's a good map I personally think it's a horrible map for Protoss due to the way the expansions are designed.
Overall a great write-up! I think Tal'Darim Altar LE needs a few more replays to produce a significant result.
these stats don't really reflect balance and random masters players don't even play the game right anyway.
i.e.: Metal should be way more z favored, and antiga should be incredibly terran favored. Dual Sight and Shakuras should be more T favored TvZ too. Can't really comment on the other matchups.
There are some errors of statistics to be honest: the pylon gateway cybercore one doesn't show what the toss is going for and what match ups. it is the most consistent protoss opening that is done (but not showing what tech route and what comes next, for all we know, it isn't a nexus first or FFE) and since toss has one of the worst matchup win rate, obviously this opening, be it 4 gate, 3 gate robo expo that the opening leads to, will have the highest rate of losing.
and since 2 gate is an opening designed to win right off, it shows the actual build can win games, unlike the one above which only shows what opening he is doing and what comes next (the rest of the build that is designed to win the game) would never get a winning rate as high as the 2 gate one
Possible Takeaways (understanding that this is not top level exclusive):
Map size: Big open maps like Tal'Darim are better balanced than small ones. This contradicts Browder's flawed reasoning that bigger maps are worse for noobs because you have to scout more area. The truth is that smaller maps are less forgiving and when someone gets an advantage you can't recover because it takes mere seconds to march into the opposing base. Scouting is fairly easy, you just rally a few units places and see how far they get. Blizzard should hire community map makes to create ladder maps, let them do it for free, or use the GSL map pool as the official ladder pool.
Macro Mechanics The macro mechanics are nice, but they really make the game more volatile, comebacks are nearly impossible, and games are quite short. I'd like Blizzard to nerf them and see what happens.
Is it possible to upload a zip of the 31k replays? I know it's alot, but I wouldn't mind grinding out some serious analysis for the next few weeks and this would make sense to do since I already worked through the violet replay.
On February 04 2012 15:03 K3Nyy wrote: Nice. Even though it's not really at the highest level, it's still interesting to see.
On February 04 2012 15:07 KawaiiRice wrote: these stats don't really reflect balance and random masters players don't even play the game right anyway.
As mentioned in the OP, I feel that databases like TLPD, Gomtv.net, and MLG provide excellent data for the highest level of play. Unfortunately the majority of players are not at the highest level so I felt this gives an interesting insight into the "average joe" masters/GM level of play which we don't see too often even though it accounts for probably the majority of people that have valid concerns about balance.
On February 04 2012 15:10 ETisME wrote: There are some errors of statistics to be honest: the pylon gateway cybercore one doesn't show what the toss is going for and what match ups. it is the most consistent protoss opening that is done (but not showing what tech route and what comes next, for all we know, it isn't a nexus first or FFE) and since toss has one of the worst matchup win rate, obviously this opening, be it 4 gate, 3 gate robo expo that the opening leads to, will have the highest rate of losing.
and since 2 gate is an opening designed to win right off, it shows the actual build can win games, unlike the one above which only shows what opening he is doing and what comes next (the rest of the build that is designed to win the game) would never get a winning rate as high as the 2 gate one
I was only providing that last statistic a bit tongue-in-cheek, but for comparison the third most played protoss opener does show quite a bit better results: Pylon, Gateway, Assimilator, Pylon, Cybernetics Core, Gateway - 2450 occurences / 53% win rate
Haha so Proxy 2 Gate has the highest win ratio for Protoss, while the standard 2 gas opener has the lowest win rate. That is a bit misleading though, since Protoss seems to have the lowest win % on the majority of the maps, so if Protoss losses to the most, then it makes sense that the build order that losses the most is the most standard Protoss opener.
On February 04 2012 15:10 ETisME wrote: There are some errors of statistics to be honest: the pylon gateway cybercore one doesn't show what the toss is going for and what match ups. it is the most consistent protoss opening that is done (but not showing what tech route and what comes next, for all we know, it isn't a nexus first or FFE) and since toss has one of the worst matchup win rate, obviously this opening, be it 4 gate, 3 gate robo expo that the opening leads to, will have the highest rate of losing.
and since 2 gate is an opening designed to win right off, it shows the actual build can win games, unlike the one above which only shows what opening he is doing and what comes next (the rest of the build that is designed to win the game) would never get a winning rate as high as the 2 gate one
This build order is a 2 gas opener, so no 1 gate fe or 4 gate. Your expansion will be way later than with 1 gas builds.
On February 04 2012 15:10 ETisME wrote: There are some errors of statistics to be honest: the pylon gateway cybercore one doesn't show what the toss is going for and what match ups. it is the most consistent protoss opening that is done (but not showing what tech route and what comes next, for all we know, it isn't a nexus first or FFE) and since toss has one of the worst matchup win rate, obviously this opening, be it 4 gate, 3 gate robo expo that the opening leads to, will have the highest rate of losing.
and since 2 gate is an opening designed to win right off, it shows the actual build can win games, unlike the one above which only shows what opening he is doing and what comes next (the rest of the build that is designed to win the game) would never get a winning rate as high as the 2 gate one
You should notice the 2nd gas is included in that winrate, so the build is for double gas P on 1 base. You can be sure it ISN'T a nexus or forge FE after this... dunno how you could possibly think a protoss is going to open double gas and then forge FE LOL
On February 04 2012 15:10 ETisME wrote: There are some errors of statistics to be honest: the pylon gateway cybercore one doesn't show what the toss is going for and what match ups. it is the most consistent protoss opening that is done (but not showing what tech route and what comes next, for all we know, it isn't a nexus first or FFE) and since toss has one of the worst matchup win rate, obviously this opening, be it 4 gate, 3 gate robo expo that the opening leads to, will have the highest rate of losing.
and since 2 gate is an opening designed to win right off, it shows the actual build can win games, unlike the one above which only shows what opening he is doing and what comes next (the rest of the build that is designed to win the game) would never get a winning rate as high as the 2 gate one
You should notice the 2nd gas is included in that winrate, so the build is for double gas P on 1 base. You can be sure it ISN'T a nexus or forge FE after this... dunno how you could possibly think a protoss is going to open double gas and then forge FE LOL
I open double gas and get a Nexus with 1 Gate (MC's Gate-Nexus-Gate-Stargate-Gate) opener vs Zerg all the time. I also like doing 3 Gate Expands vs Terran, which open double gas, get 3 Gates then expand. So there are plenty of expand builds that come of double gas openers.
we have to think that 5 months ago the game was a different version than the current one, right? So maybe these stats can't be taken too seriously since the balance could have changed even if by a few percent.
On February 04 2012 15:21 Yoshi Kirishima wrote: holy shit, the stats for shak and taldarim are a lot closer than I'd think =O
I'm surprised the TvP on antiga is so high... isn't that supposed to be an easier map as protoss since you can take 3 base easily?
Same with TvP on Dual Sight, though I'm guessing terrans usually do some kind of early aggression or 1-1-1 all-in on that map to get an easy win.
anyways thanks a lot for sharing these!
MLG Antiga Shipyards is 2 player forced cross spawn. I think most 2 player maps favor Terran because we have the most powerful rushes, which are the most effective on maps that don't require a scout.
And MLG Shattered, MLG Antiga, MLG Dual Sight, and MLG Metalopolis all still have golds expos. Which is a big deal in all TvX.
From looking at these results I bet Playhem could fix all their balancing problems just by replacing golds with normal expansions. The 4 worst balanced maps all have golds, and the 3 best balanced maps all do not have golds.
On February 04 2012 15:07 KawaiiRice wrote: these stats don't really reflect balance and random masters players don't even play the game right anyway.
i.e.: Metal should be way more z favored, and antiga should be incredibly terran favored. Dual Sight and Shakuras should be more T favored TvZ too. Can't really comment on the other matchups.
Regardless, nice post. <3 Playhem
Define "playing the game right." Not to be a dick just want to understand your point of view a bit better. I suck so i'm not going to get into anything with you but i would assume Most masters players and above have a decent idea of how to play the game hence why they're in said league. Generally speaking.
a lot of people just blindly recite TLPD stats as truth without knowing the knitty-gritty about the maps. This is an interesting look at a reasonable range of pro and non-pro on each map.... and maybe we need to see Shattered Temple move on to greener pastures.
On February 04 2012 15:10 ETisME wrote: There are some errors of statistics to be honest: the pylon gateway cybercore one doesn't show what the toss is going for and what match ups. it is the most consistent protoss opening that is done (but not showing what tech route and what comes next, for all we know, it isn't a nexus first or FFE) and since toss has one of the worst matchup win rate, obviously this opening, be it 4 gate, 3 gate robo expo that the opening leads to, will have the highest rate of losing.
and since 2 gate is an opening designed to win right off, it shows the actual build can win games, unlike the one above which only shows what opening he is doing and what comes next (the rest of the build that is designed to win the game) would never get a winning rate as high as the 2 gate one
You should notice the 2nd gas is included in that winrate, so the build is for double gas P on 1 base. You can be sure it ISN'T a nexus or forge FE after this... dunno how you could possibly think a protoss is going to open double gas and then forge FE LOL
so judging by that is it safe to say thats very indicative of 1 base stargate play? for the most part? If not can you explain why? I suck at the game so try not to flame if i'm way off pwease. want to keep my pillow tear free tonight :D
On February 04 2012 15:44 emc wrote: we have to think that 5 months ago the game was a different version than the current one, right? So maybe these stats can't be taken too seriously since the balance could have changed even if by a few percent.
This isn't meant to give deep meaning of the game, just a general idea, since I think the majority of people on TL are probably the same that are represented in the stats.
By that I mean that you shouldn't be allowed to complain about..say metalopolis unless you are a top tier terran or protoss, and same for zergs on antiga. But it does show some imbalances with maps in the ladder pool, which needs to be addressed.
Hopefully data like this should convince blizzard to change it up a bit.
On February 04 2012 15:07 KawaiiRice wrote: these stats don't really reflect balance and random masters players don't even play the game right anyway.
i.e.: Metal should be way more z favored, and antiga should be incredibly terran favored. Dual Sight and Shakuras should be more T favored TvZ too. Can't really comment on the other matchups.
Regardless, nice post. <3 Playhem
Define "playing the game right." Not to be a dick just want to understand your point of view a bit better. I suck so i'm not going to get into anything with you but i would assume Most masters players and above have a decent idea of how to play the game hence why they're in said league. Generally speaking.
But i'm silver so what do i know.
you can get into masters without having deep knowledge of what exactly you're doing and just macroing. your assumption is false because most masters don't really know and masters itself doesn't actually mean anything, it's just a stupid label blizzard decided to place on people to make them feel good.
if masters players knew how to play properly then top players wouldn't mow down ladder with ridiculous winrates.
On February 04 2012 15:07 KawaiiRice wrote: these stats don't really reflect balance and random masters players don't even play the game right anyway.
i.e.: Metal should be way more z favored, and antiga should be incredibly terran favored. Dual Sight and Shakuras should be more T favored TvZ too. Can't really comment on the other matchups.
Regardless, nice post. <3 Playhem
Define "playing the game right." Not to be a dick just want to understand your point of view a bit better. I suck so i'm not going to get into anything with you but i would assume Most masters players and above have a decent idea of how to play the game hence why they're in said league. Generally speaking.
But i'm silver so what do i know.
you can get into masters without having deep knowledge of what exactly you're doing and just macroing. your assumption is false because most masters don't really know and masters itself doesn't actually mean anything, it's just a stupid label blizzard decided to place on people to make them feel good.
if masters players knew how to play properly then top players wouldn't mow down ladder with ridiculous winrates.
It would very stupid of Playhem to balance their maps according to the top 0.0001%. They don't have the prize money to draw those kinds of players anyway. Balancing their maps to the 0.0001% would unbalance all the maps for the 99.9999%
Now, if this were a thread about GSL then I'd certainly agree with you, because everyone in the GSL is at that highest level. But I believe that attitude is too elitist for a tournament like playhem.
On February 04 2012 15:07 KawaiiRice wrote: these stats don't really reflect balance and random masters players don't even play the game right anyway.
i.e.: Metal should be way more z favored, and antiga should be incredibly terran favored. Dual Sight and Shakuras should be more T favored TvZ too. Can't really comment on the other matchups.
Regardless, nice post. <3 Playhem
Define "playing the game right." Not to be a dick just want to understand your point of view a bit better. I suck so i'm not going to get into anything with you but i would assume Most masters players and above have a decent idea of how to play the game hence why they're in said league. Generally speaking.
But i'm silver so what do i know.
you can get into masters without having deep knowledge of what exactly you're doing and just macroing. your assumption is false because most masters don't really know and masters itself doesn't actually mean anything, it's just a stupid label blizzard decided to place on people to make them feel good.
if masters players knew how to play properly then top players wouldn't mow down ladder with ridiculous winrates.
So what do you have to gain from reading TL? And how is your brain more suited to apply what you learn from here than said lowly "masters"?
This doesn't really prove a lot given that playhem events tend to be very cheese heavy, which explains some of the win rates that don't make sense. I mean the protoss build orders that are most common are fast double gas (almost always something cheesy) or 2 gate (probably proxy), if anyone needs some convincing. Nice of them to release stats and put in the effort, but also reflects very little in comparison to stats based on higher competition.
So for all those (like Blizzard) who fall back on smaller map as rush map, it is interesting to note that the average game length on Tal' Darim is about the same as it is on all maps: 11ish minutes.
On February 04 2012 15:10 ETisME wrote: There are some errors of statistics to be honest: the pylon gateway cybercore one doesn't show what the toss is going for and what match ups. it is the most consistent protoss opening that is done (but not showing what tech route and what comes next, for all we know, it isn't a nexus first or FFE) and since toss has one of the worst matchup win rate, obviously this opening, be it 4 gate, 3 gate robo expo that the opening leads to, will have the highest rate of losing.
and since 2 gate is an opening designed to win right off, it shows the actual build can win games, unlike the one above which only shows what opening he is doing and what comes next (the rest of the build that is designed to win the game) would never get a winning rate as high as the 2 gate one
You should notice the 2nd gas is included in that winrate, so the build is for double gas P on 1 base. You can be sure it ISN'T a nexus or forge FE after this... dunno how you could possibly think a protoss is going to open double gas and then forge FE LOL
I open double gas and get a Nexus with 1 Gate (MC's Gate-Nexus-Gate-Stargate-Gate) opener vs Zerg all the time. I also like doing 3 Gate Expands vs Terran, which open double gas, get 3 Gates then expand. So there are plenty of expand builds that come of double gas openers.
Sorry, it's just something I rarely see. Even if MC did it one time, it happens almost never. I just did an analysis in sc2gears of every single korean replay I have (and I don't remember deleting any, though maybe some really old ones aren't in there), and of the 72 replays that start with "Pylon, Gateway, Assimilator, Pylon, Cybernetics Core, Assimilator," only 1 of them has the next building as a nexus (which lost). Only 7 have a nexus after another pylon, and the record there is 3-4.
So okay, perhaps ~10% of the time people will expand immediately after going double gas, but it's not common, and it is a build that doesn't win very much.
On February 04 2012 15:10 ETisME wrote: There are some errors of statistics to be honest: the pylon gateway cybercore one doesn't show what the toss is going for and what match ups. it is the most consistent protoss opening that is done (but not showing what tech route and what comes next, for all we know, it isn't a nexus first or FFE) and since toss has one of the worst matchup win rate, obviously this opening, be it 4 gate, 3 gate robo expo that the opening leads to, will have the highest rate of losing.
and since 2 gate is an opening designed to win right off, it shows the actual build can win games, unlike the one above which only shows what opening he is doing and what comes next (the rest of the build that is designed to win the game) would never get a winning rate as high as the 2 gate one
You should notice the 2nd gas is included in that winrate, so the build is for double gas P on 1 base. You can be sure it ISN'T a nexus or forge FE after this... dunno how you could possibly think a protoss is going to open double gas and then forge FE LOL
so judging by that is it safe to say thats very indicative of 1 base stargate play? for the most part? If not can you explain why? I suck at the game so try not to flame if i'm way off pwease. want to keep my pillow tear free tonight :D
Depends on many, many factors. First off you have to take into account matchup: PvP - could mean anything. Double gas on 1base is very standard. For the most part you can rule out 4gate, though sometimes people will fake you with 2nd gas and still 4gate. In PvP specifically you can probably also rule out 1gate nexus for the mast part as you'd do that build with 1 gas. Could be standard robo build into immortals or colossi, could be stargate build, could be blink+obs, could be blink, could be dt, could be 3gate expand. The list goes on, because double gas is pretty much required for toss especially in that matchup.
PvT - could be 3gate voidray, could be blink+obs, could be 4gate blink, could be 3gate sentry expand, could be 1 gate robo sentry+obs expand, could be a phoenix build, could be DT, could be a wacky templar rush, could be double gas 4gate. I'm sure there are some I am missing
PvZ - well you know they're not forge FEing. The most common occurrence is probably 3gate sentry expand, but of course it could still be 1gate sentry expand, some 1-base blink build, DT expand, 1gate voidray expand, some weird robo warp prism + sentry build, and of course there are plenty of other alternatives that are less common like 1-base colossus or something silly like that.
There are so many more factors that go into what build your opponent is doing than just if they get double gas or not, but it certainly helps narrow it down. Keep tabs not only of gas but other indicative elements such as the amount of chrono saved, where the chrono is being used, and how many sentries are made early game. If your opponent gets double gas and you scout a low number of sentries, you better prepare for the possibility of DT or stargate (or blink if u see a high number of stalkers).
Pros say map stats are bullshit until they are in tournaments.
Maps goes into tournaments, pros say it's bullshit because it's not Koreans pros.
Maps go into Korean tournaments, pros claim it's bullshit because they need thousands of games.
Playhem released stats with thousands of games. Pros claim it's bullshit because it's not thousands of games with only the very highest level of players.........
Sigh...............
But then they applaud tournaments for using Blizzard maps that go no pro testing whatsoever and are not designed for tournament play....
This is an endless circle I swear. They could probably release stats with 15,000 games between MvP and NesTea and pros would say it's bullshit because it's not Flash vs. Jaedong..............
On February 04 2012 16:13 Diamond wrote: Pros say map stats are bullshit until they are in tournaments.
Maps goes into tournaments, pros say it's bullshit because it's not Koreans pros.
Maps go into Korean tournaments, pros claim it's bullshit because they need thousands of games.
Playhem released stats with thousands of games. Pros claim it's bullshit because it's not thousands of games with only the very highest level of players.........
Sigh...............
But then they applaud tournaments for using Blizzard maps that go no pro testing whatsoever and are not designed for tournament play....
This is an endless circle I swear. They could probably release stats with 15,000 games between MvP and NesTea and pros would say it's bullshit because it's not Flash vs. Jaedong..............
This isn't anything new. Pros want to play on old maps because that's what they've already practiced.
The only problem here is that the community tends to listen to pros. In this particular instance (furthering the map scene), listening to pros is actually hugely detrimental to progress.
As you have said before diamond, pros were adamantly fighting against the removal of steppes of war for quite some time (just as an example). If we listen to pros we get absolutely nowhere with mapmaking.
On February 04 2012 15:07 KawaiiRice wrote: these stats don't really reflect balance and random masters players don't even play the game right anyway.
i.e.: Metal should be way more z favored, and antiga should be incredibly terran favored. Dual Sight and Shakuras should be more T favored TvZ too. Can't really comment on the other matchups.
Regardless, nice post. <3 Playhem
Define "playing the game right." Not to be a dick just want to understand your point of view a bit better. I suck so i'm not going to get into anything with you but i would assume Most masters players and above have a decent idea of how to play the game hence why they're in said league. Generally speaking.
But i'm silver so what do i know.
you can get into masters without having deep knowledge of what exactly you're doing and just macroing. your assumption is false because most masters don't really know and masters itself doesn't actually mean anything, it's just a stupid label blizzard decided to place on people to make them feel good.
if masters players knew how to play properly then top players wouldn't mow down ladder with ridiculous winrates.
It would very stupid of Playhem to balance their maps according to the top 0.0001%. They don't have the prize money to draw those kinds of players anyway. Balancing their maps to the 0.0001% would unbalance all the maps for the 99.9999%
Now, if this were a thread about GSL then I'd certainly agree with you, because everyone in the GSL is at that highest level. But I believe that attitude is too elitist for a tournament like playhem.
Haha I thought that was an awesome response. You're right, these games are for the rest of the players at the top levels besides that group of top ~~300 players that take this game as a job. And sometimes that's much more interesting anyway.
Edit: And anyway, there have been COUNTLESS times I've seen top GM players losing to players who are in the Masters league on streams.
Antiga mains are so vulnerable to drops, which is probably why TvP is so favored there. Tal-Darim's awesome racial balance is offset by its stupidity in the PvP matchup, and to a lesser extent, ZvZ.
On February 04 2012 15:07 KawaiiRice wrote: these stats don't really reflect balance and random masters players don't even play the game right anyway.
i.e.: Metal should be way more z favored, and antiga should be incredibly terran favored. Dual Sight and Shakuras should be more T favored TvZ too. Can't really comment on the other matchups.
Regardless, nice post. <3 Playhem
Define "playing the game right." Not to be a dick just want to understand your point of view a bit better. I suck so i'm not going to get into anything with you but i would assume Most masters players and above have a decent idea of how to play the game hence why they're in said league. Generally speaking.
But i'm silver so what do i know.
you can get into masters without having deep knowledge of what exactly you're doing and just macroing. your assumption is false because most masters don't really know and masters itself doesn't actually mean anything, it's just a stupid label blizzard decided to place on people to make them feel good.
if masters players knew how to play properly then top players wouldn't mow down ladder with ridiculous winrates.
It would very stupid of Playhem to balance their maps according to the top 0.0001%. They don't have the prize money to draw those kinds of players anyway. Balancing their maps to the 0.0001% would unbalance all the maps for the 99.9999%
Now, if this were a thread about GSL then I'd certainly agree with you, because everyone in the GSL is at that highest level. But I believe that attitude is too elitist for a tournament like playhem.
Regardless of how a map is balanced for the highest level it will most likely not affect lower level play, as seen by some of the stats here...... I don't see how trying to make a map 50% win/loss (which btw would amaze me if this ever gets accomplished) is going to "unbalance" a map for the 99%. That's an extremely large stretch. I argue it would have little to no effect at all to winrates for other players.
The reason I'm being sort of critical here is because playhem is trying to use these stats to represent some sort of statement about map balance when they also include thousands of games where people lose because of millions of mistakes that have nothing to do with map balance because they aren't as good.
On February 04 2012 15:07 KawaiiRice wrote: these stats don't really reflect balance and random masters players don't even play the game right anyway.
i.e.: Metal should be way more z favored, and antiga should be incredibly terran favored. Dual Sight and Shakuras should be more T favored TvZ too. Can't really comment on the other matchups.
Regardless, nice post. <3 Playhem
Define "playing the game right." Not to be a dick just want to understand your point of view a bit better. I suck so i'm not going to get into anything with you but i would assume Most masters players and above have a decent idea of how to play the game hence why they're in said league. Generally speaking.
But i'm silver so what do i know.
you can get into masters without having deep knowledge of what exactly you're doing and just macroing. your assumption is false because most masters don't really know and masters itself doesn't actually mean anything, it's just a stupid label blizzard decided to place on people to make them feel good.
if masters players knew how to play properly then top players wouldn't mow down ladder with ridiculous winrates.
So what do you have to gain from reading TL? And how is your brain more suited to apply what you learn from here than said lowly "masters"?
On February 04 2012 15:07 KawaiiRice wrote: these stats don't really reflect balance and random masters players don't even play the game right anyway.
i.e.: Metal should be way more z favored, and antiga should be incredibly terran favored. Dual Sight and Shakuras should be more T favored TvZ too. Can't really comment on the other matchups.
Regardless, nice post. <3 Playhem
Define "playing the game right." Not to be a dick just want to understand your point of view a bit better. I suck so i'm not going to get into anything with you but i would assume Most masters players and above have a decent idea of how to play the game hence why they're in said league. Generally speaking.
But i'm silver so what do i know.
you can get into masters without having deep knowledge of what exactly you're doing and just macroing. your assumption is false because most masters don't really know and masters itself doesn't actually mean anything, it's just a stupid label blizzard decided to place on people to make them feel good.
if masters players knew how to play properly then top players wouldn't mow down ladder with ridiculous winrates.
fair enough, that does make sense. Yeah had to assume cuz the only thing i've been close to top tier at was fps and the leagues were generally a good representation of skill. So the main thing to master doing first is macro and mechanics and then tactics and what not after you got that down? Is that fair to say?
On February 04 2012 16:34 KawaiiRice wrote: The reason I'm being sort of critical here is because playhem is trying to use these stats to represent some sort of statement about map balance when they also include thousands of games where people lose because of millions of mistakes that have nothing to do with map balance because they aren't as good.
The underlying meaning behind the analysis was more along the lines that map design can have an impact on balance, and that neither maps nor patches alone will lead to a perfect game but that changes have to be made in order to eventually find the best balance. Everyone knows how different BW was in the earliest days compared to now, and change is the only thing that will facilitate SC2's progress toward a better game.
While you're right that mistakes are factored in just as the map balance is, with a large enough sample size you should have mistakes factoring in roughly evenly for all maps/matchups unless something else is influencing some races to be more prone to game-losing mistakes. That might be an entirely different/interesting issue about how the racial mechanics work in SC2.
tl;dr, Tournaments should try out some new stuff and not just cling to old maps.
On February 04 2012 15:07 KawaiiRice wrote: these stats don't really reflect balance and random masters players don't even play the game right anyway.
i.e.: Metal should be way more z favored, and antiga should be incredibly terran favored. Dual Sight and Shakuras should be more T favored TvZ too. Can't really comment on the other matchups.
Regardless, nice post. <3 Playhem
Define "playing the game right." Not to be a dick just want to understand your point of view a bit better. I suck so i'm not going to get into anything with you but i would assume Most masters players and above have a decent idea of how to play the game hence why they're in said league. Generally speaking.
But i'm silver so what do i know.
you can get into masters without having deep knowledge of what exactly you're doing and just macroing. your assumption is false because most masters don't really know and masters itself doesn't actually mean anything, it's just a stupid label blizzard decided to place on people to make them feel good.
if masters players knew how to play properly then top players wouldn't mow down ladder with ridiculous winrates.
It would very stupid of Playhem to balance their maps according to the top 0.0001%. They don't have the prize money to draw those kinds of players anyway. Balancing their maps to the 0.0001% would unbalance all the maps for the 99.9999%
Now, if this were a thread about GSL then I'd certainly agree with you, because everyone in the GSL is at that highest level. But I believe that attitude is too elitist for a tournament like playhem.
Regardless of how a map is balanced for the highest level it will most likely not affect lower level play, as seen by some of the stats here...... I don't see how trying to make a map 50% win/loss (which btw would amaze me if this ever gets accomplished) is going to "unbalance" a map for the 99%. That's an extremely large stretch. I argue it would have little to no effect at all to winrates for other players.
The reason I'm being sort of critical here is because playhem is trying to use these stats to represent some sort of statement about map balance when they also include thousands of games where people lose because of millions of mistakes that have nothing to do with map balance because they aren't as good.
On February 04 2012 15:07 KawaiiRice wrote: these stats don't really reflect balance and random masters players don't even play the game right anyway.
i.e.: Metal should be way more z favored, and antiga should be incredibly terran favored. Dual Sight and Shakuras should be more T favored TvZ too. Can't really comment on the other matchups.
Regardless, nice post. <3 Playhem
Define "playing the game right." Not to be a dick just want to understand your point of view a bit better. I suck so i'm not going to get into anything with you but i would assume Most masters players and above have a decent idea of how to play the game hence why they're in said league. Generally speaking.
But i'm silver so what do i know.
you can get into masters without having deep knowledge of what exactly you're doing and just macroing. your assumption is false because most masters don't really know and masters itself doesn't actually mean anything, it's just a stupid label blizzard decided to place on people to make them feel good.
if masters players knew how to play properly then top players wouldn't mow down ladder with ridiculous winrates.
So what do you have to gain from reading TL? And how is your brain more suited to apply what you learn from here than said lowly "masters"?
stfu
someone lost to a masters player who proxy 2 gated.
Aren't you like 16 years old? Seems like hanging around that angry nerd picnic is having an adverse effect on your impressionable mind.
On February 04 2012 16:34 KawaiiRice wrote: The reason I'm being sort of critical here is because playhem is trying to use these stats to represent some sort of statement about map balance when they also include thousands of games where people lose because of millions of mistakes that have nothing to do with map balance because they aren't as good.
The underlying meaning behind the analysis was more along the lines that map design can have an impact on balance, and that neither maps nor patches alone will lead to a perfect game but that changes have to be made in order to eventually find the best balance. Everyone knows how different BW was in the earliest days compared to now, and change is the only thing that will facilitate SC2's progress toward a better game.
While you're right that mistakes are factored in just as the map balance is, with a large enough sample size you should have mistakes factoring in roughly evenly for all maps/matchups unless something else is influencing some races to be more prone to game-losing mistakes. That might be an entirely different/interesting issue about how the racial mechanics work in SC2.
tl;dr, Tournaments should try out some new stuff and not just cling to old maps.
So essentially, at a master-grandmaster level, protoss is kind of a joke on most of these maps? Cant really say im surprised. I am surprised about antiga though, I would have figured that would be more PvT favored.
It's sad that PvP is so... dumb on Tal'Darim Altar. The balance in the non-mirror matchups looks great and I really enjoy playing the map. But every time I get a PvP on it I just go straight to frown town.
Great work compiling all of this data, and thanks for all of the content in general from Playhem.
You have to look at when the game ended (early game, mid game, late game) and the relative skill level of players (low-masters vs koreans like Violet or Zenex players means absolutely nothing, low masters will have 5% winrate on any map)
Thanks for this regardless though, nice statistics.
On February 04 2012 14:57 godulous wrote: Tal'Darim Altar LE get's a special shout out for being possibly the most balanced map currently in our pool, it scored nearly 50% win rates for every match up and overall race win rates.
I have to disagreee with this. The most balanced map is edit: mistake in my calculations (damn copy/paste) MLG metalopolis and Tal'Darim Altar LE are tied #1
You are making mistake of thinking closest to 50% is the most balanced, when you really should be thinking that closest to average win% of the matchup is most balanced. Also consider the more further win% is from average, more meaningful it is. So you should get something like ABS((TvPa-TvPm)^2)+ABS((PvZa-TvZm)^2)+ABS((ZvTa-ZvTm)^2) where TvPa is average win% for TvP and TvPm is the win% for map and so on. Lowest score map wins.
Think about situation where for example we have start that gives 90% winrate for some matchup. Only way to get even close 50% winrate for that matchup is to make teh map hevily unbalanced for other end.
On February 04 2012 16:57 cjin wrote: Think about situation where for example we have start that gives 90% winrate for some matchup. Only way to get even close 50% winrate for that matchup is to make teh map hevily unbalanced for other end.
That was the whole point of this thread. Once you consider a map as one of the dials you can tweak in balancing the game, you can fine tune things like averages that aren't 50% (which they should ideally be). In that case, a 'heavily unbalanced map' by your standards would be a good thing. If unit abilities and build times/costs can be adjusted to correct win rates, why can't maps?
On February 04 2012 17:00 T.O.P. wrote: How about only taking top 8 of each playhem and analyzing those replays.
I'll see if I can do this tomorrow and post my findings (though the sample size will be much smaller).
Ingenius playhem . I wish tournaments adjusted the map pool away from certain maps that are considered imbalanced but with such a large sample size tournaments can naturally do this while being nearly entirely subjective. hats off to Playhem ;D
On February 04 2012 14:57 godulous wrote: Tal'Darim Altar LE get's a special shout out for being possibly the most balanced map currently in our pool, it scored nearly 50% win rates for every match up and overall race win rates.
I have to disagreee with this. The most balanced map is edit: mistake in my calculations (damn copy/paste) MLG metalopolis and Tal'Darim Altar LE are tied #1
You are making mistake of thinking closest to 50% is the most balanced, when you really should be thinking that closest to average win% of the matchup is most balanced. Also consider the more further win% is from average, more meaningful it is. So you should get something like ABS((TvPa-TvPm)^2)+ABS((PvZa-TvZm)^2)+ABS((ZvTa-ZvTm)^2) where TvPa is average win% for TvP and TvPm is the win% for map and so on. Lowest score map wins.
Think about situation where for example we have start that gives 90% winrate for some matchup. Only way to get even close 50% winrate for that matchup is to make teh map hevily unbalanced for other end.
I'm sorry, I believe your logic is flawed here. You don't want to be closer to the average winrate of the matchup. This is relating it to other maps. This is unimportant. What matters are the races and how they interact on a single map. How they interact on other maps is irrelevant. What you ideally want is a map that has a 50% winrate across the board, because regardless of current trends in balance, what you want is balance on any given map. If that map is balanced, who cares if the overall average protoss winrate in the matchup of PvZ is 55% for example? You don't want the map to show 55% in that winrate, you still want it to show 50% in that winrate so that you have a balanced match to play.
On February 04 2012 17:02 godulous wrote: That was the whole point of this thread. Once you consider a map as one of the dials you can tweak in balancing the game, you can fine tune things like averages that aren't 50% (which they should ideally be). In that case, a 'heavily unbalanced map' by your standards would be a good thing. If unit abilities and build times/costs can be adjusted to correct win rates, why can't maps?
Blizzard is the one making changes to unit abilities and build times/costs, but not to tournament maps. If Blizzard makes a change into something, you would first need to play tons of games on new settings to even know how much effect the change had to win%.
Also how you would count playerskill? if we could mesure playerskill with lets say 1-100 and average skill of T turns to be 60 and average skill of Z 40, if you would balance maps for 50% winrate it would mean that T60 would equal Z40 in win%, how far they would get in tournaments. and Z40 would crush kill ´n destroy T40 like a bug. Ideally T40 and Z40 should get just as far.
As interesting as the map stats are, I'm also intrigued by the word cloud. Looks like there's an interesting variety of bm going around, though the majority of talk is gm or neutral.
That doesn't even count offensive gg, and I'm sure I missed some words. : P
On February 04 2012 17:33 Ryuu314 wrote: interesting how protoss has the saddest winrates while at top top level protoss is doing decently well.
if we take top top level to be code S, protoss is doing really really really badly.
A few fun facts:
A protoss hasnt been in a Code S final since May 2011. (in 3-4 months it will be a year)
in the last 2 GSL's of 2011, there were 2 protoss in the gsl ro8, when you combine the seasons (2 in one, none in the other), each season had 5 protoss in the ro32.
This season was a little better, 8 in the ro32 (thats 1/4 the players)
There have been 3 protoss GSL finalists (code s + special event + open seasons), compare that to 20 terrans and 11 zergs. (the chance of this happening has a decimal point followed by lots and lots of zeros - im embarrased to say ive forgotten how to figure it out)
So, in terms of statistics, at the top top level, protoss is doing unimaginably poorly, and has been since the beginning of GSL.
While statistics do not prove balance or imbalance in anyway, and i am not stating protoss is underpowered, the statement that protoss are doing decently well at top top level simply is as far from the truth as possible.
How can anyone argue that this is not a good sample for analysis?
It is high-level players (I know they aren't S-Class, but they are high-level nonetheless) over THOUSANDS AND THOUSANDS of tournament games across all matchups and maps. Maybe this doesn't say a whole lot about balance in the GSL, but it certainly indicates the balance in general.
I feel so sad for protoss from those statistics. And after all the hurdle and it being the worst race on all maps the only successful build it is the two gate lol.
On February 04 2012 17:41 jubil wrote: As interesting as the map stats are, I'm also intrigued by the word cloud. Looks like there's an interesting variety of bm going around, though the majority of talk is gm or neutral.
That doesn't even count offensive gg, and I'm sure I missed some words. : P
Haha, yeah, I noticed that as well. Although most of the circled words seem to be of low occurence as their font size is baseline. Anyways, cool stuff.
How big is the variance in % of wrong maps being played? (I.e. playing on ESL iCCup Sanshorn Mist AE by ESL, which is outdated, instead of playing on ESV Sanshorn Mist AE by SUPEROUMAN, which is uptodate. Or playing on GSL Dual Sight by GomTV (outdated) instead of GSL_Dual Sight by TeamClash (uptodate) ) Also, thanks to how Blizzard's map system works, there are rolling changes to the maps. MLG Antiga and MLG Dual Sight were changed to no-gold bases at some point. MLG Metal features no more gold bases since last week. And MLG Shakuras was the outdated version with the backdoor rocks for the last 10 months on EU. This stuff will alter the statistics and is very hard to keep track of.
On February 04 2012 15:01 0kz wrote: wow overall maps look pretty balanced..
what? out of the entire map pool only ONE map is balanced. the rest have 1 or more matchups that aren't even remotely balanced.
Edit: and this is the issue with trying to balance the game. the maps play such a huge factor it's really difficult to say what's "imbalanced" or not.
One could say a 5% margin is pretty balanced considering how the game could be. I think with a margin that is ~5% the better player should still win most of the time, unless they are not considerably better etc. Obviously it isn't perfect, and obviously you want every map to be 50-50-50, but this is pretty good in my opinion.
On February 04 2012 15:21 Yoshi Kirishima wrote: holy shit, the stats for shak and taldarim are a lot closer than I'd think =O
I'm surprised the TvP on antiga is so high... isn't that supposed to be an easier map as protoss since you can take 3 base easily?
Same with TvP on Dual Sight, though I'm guessing terrans usually do some kind of early aggression or 1-1-1 all-in on that map to get an easy win.
anyways thanks a lot for sharing these!
Antiga is a great map for terran against toss. All the early agreesion on that map can be very effective and late game once terran has a gold, it would be super easy for terran. The gsl version one is balanced.
I think this shows us how much a map can contribute to game balance. For example, a few months ago, 11/11 opening against zergs and 111 against toss are extremely imbalanced, but maps encourage terrans to do that as well. Now when I look at it, the only tournament you can see 111 all the time is IEM simply because of the maps they are using.
I really hope all the major events can learn from gsl and pay more attention to the maps they are using. Of course, do not learn gsl by using Cross Fire.
Wow, amazing stats. Interesting to see that Tal'Darim is such a great map, I mean it's always been one of my favourites, but I know so many people that hate it, so it's a bit surprising none the less.
Regardless, massive props for doing this, Playhem ftw
On February 04 2012 15:07 KawaiiRice wrote: these stats don't really reflect balance and random masters players don't even play the game right anyway.
Really?
First of all they're mid/high masters and GM. They're no GSL pro players, but they do know their way around. Also it's not like they're playing against GSL pro players, they're playing against equal skill level.
Also, by your own logic, your opinion and experience shouldn't matter because you're very far from top players and have no idea how to play the game compared to them.
anyone can enter these playhem OPENS. Many many mid master players play, and the game may be imba at mid master level, but balanced at GM level, and imba for the other race at GSL level, meaning these stats would mean nothing. he has a valid point.
Big sample size but the avg game lenght is just horribly short. I wouldn't use the winrates for anything else but checking which maps are good to cheese/all in on. Still nice to see how big differences there are between maps and nice to see those taldarim winrates.
Chat cloud is the most surprising thing about these stats. How can faggot and l2p be so far from the middle!? :D But seriously, apparently tournament play is more mannered than ladder play on avg, where most of the chat is telling opponent how bad they suck (yeah, even at GM/high master).
But clearly shows the flaws of some maps. Meta has appeared terrible for toss ever since FFE became standard that "has" to be done. Shattered being terran favored overall, and so on. Indeed surprising taldarim is supposedly the most balanced one.
On February 04 2012 18:51 Entteri wrote: Big sample size but the avg game lenght is just horribly short. I wouldn't use the winrates for anything else but checking which maps are good to cheese/all in on. Still nice to see how big differences there are between maps and nice to see those taldarim winrates.
Game length is given in real time. So multiply it by *1.38 for ingame length. I don't think 11min real time is that low for average game. 15min ingame game has often got to 3v3base situation, and that's only average. For every 10min game there's 20min game etc.
This only strengthens the idea that zergs early third hatch is a huge problem in PvZ, as Tal'darim, being the only map where this is hard to pull off, is the most balanced in that matchup (and all other matchups for that matter).
Such a shame the most balanced map is the worst map for PvP though =|
I`m glad Taldarim has such a balanced race-win-percentage. Also the fact that taldarim games aren`t longer in average is surprising. thanks for sharing
Have you considered providing these 31.5k replays to that scientific study that was looking at skill development? I don't know if they are exactly what they're looking for, but that's a lot of yummy, sweet data.
Would it be possible to get graphs of the time series of winrates of each matchup per map? We have a 5 month average, would be curious to see how this varied more specifically according to metagame/patch
On February 04 2012 15:07 KawaiiRice wrote: these stats don't really reflect balance
Um, yes they do, except only for the lower-than-pro tier players, who kind of constitute the majority of the playerbase. What's wrong with tournament organizers optimizing the balance of their tournament according to the statistics of their target demographic?
On February 04 2012 15:07 KawaiiRice wrote: random masters players don't even play the game right anyway.
And this invalidates the relevancy of this data in this particular context, how, exactly? Also, lol elitist.
On February 04 2012 15:07 KawaiiRice wrote: these stats don't really reflect balance
Um, yes they do, except only for the lower-than-pro tier players, who kind of constitute the majority of the playerbase. What's wrong with tournament organizers optimizing the balance of their tournament according to the statistics of their target demographic?
On February 04 2012 19:07 Excludos wrote: This only strengthens the idea that zergs early third hatch is a huge problem in PvZ, as Tal'darim, being the only map where this is hard to pull off, is the most balanced in that matchup (and all other matchups for that matter).
Such a shame the most balanced map is the worst map for PvP though =|
I disagree. I think the lack of a zerg third on taldarim (combined with the shape of the protoss natural) puts zerg behind. However, the strength of mutalisks on taldarim puts them even again.
This is purely opinion though (no evidence at all).
On February 04 2012 19:07 Excludos wrote: This only strengthens the idea that zergs early third hatch is a huge problem in PvZ, as Tal'darim, being the only map where this is hard to pull off, is the most balanced in that matchup (and all other matchups for that matter).
Such a shame the most balanced map is the worst map for PvP though =|
I disagree. I think the lack of a zerg third on taldarim (combined with the shape of the protoss natural) puts zerg behind. However, the strength of mutalisks on taldarim puts them even again.
This is purely opinion though (no evidence at all).
Well, if mutalisks is able to put them ahead when they're apparantly "behind", doesn't that agree to the fact that mutalisks of 3 base is too strong? As then zergs are ahead (or even if you like) and get more ahead because of the mutalisks that would otherwise have just gotten them even.
wow well done really nice statistics but i wouldnt agree that taldarim is the most balanced map just because it has winrates that are so close together since on TDA a lot depends on spawning positions (also it has least games played which is an indicator of people not liking this map too much)
great thread nonetheless
and
On February 04 2012 15:07 KawaiiRice wrote: these stats don't really reflect balance and random masters players don't even play the game right anyway.
i.e.: Metal should be way more z favored, and antiga should be incredibly terran favored. Dual Sight and Shakuras should be more T favored TvZ too. Can't really comment on the other matchups.
Regardless, nice post. <3 Playhem
dual sight T favored?
im not nearly as good as you but from what i have seen this map is quite good for zerg
On February 04 2012 19:07 Excludos wrote: This only strengthens the idea that zergs early third hatch is a huge problem in PvZ, as Tal'darim, being the only map where this is hard to pull off, is the most balanced in that matchup (and all other matchups for that matter).
Such a shame the most balanced map is the worst map for PvP though =|
I disagree. I think the lack of a zerg third on taldarim (combined with the shape of the protoss natural) puts zerg behind. However, the strength of mutalisks on taldarim puts them even again.
This is purely opinion though (no evidence at all).
Well, if mutalisks is able to put them ahead when they're apparantly "behind", doesn't that agree to the fact that mutalisks of 3 base is too strong? As then zergs are ahead (or even if you like) and get more ahead because of the mutalisks that would otherwise have just gotten them even.
ever tried fast muta vs 7 gate?
doest work too well if we are entirely honest right?
On February 04 2012 19:07 Excludos wrote: This only strengthens the idea that zergs early third hatch is a huge problem in PvZ, as Tal'darim, being the only map where this is hard to pull off, is the most balanced in that matchup (and all other matchups for that matter).
Such a shame the most balanced map is the worst map for PvP though =|
I disagree. I think the lack of a zerg third on taldarim (combined with the shape of the protoss natural) puts zerg behind. However, the strength of mutalisks on taldarim puts them even again.
This is purely opinion though (no evidence at all).
Well, if mutalisks is able to put them ahead when they're apparantly "behind", doesn't that agree to the fact that mutalisks of 3 base is too strong? As then zergs are ahead (or even if you like) and get more ahead because of the mutalisks that would otherwise have just gotten them even.
No, I was specifically talking about mutalisks on taldarim.
If you look at Shattered Temple, which also has rocks on the third, then I think it is fair to say that the map is balanced for other reasons. In this case it is the wide choke to the natural and the long distance from the bottom of the ramp to the natural minerals.
Every map has pros and cons for different reasons.
On February 04 2012 19:35 Micket wrote: Incorrect. This doesn't show balance, just win rates of lower level play on maps.
Yes and those win rates of "lower level play" encapsulate the balance in this particular tournament, under a specific set of circumstances and a specific player pool. What exactly is so hard to understand about that?
On February 04 2012 16:13 Diamond wrote: Pros say map stats are bullshit until they are in tournaments.
Maps goes into tournaments, pros say it's bullshit because it's not Koreans pros.
Maps go into Korean tournaments, pros claim it's bullshit because they need thousands of games.
Playhem released stats with thousands of games. Pros claim it's bullshit because it's not thousands of games with only the very highest level of players.........
Sigh...............
But then they applaud tournaments for using Blizzard maps that go no pro testing whatsoever and are not designed for tournament play....
This is an endless circle I swear. They could probably release stats with 15,000 games between MvP and NesTea and pros would say it's bullshit because it's not Flash vs. Jaedong..............
I agree with Diamond, I've been noticing this trend, every time there's a release of statistics that suit what the previous group whined about, and when they get their specifications for replay pool met, they complain about something else.
On February 04 2012 15:02 lee365 wrote: Proxy 2 gate represent. 62% winrate might make it one of the most effective cheeses in the game
It doesn't have to be offensive proxy 2 gates. Probably a lot of defensive ones to counter scouted 6 or 7 pools etc. Which is pretty easy to win once the rush has been stopped.
On February 04 2012 15:02 lee365 wrote: Proxy 2 gate represent. 62% winrate might make it one of the most effective cheeses in the game
It doesn't have to be offensive proxy 2 gates. Probably a lot of defensive ones to counter scouted 6 or 7 pools etc. Which is pretty easy to win once the rush has been stopped.
I think going 2 gate vs zerg is just the stronger opening, especially on big maps. On big maps, many zerg (including myself) cut a particular corner: the drone scout. Since its such a huge map, we assume that saving that drone and just sending the 2 initial lings would be enough, just to encounter 3 zealots on their way to your base.
These 3-4 zealots force so many lings if they manage to get into the mineral line, its not even funny, and building a spine crawler on reaction is just more economical damage to the zerg, since he still has to build lings, otherwise the zealots just kill the building spine crawler. If you go for gas first then pool into speedling expand, and drone scout the 2 gate, its better, since you can just catch those zealots in the open, but its still roughly 4 drones less than versus a standard gateway into cyber opening.
Watching the replays of those games, zerg falls behind even if they kill those zealots with minimal amount of zerglings. Usually if a zerg goes gas and then pool, and then expand on 18, they are stuck at 18 supply for quite some time, since 300 for hatchery, 150 for queen, 100/100 for gas is very expensive and doesnt let you sneak in that 100 mineral overlord that you need to get some more supply. If you look at worker count during that stage, protoss is always ahead, but vs a standard gateway cyber opening, zerg follows this up with double drones from both hatcheries and immediately catches up in worker count and then surpasses the protoss.
If the protoss went 2 gateway, they force the zerg to have 4 less drones during that 18 supply period where they usually are behind in workers anyways, and that just puts zerg in a terrible position in my opinion.
I think going 2 gate vs zerg is like the 11/11 rax against zerg, almost always viable, even if it you dont do much with it, zerg has to react and build lings, which almost definitely puts him in a worse position than if you wouldnt have gone 2 gate.
I really dont believe the 2 gate statistics from the playhem stem from proxy gates, since proxy gates suck balls.
People seem to be surprised by the "short" average of the games, but...
Watching various tournaments, it seems that the long "epic" games are less common than we tend to think. I mean, for the top level players, I'm surprised to see games that run for 30+ minutes - even when they are macro games without cheese. Granted, this is just perception and opinion. But players in tournaments shouldn't really be trying to draw games out - they have games to play, and if they can shorten their games leading to the finals they will be better off when the get towards the end of the tournament in terms of game fatigue. (People that slog their way through open brackets to get into the top 8 get props for me, especially when they show up to face "fresher" opponents that only were in pool play or were direct seeds.)
Also, since it's an "average" game length, you have to realize that the time for each game just means that there are more ,or shorter, "quick" games and fewer, or longer, "long" games. As even the pros can tell you, good cheese sometimes is an effective way of just bashing your way up to the point where you're playing opponents that will take all your skill to beat.
(Then again, my average personal game times are pretty short, but I'm faaar lower league and the first push from a 3rax can still end a game outright in 9 minutes.)
On February 04 2012 19:07 Excludos wrote: This only strengthens the idea that zergs early third hatch is a huge problem in PvZ, as Tal'darim, being the only map where this is hard to pull off, is the most balanced in that matchup (and all other matchups for that matter).
Such a shame the most balanced map is the worst map for PvP though =|
I disagree. I think the lack of a zerg third on taldarim (combined with the shape of the protoss natural) puts zerg behind. However, the strength of mutalisks on taldarim puts them even again.
This is purely opinion though (no evidence at all).
Well, if mutalisks is able to put them ahead when they're apparantly "behind", doesn't that agree to the fact that mutalisks of 3 base is too strong? As then zergs are ahead (or even if you like) and get more ahead because of the mutalisks that would otherwise have just gotten them even.
ever tried fast muta vs 7 gate?
doest work too well if we are entirely honest right?
2base fast mutas can hold 6 or 7gate if the zerg knows its coming. It will be tight, and most likely he will lose the natural. But then he has 10+ mutalisks left and can just rape both of protosses mineral lines afterwards. (Also depends on the type of allin from the toss. If its zealot-stalker, its harder. But since lingspeed will almost always be done by then, most opt for a zealot-sentry force)
On February 04 2012 19:07 Excludos wrote: This only strengthens the idea that zergs early third hatch is a huge problem in PvZ, as Tal'darim, being the only map where this is hard to pull off, is the most balanced in that matchup (and all other matchups for that matter).
Such a shame the most balanced map is the worst map for PvP though =|
I disagree. I think the lack of a zerg third on taldarim (combined with the shape of the protoss natural) puts zerg behind. However, the strength of mutalisks on taldarim puts them even again.
This is purely opinion though (no evidence at all).
Well, if mutalisks is able to put them ahead when they're apparantly "behind", doesn't that agree to the fact that mutalisks of 3 base is too strong? As then zergs are ahead (or even if you like) and get more ahead because of the mutalisks that would otherwise have just gotten them even.
No, I was specifically talking about mutalisks on taldarim.
If you look at Shattered Temple, which also has rocks on the third, then I think it is fair to say that the map is balanced for other reasons. In this case it is the wide choke to the natural and the long distance from the bottom of the ramp to the natural minerals.
Every map has pros and cons for different reasons.
Shattered would have a lot better statistics in PvZ if the ledge behind your mineral line wasn't there. That + the fact that the zerg can take a quick, uncontested, gold, skews the numbers back. I still hold my belief that the quick third hatch, which is normal on any maps that allow it, allows for a massive amount of muta numbers which is near impossible to deal with as toss.
I really don't agree with people who say that only GSL players are the ones who play the game well, so we should look to them for balance. The reason is that the very top players can use units in a way 99% of people cannot. For example, for most people banelings are an extremely effective counter to marines, but MKP can micro marines so well that you would say that marines are the counter to banelings. If you took this into consideration then sure the game would be balanced better for GSL, but what about the 99.9% of the rest of the people that play and enjoy SC2?
For this reason I think that the survey presented here is a good sample, because no matter what some elitists may think, masters players all have some ability in the game, and are a good representation of how most people will experience the game in the real world (ie on ladder).
i dont see how this is an accurate representation of balance. the more a player wins, the more he gets to play. can we see the results from the first round only of playhem tournaments?
wouldn't have expected the balance so good at the playhem tournaments, but they are always nice to watch and thanks for the effort. And haha 2 gate so strong, my standard opening. But i doubt the next steps are close to mine.
PS: i think game balance is unimportant, whats important is that maps are balanced for the players in the tournament. So if those turn close to 50% then i think the tournament is doing a really good job. (though if the game is strongly imbalanced there will be an underrepresentation of a race)
On February 04 2012 22:10 deathly rat wrote: I really don't agree with people who say that only GSL players are the ones who play the game well, so we should look to them for balance. The reason is that the very top players can use units in a way 99% of people cannot. For example, for most people banelings are an extremely effective counter to marines, but MKP can micro marines so well that you would say that marines are the counter to banelings. If you took this into consideration then sure the game would be balanced better for GSL, but what about the 99.9% of the rest of the people that play and enjoy SC2?
For this reason I think that the survey presented here is a good sample, because no matter what some elitists may think, masters players all have some ability in the game, and are a good representation of how most people will experience the game in the real world (ie on ladder).
Well, if you want to play just ladder then yes.
But if you want to watch tournaments with good players, balance is issue and it shouldn't be just "Fuck, it's not balanced for them anyway, why should i watch that?".
To me balance on ladder is lesser issue. I always lose because my mistake and i can get rid of doing that mistake.
By the way, about the average game time. I assume the OP didn't change the settings in SC2Gears, so its real time, not ingame blizzard time. So u have to multiply the time by about ~1,38 or so to get what the ingame timer would show u
It's like if you spent 30 seconds setting up a joke, but never delivered the punchline. There isn't a thesis, and there's no analysis of the data. There's only a lot of data, but no one has interpreted it. I guess what I'm saying is, who cares?
Hey Playhem guys. And the guys who run daily tourneys. All of you should get in contact with map makers to test there maps! I see you already were using Cloud Kingdom which is good.
On February 04 2012 23:35 -_- wrote: Hey Playhem guys. And the guys who run daily tourneys. All of you should get in contact with map makers to test there maps! I see you already were using Cloud Kingdom which is good.
We do get in contact with mapmakers and leave them feedback through e-mail or here on TL (: It's up to them to actually do something with our data.
On February 04 2012 20:32 felisconcolori wrote: People seem to be surprised by the "short" average of the games, but...
Watching various tournaments, it seems that the long "epic" games are less common than we tend to think. I mean, for the top level players, I'm surprised to see games that run for 30+ minutes - even when they are macro games without cheese. Granted, this is just perception and opinion. But players in tournaments shouldn't really be trying to draw games out - they have games to play, and if they can shorten their games leading to the finals they will be better off when the get towards the end of the tournament in terms of game fatigue. (People that slog their way through open brackets to get into the top 8 get props for me, especially when they show up to face "fresher" opponents that only were in pool play or were direct seeds.)
Also, since it's an "average" game length, you have to realize that the time for each game just means that there are more ,or shorter, "quick" games and fewer, or longer, "long" games. As even the pros can tell you, good cheese sometimes is an effective way of just bashing your way up to the point where you're playing opponents that will take all your skill to beat.
(Then again, my average personal game times are pretty short, but I'm faaar lower league and the first push from a 3rax can still end a game outright in 9 minutes.)
I believe the 10min average is based upon real minutes, not the Blizzard style.
On February 04 2012 23:53 magnaflow wrote: I believe the 10min average is based upon real minutes, not the Blizzard style.
Most definitely real time.
Blizzard needs to get rid of their silly in-game time, I really don't see the point. "Faster" is the default time for ladder so that's not an excuse either.
On February 04 2012 23:53 magnaflow wrote: I believe the 10min average is based upon real minutes, not the Blizzard style.
Most definitely real time.
Blizzard needs to get rid of their silly in-game time, I really don't see the point. "Faster" is the default time for ladder so that's not an excuse either.
Should have done it from the beginning now its too late imo. All our timings and measurements are now in blizzardtime and would throw everyone off if it was changed now. Similar to what happened to APM it was wrong both before and after but now that they changed it, its useless because its not what we've become accustomed to.
And if these charts are in real time then it suddenly becomes hard to figure out how long the games really are because its not in the format we normally see them as.
On February 04 2012 23:53 magnaflow wrote: I believe the 10min average is based upon real minutes, not the Blizzard style.
Most definitely real time.
Blizzard needs to get rid of their silly in-game time, I really don't see the point. "Faster" is the default time for ladder so that's not an excuse either.
Should have done it from the beginning now its too late imo. All our timings and measurements are now in blizzardtime and would throw everyone off if it was changed now. Similar to what happened to APM it was wrong both before and after but now that they changed it, its useless because its not what we've become accustomed to.
And if these charts are in real time then it suddenly becomes hard to figure out how long the games really are because its not in the format we normally see them as.
Better late than never, people will adapt quicker than you think.
Like every other statistic you have to take it with a grain of salt but I feel like this is about as good of a picture as we are going to get. Not necessarily of the 0.0001% of players but like top 3% is good enough for me as an average joe starcraft player.
Hot damn that's almost unbelievably well balanced imo, not to say there are no balance issues but still, cool statistics to see.
Note that 5 moths ago = from september Last real patch was 20 September 2011 The 8 November 2011 made only minor balance changes (lower EMP radius, lower protoss upgrade costs)
On February 04 2012 15:02 lee365 wrote: Proxy 2 gate represent. 62% winrate might make it one of the most effective cheeses in the game
I think the fact its used to so often in PvP vs. a standard opening is what is leading to it being the highest win rate vs standard P opening being one of the worst.
Spurious events are awesome aren't they?
On February 05 2012 01:25 ZisforZerg wrote: who knew TDA was so balanced O>O
Everyone did actually The only thing thats sucks about TDA is the 4 gate v 4 gate scenarios
Excellent post. Really great stuff! TalDarim is a well balanced map according to this, but most people hate it because of PvP and the rocks on third. The thing to take away is that a map can be balanced, but that doesn't necessarily make it good.
After working for Playhem for 3 months, I can't believe how many games I have seen, and finding out that there have been 31.5k plus games played. Long Live Playhem!
Well, even more support for Tal'Darim Altar being my favorite map. Hell, I even like it PvP. Very balanced PvP map, it is. Never strays from 50%.
I don't think these statistics mean TOO much, though. They're over five months, and there were some pretty significant balance changes and metagame shifts in the past five months.
I was surprised at how low/high some of the match ups were on these maps. This is some damn fine work though, I wonder if we could get the bigger tournaments to do the same after an event.
Unfortunately the amount of data given this way is just a tad little, i'd be interested to know the exact numbers you have for each matchup by the way. For example tal darim has roughyl 1500 games in total, how many of those were PvZ? The best way to represent data like this is simple confidence intervals, that way you can see directly how valuable the data are and more importantly, if any differ significantly from 50%. With these amounts of games i suspect that anything over 52 (or below 48) is practically significant at 5% though so there is definately something going on.
One more thing I'd like to see stats of are mirror matches. How often wins the player with more skill (measured by ladder rating or ladder win percentage in the particular matchup for example)? I''ve never seen someone look into this but mirrors DO affect balance overall and shouldn't be ignored. If a mirror is very luck based, let's say PvP for sake of argument, then it is more likely a good player get's knocked out by a worse player. Consequently the race will perform WORSE in other matchups as the mirrors don't act as a selection for better players. Vice versa, if the mirror features almost no luck, for sake of argument let's take TvT, then mirror matches will more consistenly weed out the worse players. As a result that race will perform BETTER in other matchups and skew the statistics in that way. I suspect that for a long time already P has worse stats and T has better stats because of this effect, which seems to be confirmed by the fact that top players tend to have worse PvP percentages then TvT ones.
Overall I'd rather see balancing being done by changing the game (unit stats etc.) then by changing the maps. Balanced maps are often still bad because they don't produce interesting games, for example Tal darim may look very balanced but i'm not a fan of the map at all. PvP is terrible on it, the cliff at the natural means colossus or siege tanks attacks on it are very popular and PvZ is completely focussed around muta play on it. 50% maps are not the holy grail, fun maps with close to 50% are..
On February 05 2012 01:59 Markwerf wrote: The best way to represent data like this is simple confidence intervals, that way you can see directly how valuable the data are and more importantly, if any differ significantly from 50%. With these amounts of games i suspect that anything over 52 (or below 48) is practically significant at 5% though so there is definately something going on.
How are they supposed to give confidence intervals? The numbers give are the exact numbers they have. I see no real reason to try and use statistical analysis to find how big % of all starcraft 2 games being played are actually being played in the playhem daily and then how likely they are to represent the total.
On February 05 2012 01:22 Thombur wrote: Hot damn that's almost unbelievably well balanced imo, not to say there are no balance issues but still, cool statistics to see.
Note that 5 moths ago = from september Last real patch was 20 September 2011 The 8 November 2011 made only minor balance changes (lower EMP radius, lower protoss upgrade costs)
The only stats that surprised me were the Cloud Kingdom and Tal stats.... but I think that's because I have the Idea of 3 months ago Tal in my head... Cloud did legitimately throw me off.... very interesting how Terrans aren't doing so well there.
On February 05 2012 01:59 Markwerf wrote: The best way to represent data like this is simple confidence intervals, that way you can see directly how valuable the data are and more importantly, if any differ significantly from 50%. With these amounts of games i suspect that anything over 52 (or below 48) is practically significant at 5% though so there is definately something going on.
How are they supposed to give confidence intervals? The numbers give are the exact numbers they have. I see no real reason to try and use statistical analysis to find how big % of all starcraft 2 games being played are actually being played in the playhem daily and then how likely they are to represent the total.
the small problem with these numbers is that you don't know exactly how many games they are based on, confidence intervals are just the (generally agreed upon) clearest way to show the data and some insight in their reliability at the same time. You can easily make confidence intervals of these numbers based on a t-statistic, you don't really need to know how big % of all sc2 games are played in the daily as you can safely assume it's a irrelevant small number (thus you don't need to correct).
The minor problem I have now is that I don't know how many games for example the numbers for tal darim are based on. I only know the total of those matchups plus the mirrors is around 1500.. Confidence intervals are easy to calculate and only provide more info..
The average game lengths seem rather short to me. Is that due to a majority of these being EU where there are a lot more hard timings or are my games just longer than average?
On February 05 2012 03:57 LilTip wrote: The average game lengths seem rather short to me. Is that due to a majority of these being EU where there are a lot more hard timings or are my games just longer than average?
On February 05 2012 01:45 Acritter wrote: Well, even more support for Tal'Darim Altar being my favorite map. Hell, I even like it PvP. Very balanced PvP map, it is. Never strays from 50%.
I don't think these statistics mean TOO much, though. They're over five months, and there were some pretty significant balance changes and metagame shifts in the past five months.
I just want to let you know that PvP winrate is always 100%.
On February 04 2012 15:07 KawaiiRice wrote: these stats don't really reflect balance and random masters players don't even play the game right anyway.
i.e.: Metal should be way more z favored, and antiga should be incredibly terran favored. Dual Sight and Shakuras should be more T favored TvZ too. Can't really comment on the other matchups.
Regardless, nice post. <3 Playhem
Define "playing the game right." Not to be a dick just want to understand your point of view a bit better. I suck so i'm not going to get into anything with you but i would assume Most masters players and above have a decent idea of how to play the game hence why they're in said league. Generally speaking.
But i'm silver so what do i know.
A lot of master league players just have a build or a playstyle that they use regardless of the map or the situation. I obviously can't speak for all master players, but beyond certain tactics on a few maps, most maps play almost the exact same. Only when you start getting into grandmaster/pro level do people really start fine tuning their builds to fit the map.
Antiga PvT is really really gay because in the usual engagement zone for most pvt battles, terran can airlift into protoss' main and be able to hit a low ground protoss army as well as the buildings in his main. This isn't true of any other maps.
On February 04 2012 15:10 ETisME wrote: There are some errors of statistics to be honest: the pylon gateway cybercore one doesn't show what the toss is going for and what match ups. it is the most consistent protoss opening that is done (but not showing what tech route and what comes next, for all we know, it isn't a nexus first or FFE) and since toss has one of the worst matchup win rate, obviously this opening, be it 4 gate, 3 gate robo expo that the opening leads to, will have the highest rate of losing.
and since 2 gate is an opening designed to win right off, it shows the actual build can win games, unlike the one above which only shows what opening he is doing and what comes next (the rest of the build that is designed to win the game) would never get a winning rate as high as the 2 gate one
You should notice the 2nd gas is included in that winrate, so the build is for double gas P on 1 base. You can be sure it ISN'T a nexus or forge FE after this... dunno how you could possibly think a protoss is going to open double gas and then forge FE LOL
I open double gas and get a Nexus with 1 Gate (MC's Gate-Nexus-Gate-Stargate-Gate) opener vs Zerg all the time. I also like doing 3 Gate Expands vs Terran, which open double gas, get 3 Gates then expand. So there are plenty of expand builds that come of double gas openers.
Sorry, it's just something I rarely see. Even if MC did it one time, it happens almost never. I just did an analysis in sc2gears of every single korean replay I have (and I don't remember deleting any, though maybe some really old ones aren't in there), and of the 72 replays that start with "Pylon, Gateway, Assimilator, Pylon, Cybernetics Core, Assimilator," only 1 of them has the next building as a nexus (which lost). Only 7 have a nexus after another pylon, and the record there is 3-4.
So okay, perhaps ~10% of the time people will expand immediately after going double gas, but it's not common, and it is a build that doesn't win very much.
On February 04 2012 15:10 ETisME wrote: There are some errors of statistics to be honest: the pylon gateway cybercore one doesn't show what the toss is going for and what match ups. it is the most consistent protoss opening that is done (but not showing what tech route and what comes next, for all we know, it isn't a nexus first or FFE) and since toss has one of the worst matchup win rate, obviously this opening, be it 4 gate, 3 gate robo expo that the opening leads to, will have the highest rate of losing.
and since 2 gate is an opening designed to win right off, it shows the actual build can win games, unlike the one above which only shows what opening he is doing and what comes next (the rest of the build that is designed to win the game) would never get a winning rate as high as the 2 gate one
You should notice the 2nd gas is included in that winrate, so the build is for double gas P on 1 base. You can be sure it ISN'T a nexus or forge FE after this... dunno how you could possibly think a protoss is going to open double gas and then forge FE LOL
so judging by that is it safe to say thats very indicative of 1 base stargate play? for the most part? If not can you explain why? I suck at the game so try not to flame if i'm way off pwease. want to keep my pillow tear free tonight :D
Depends on many, many factors. First off you have to take into account matchup: PvP - could mean anything. Double gas on 1base is very standard. For the most part you can rule out 4gate, though sometimes people will fake you with 2nd gas and still 4gate. In PvP specifically you can probably also rule out 1gate nexus for the mast part as you'd do that build with 1 gas. Could be standard robo build into immortals or colossi, could be stargate build, could be blink+obs, could be blink, could be dt, could be 3gate expand. The list goes on, because double gas is pretty much required for toss especially in that matchup.
PvT - could be 3gate voidray, could be blink+obs, could be 4gate blink, could be 3gate sentry expand, could be 1 gate robo sentry+obs expand, could be a phoenix build, could be DT, could be a wacky templar rush, could be double gas 4gate. I'm sure there are some I am missing
PvZ - well you know they're not forge FEing. The most common occurrence is probably 3gate sentry expand, but of course it could still be 1gate sentry expand, some 1-base blink build, DT expand, 1gate voidray expand, some weird robo warp prism + sentry build, and of course there are plenty of other alternatives that are less common like 1-base colossus or something silly like that.
There are so many more factors that go into what build your opponent is doing than just if they get double gas or not, but it certainly helps narrow it down. Keep tabs not only of gas but other indicative elements such as the amount of chrono saved, where the chrono is being used, and how many sentries are made early game. If your opponent gets double gas and you scout a low number of sentries, you better prepare for the possibility of DT or stargate (or blink if u see a high number of stalkers).
helped alot actually, i appreciate you taking the time to answer rather thoroughly.
I like your casts dude, some of the more informative casts outs there to be honest and you got the whole news / sportcaster head movement thing going on :D. At first i kinda thought it might be a little too much movement ( then i watched a few sports and news streams because i don't have cable ) and your pretty spot on. Actually trying to think of some negative criticism or something you could work on. I'd probably say maybe practice the play by play a bit of your casting more more, not that your weak at it but your in game and post game analysis is without a doubt your strong points so if you worked if your strong at doing both you really shouldn't have any problem finding work casting for any tournament or league. I'd actually really like to see you cast with wheat, dan, husky or either of the plot brothers. Maybe not as a permanent thing but i think it's something both the community and yourself could benefit from. Your clearly passionate about about what you do so i think you'd benefit and learn so much from it and the community would get some epic casts at the same time. Definately very comfortable and natural in front of a camera.
You took the time to answer my questions and help me out so i can at least do the same.
This is great stuff ... is the full replay pack available somewhere? I'd love to use it to look at win rates based on matchup and the length of the game...
On February 05 2012 03:57 OnYourWifi wrote: What are the winrates for each matchup for the various game lengths? SC2Gears reports this.
Now this would be some good data statistically.
Could you report on the winrates per matchup on various game length OP?
It doesn't report it directly. You can get an individual players' WR by game length, and you can get their WR by matchup. But you can't get matchup win rates for a group of players. You have to get a group of replays, paste the data into Excel, and then do a bunch of stuff. It's not rocket science, but it is work. (Work I'd be willing to do if someone has a zip file with 30K replays :D)
In general these stats tend to reinforce general forum opinion.
Shattered favors Terran in TvZ
Dual Sight sucks for Protoss
etc.
What I find interesting is that the stats for Shakuras in particular are so close. From my own experience I would have expected Terrans to have a larger advantage there in TvZ especially, ditto with Metalopolis and Zerg.
I think the variable that is not being specifically mentioned in the OP is the amount of early game all ins present across the replays.I think that might be a big contributing factor in the stats on these maps but I'm not entirely sure i have a good iea of how it sways it.
I'd really like to see an analytical post done by some of the guys who actually watched all of the replays, as much as I am excited to see this kind of quantitative data, there's a lot to speculate on without actually seeing the replays for myself.
On February 05 2012 06:24 AfricanPsycho wrote: The most favoured protoss map is 53% (pvz), come on blizz we need more P favoured maps please.
Looks like there are no PvT favoured maps, only vice versa. Its tuff to be a toss these days seriously!
Nope, cloud kingdom has TvP 45% which means PvT 55%
I wonder why cloud kingdom is the only one where protoss wins vs terran?
I bet you it is really hard for terran to drop protoss on that map, because you can simultaneously defend the third and main from drops on the left, and defend nat and main from drops on the right.
On February 05 2012 06:24 AfricanPsycho wrote: The most favoured protoss map is 53% (pvz), come on blizz we need more P favoured maps please.
Looks like there are no PvT favoured maps, only vice versa. Its tuff to be a toss these days seriously!
Nope, cloud kingdom has TvP 45% which means PvT 55%
I wonder why cloud kingdom is the only one where protoss wins vs terran?
I bet you it is really hard for terran to drop protoss on that map, because you can simultaneously defend the third and main from drops on the left, and defend nat and main from drops on the right.
I wouldn't take this too seriously but I'd definitely still recommend keeping this info in mind. I love shattered temple as a zerg because the middle is so wide open but now I find out it's heavily Terran favored. I'm still going not going to veto the map because my playstyle really fits the map and I do well on it. So again I wouldn't take this all that seriously when considering your own play and which maps to veto. Not having a P favored map isn't necessarily a bad thing, as long as the percentages are close to 50% that's all that really matters.
On February 05 2012 03:57 OnYourWifi wrote: What are the winrates for each matchup for the various game lengths? SC2Gears reports this.
Now this would be some good data statistically.
Could you report on the winrates per matchup on various game length OP?
It doesn't report it directly. You can get an individual players' WR by game length, and you can get their WR by matchup. But you can't get matchup win rates for a group of players. You have to get a group of replays, paste the data into Excel, and then do a bunch of stuff. It's not rocket science, but it is work. (Work I'd be willing to do if someone has a zip file with 30K replays :D)
Oh damn, I didn't realize it only did it on a per-player basis. That sucks.
Shattered temple stats really surprise me. zvp vs FFE you can't take a fast third, and zvt seems balanced to me. Maybe ease of 4 bases for terran? I would assume it'd be worse than shakuras for sure, late game zvt is impossible on that map.
if there's anything to gain from this it's that map balance should really become a priority for blizzard as opposed to needlessly shifting the bunker build timing back and forth 5 seconds. if we could only have stats for non-blizz maps as well. i have a feeling that they will be dramatically better than some of the ones in the OP
On February 06 2012 06:48 TheDraken wrote: if there's anything to gain from this it's that map balance should really become a priority for blizzard as opposed to needlessly shifting the bunker build timing back and forth 5 seconds. if we could only have stats for non-blizz maps as well. i have a feeling that they will be dramatically better than some of the ones in the OP
Several of the maps in the OP are non-Blizzard maps.
One thing I have to ask is whether you cut off the replays of certain maps at the point where you changed the maps. i.e. When the golds were removed from Playhem.
It'd be interesting to see how that effects TvX balance win-rates.
I guess things are more balanced than they seem. I am still shocked that Tal D' is as the most balanced. That map has so many weird issues, but I guess they apply to all races.
I think this truly shows the impact of maps on a particular match up. I am a Zerg player myself and I find myself enjoying a particular when played on specific maps (Say Protoss on Tal'Darim) yet I feel continuously helpless on Shattered Temple. Starcraft is really a game of imperfect information but I don't think you should always straight up lose a game if you didn't completely scout or correctly guess his strategy (say 6 gate vs 6 gate robo). On maps like Tal'Darim altar, given that I have strong macro, I can adjust my unit composition to counter the unit composition of my opponent where as on Shattered it's too late regardless of positions. I could say the same of TvZ on most maps, versus on Shakuras Plateau or again Shattered Temple.
In any case, I think maps are 100% the reason we have so much balance problems (evident by ladder statistics) and I wish that Blizzard would stop being retarded and use the balanced maps used in the major tournaments aka GSL. Those maps are gorgeous and well-tested.
I think this truly shows the impact of maps on a particular match up. I am a Zerg player myself and I find myself enjoying a particular when played on specific maps (Say Protoss on Tal'Darim) yet I feel continuously helpless on Shattered Temple. Starcraft is really a game of imperfect information but I don't think you should always straight up lose a game if you didn't completely scout or correctly guess his strategy (say 6 gate vs 6 gate robo). On maps like Tal'Darim altar, given that I have strong macro, I can adjust my unit composition to counter the unit composition of my opponent where as on Shattered it's too late regardless of positions. I could say the same of TvZ on most maps, versus on Shakuras Plateau or again Shattered Temple.
In any case, I think maps are 100% the reason we have so much balance problems (evident by ladder statistics) and I wish that Blizzard would stop being retarded and use the balanced maps used in the major tournaments aka GSL. Those maps are gorgeous and well-tested.
Blizzard is moving toward the right path with the map pool. The quicker seasons means they can cycle out bad maps quickly and still have time to try new things. With recent map contest, which blizzard has taken note of it, I am sure we will see better maps as time goes on.
I really wish they could just accept that some maps should be held back until players are out of the lower leagues. It would allow the higher level players to enjoy the more complex, macro maps; while allowing the newer players to cut their teeth on the smaller maps. There are so many bad players I used to encounter that would build a hidden second base on every large map. Tailoring the map pool to allow players to grow into the larger maps seems like the best way keep everyone happy and learning.
If there was SIGNIFICANT imbalances as claimed on some maps, it would've shown up here. Fairly random sampling of players that do online tournaments with no entry fee and low prize money available. The highlighted winrates are fairly confirming of some widely-held beliefs on map balance, and also serve to disprove a lot of contentions frequently put forth. I.E. That Taldarim Altar really favors a particular race.
On February 06 2012 18:40 Danglars wrote: If there was SIGNIFICANT imbalances as claimed on some maps, it would've shown up here. Fairly random sampling of players that do online tournaments with no entry fee and low prize money available. The highlighted winrates are fairly confirming of some widely-held beliefs on map balance, and also serve to disprove a lot of contentions frequently put forth. I.E. That Taldarim Altar really favors a particular race.
Well it is not uncommon for people to claim so and so map is imbalanced because of so and so spawning positions, so maybe different spawn positions averaging out is a factor? I do agree that significant imbalance at this level overall would probably have shown up in this set of replays, but there is also information we don't have which makes some of us less willing to pass judgment aside from what is quite clear anyways.
On February 07 2012 12:08 Sroobz wrote: These stats mean nothing...I think every single Z and T Korean would agree that Dual Site is the most Z favored map out there.
Protoss is down in every matchup on every map here except for Cloud Kingdom and for PvZ on Shakuras Plateau. And this is most extreme on Dualsight.
As a I result I think what would be interesting is a critical comparison of the features of Dualsight compared with those of Cloud Kingdom. Could provide some valuable insights into how and why a map can be Protoss favoured and maybe explain away some of the heavier map win rate imbalances (i.e. demonstrate that its a problem with map design rather than a racial imbalance).
On February 08 2012 02:16 Lightspeaker wrote: Protoss is down in every matchup on every map here except for Cloud Kingdom and for PvZ on Shakuras Plateau. And this is most extreme on Dualsight.
As a I result I think what would be interesting is a critical comparison of the features of Dualsight compared with those of Cloud Kingdom. Could provide some valuable insights into how and why a map can be Protoss favoured and maybe explain away some of the heavier map win rate imbalances (i.e. demonstrate that its a problem with map design rather than a racial imbalance).
Small ramps versus Open Ground = Forcefields are king or Forcefields do nothing. Some may argue otherwise but I truly believe the sentry is the most important unit in the game for any race.
Forge expand is hard on an open natural like dualsight so players that don't feel they can risk it go for either a 1 gate expand which can easily lead to your nexus being cancelled multiple times by mass speedling leaving you behind economically and on tech and army. Or you can go for a saf 3 gate expand which forces you to do a zealot sentry pressure which is such an old build that Zerg has learnt to handle it safely on maps that are GOOD for a zealot sentry pressure and Dualsight is a perfect example of a bad map for such a build. All open ground AND multiple attack paths? There's not enough sentries in the universe to have enough Forcefields to keep your sentries from dying AND cause enough economic damage to put you into a lead/equal position.
I'm not saying it's impossible to win just that most of the time you're just going to have to be better than your opponent rather than outplaying him.
Edit: all IMO obviously. Feel free to call me bad if you think I'm wrong.
Shouldn't we adjust the win rates by something like an overall expected win rate given the relative ratings of the players? If one race's player set has a generally higher rating, then that will skew win rates on certain maps compared to others. Need to find an expected winrate for each map dataset based on relative rating.
On February 08 2012 21:38 IgnE wrote: Shouldn't we adjust the win rates by something like an overall expected win rate given the relative ratings of the players? If one race's player set has a generally higher rating, then that will skew win rates on certain maps compared to others. Need to find an expected winrate for each map dataset based on relative rating.
Given the number of players that would have been involved in this massive number of replays, any variation in the average ability of players for each race is going to be tiny.
I love antiga in PvT i don't know why it's so T favored. And also i can't belive ZvP metalopolis is so close to 50%, i guess it's because of close by air spawns because cross spawns gotta be like 35% win for protoss.
It would be really freaking awesome if there was a breakdown as to # of games for each pairing. If that was available, I could run stats to display any forms of significance between race pairings. Because the sample size (# of replays) is huge, running an ANOVA and post hoc t-tests should show some very interesting results. If anyone has access to this data, please please please PM me.
edit
Because there's a large sample size, a difference of 43% and 54% can have significance because the data could be normally distributed.
If people would like to see any further in depth probability of significance statistics, please support this post
As somebody who follows both BW and SC2, I disagree with the obsession of the SC2 community with very balanced maps. A map pool with 50% win rate for all match-ups would be a very boring map pool. 40-60% win rates are good enough, provided that not all maps in the pool favor the same race in any match-up.
In order to create a map pool with all maps being very closely balanced, you'd pretty much have to design all the maps very similarly to each other. That defeats much of the purpose of having a map pool, which is having players devise a diverse set of strategies, unit compositions, timings and push paths.
On February 09 2012 08:56 andrewlt wrote: As somebody who follows both BW and SC2, I disagree with the obsession of the SC2 community with very balanced maps. A map pool with 50% win rate for all match-ups would be a very boring map pool. 40-60% win rates are good enough, provided that not all maps in the pool favor the same race in any match-up.
In order to create a map pool with all maps being very closely balanced, you'd pretty much have to design all the maps very similarly to each other. That defeats much of the purpose of having a map pool, which is having players devise a diverse set of strategies, unit compositions, timings and push paths.
While you're right to a degree we still need to keep that variance within a certain margin, you wouldn't want a map with a 40% win rate as a starting map in a Bo1 round of 256 or you'll screw up the racial balance throughout the entire tournament. At the same time it's also important that maps in the pool that are not starting maps have a slight racial imbalance since we use a loser-chooses-next format. This leads to the most interesting match ups and clearest determination of skill since it'll allow series to be played out with more games more frequently and most often determine which player was actually best and not just which one got lucky with maps.
I try to put the most balanced maps early in the tournament and as starting maps, while the ones with racial skews are moved to the map pool that losers can choose from, and maps with extreme percentages get removed entirely. Hopefully that sheds some light on the process we use when adjusting our pool.
Very interesting. I've always guessed that there are no actually Protoss favored maps... this seems to say the same thing. So for the sentiment of balancing the game with maps, maybe it can't be done. I wonder what the recent patch or HotS will do for this sort of thing.
On February 14 2012 01:28 Blacklizard wrote: Very interesting. I've always guessed that there are no actually Protoss favored maps... this seems to say the same thing. So for the sentiment of balancing the game with maps, maybe it can't be done. I wonder what the recent patch or HotS will do for this sort of thing.
You didn't bother reading the stats I guess. Cloud Kingdom is decidedly protoss favored by a small, acceptable margin.
On February 04 2012 16:34 KawaiiRice wrote: The reason I'm being sort of critical here is because playhem is trying to use these stats to represent some sort of statement about map balance when they also include thousands of games where people lose because of millions of mistakes that have nothing to do with map balance because they aren't as good.
The underlying meaning behind the analysis was more along the lines that map design can have an impact on balance, and that neither maps nor patches alone will lead to a perfect game but that changes have to be made in order to eventually find the best balance. Everyone knows how different BW was in the earliest days compared to now, and change is the only thing that will facilitate SC2's progress toward a better game.
While you're right that mistakes are factored in just as the map balance is, with a large enough sample size you should have mistakes factoring in roughly evenly for all maps/matchups unless something else is influencing some races to be more prone to game-losing mistakes. That might be an entirely different/interesting issue about how the racial mechanics work in SC2.
tl;dr, Tournaments should try out some new stuff and not just cling to old maps.
Good post, I tend to agree. Just sheer numbers makes these stats pretty good.
However, people should get off KawaiiRice because he also is stating quite a bit of truth, and thankfully isn't being so politically correct that there is nothing to be gleaned from what he is saying. You have to balance for the very top to balance the game for the rest... that is the best way to do it. Blizzard knows this and is attempting to put in balance tweaks for low level in HotS to compensate a little... the statement David Kim made about hellions starting in x mode is a nice example IMO.
On February 04 2012 16:50 Splynn wrote: It's sad that PvP is so... dumb on Tal'Darim Altar. The balance in the non-mirror matchups looks great and I really enjoy playing the map. But every time I get a PvP on it I just go straight to frown town.
Great work compiling all of this data, and thanks for all of the content in general from Playhem.
1. It is very very interesting that a map with no ramp at main is the most balanced for this level of play. What does that say about the validity of forcefields? It screams to me that distance is the only way a Protoss is safe early game against many powerful early game tactics. This has to change if map variety is ever going to be anything but huge maps in tournaments.
2. PvP 4 gate vs 4 gate... people should just learn to play early game perfectly and suddenly 4 gate vs 4 gate is a nice stable game that can go to mid and even late game. It's not perfect, but it's just as good if not better than other BO problems on other maps PvP.
Doesn't anyone remember a little game called Brood War?
Maps were NOT race balanced in that. They were never meant to be. Map adversity is what made the tournaments interesting. I want different maps that force me to change up my strategies. I voted down shakuras plateau till I learned how to do solid multi-drop openings vs Z. Now I can play the map. Is it Z favoured? Sure. Still fun.
On February 14 2012 11:55 DaemonX wrote: Doesn't anyone remember a little game called Brood War?
Maps were NOT race balanced in that. They were never meant to be. Map adversity is what made the tournaments interesting. I want different maps that force me to change up my strategies. I voted down shakuras plateau till I learned how to do solid multi-drop openings vs Z. Now I can play the map. Is it Z favoured? Sure. Still fun.
Map balance = boring.
wat? Are you seeing the SPL latelty? or, have you follow the Pro Scene BW at least this 5 last years? almost all maps were X race favorate, the only one wiith 50/50 in all races was Fighting Spirtit (gad! what a perfect map <3 <3 <3)
On February 14 2012 11:55 DaemonX wrote: Doesn't anyone remember a little game called Brood War?
Maps were NOT race balanced in that. They were never meant to be. Map adversity is what made the tournaments interesting. I want different maps that force me to change up my strategies. I voted down shakuras plateau till I learned how to do solid multi-drop openings vs Z. Now I can play the map. Is it Z favoured? Sure. Still fun.
Map balance = boring.
wat? Are you seeing the SPL latelty? or, have you follow the Pro Scene BW at least this 5 last years? almost all maps were X race favorate, the only one wiith 50/50 in all races was Fighting Spirtit (gad! what a perfect map <3 <3 <3)
Dude, I know, I am agreeing with you, that was my point!! BW didn't have race balance on maps, why does SC2 need it suddenly?
On February 14 2012 22:34 Mentalizor wrote: Isn't Shakuras and Cloud Kingdom the only maps where protoss has just a single 50%+ winrate in any matchup?
I think the winrates are slightly skewed by the fact that less korean protosses play playhem compared to T/Z by a decent bit.
On February 19 2012 08:27 DaemonX wrote: Silly post, in terms of title. A look at balance my ass.
According to this, late game TvP is broken in TERRANS FAVOUR. -> Lolwat.
Interesting stats, good contribution but the title is a silly claim.
Hah, have you seen TLPD winrate graphs? They do indeed agree to that TvP is terran favored from July 2011 until now, with only one months exception. You a terran player mby?
On February 19 2012 08:27 DaemonX wrote: Silly post, in terms of title. A look at balance my ass.
According to this, late game TvP is broken in TERRANS FAVOUR. -> Lolwat.
Interesting stats, good contribution but the title is a silly claim.
Yeah totally the stats from 31k + games from mostly masters players and progamers must be wrong and your (I must say quite eloquent and erudite, and certainly unbiased) opinion must be the right one.
On February 04 2012 15:07 KawaiiRice wrote: these stats don't really reflect balance and random masters players don't even play the game right anyway.
i.e.: Metal should be way more z favored, and antiga should be incredibly terran favored. Dual Sight and Shakuras should be more T favored TvZ too. Can't really comment on the other matchups.
Regardless, nice post. <3 Playhem
Define "playing the game right." Not to be a dick just want to understand your point of view a bit better. I suck so i'm not going to get into anything with you but i would assume Most masters players and above have a decent idea of how to play the game hence why they're in said league. Generally speaking.
But i'm silver so what do i know.
you can get into masters without having deep knowledge of what exactly you're doing and just macroing. your assumption is false because most masters don't really know and masters itself doesn't actually mean anything, it's just a stupid label blizzard decided to place on people to make them feel good.
if masters players knew how to play properly then top players wouldn't mow down ladder with ridiculous winrates.
starcraft II isn't supported by the upper .01% that play the game "right"- its supported by the bronze through masters players. These are the ones that allow there to be mlg's, gsl's, ipl's, ect... you need to balance the game for the 99.9 percent of the people to keep them interested. Just because MC can win pvz on dual sight doesnt mean i can.. what fun is that for me or anyone else in the 99.9% category.
On February 19 2012 08:27 DaemonX wrote: Silly post, in terms of title. A look at balance my ass.
According to this, late game TvP is broken in TERRANS FAVOUR. -> Lolwat.
Interesting stats, good contribution but the title is a silly claim.
Huh? I see no breakdown of late game and early game in the stats. It just shows overall Terrans win slightly more than Protoss. Am I missing something?
I don't think these reflect true balance, as they are from a period of time which included, I believe, 2 patches. Still, interesting that Tal'Darim is apparently most balanced.
So of the maps in the Season 6 pool, it looks like this:
Shakuras & Tal'Darim: balanced. Metalopolis: Bad for Protoss Shattered: Bad for Protoss, good for Terran Antiga: bad for Protoss (v Terran)
Only 1 of the 6 listed actually favored Protoss, and then only v Terran: Cloud Kingdom.
Q: what makes Cloud better for Protoss and worse for Terran, and how can the other maps be tweaked to make them more balanced? (I'm guessing it has something to do with drops and spawn positions?) Then, how can Cloud be tweaked to make it more balanced too?
EDIT: popping to a separate thread to discuss map modifications specifically.