EPT
Maps in the Balance - Page 2
| Forum Index > Final Edits |
|
Guybrush
Spain4744 Posts
| ||
|
Xeln4g4
Italy1209 Posts
TvP: 8-16 favour P vs T TvP: 5-14 favour P vs T PvZ: 22-11 favour P vs Z TvZ: 11-3 favour T vs Z TvP: 4-0 favour T vs P TvZ: 14-29 favour Z vs T PvZ: 14-24 favour Z vs P PvZ: 3-7 favour Z vs P This tell me that P is quite favoured on T while Z is quite favoured on P. Nothing else. Considering that the analysis was carried on 4 different maps, so different playground, i COULD state that: Zerg have a balance mu (ZvT) and an easy mu (ZvP) Protoss have an easy mu (PvT) and an imbalanced mu (PvZ) Terran have a balaced mu (TvZ) and an imbalanced mu (TvP) Map issue or race balance issue? Beside this, great article as usual. Very good JOB. | ||
|
Mindcrime
United States6899 Posts
On January 22 2008 01:18 IH4t3z3rg wrote: With the premise that imho the number of games played on this maps it's absolutely not enough to state what mu is imba or not (u need more played matches for this kind of analysis ...) it's funny to see how the same number can lead to different opinions: TvP: 8-16 favour P vs T TvP: 5-14 favour P vs T PvZ: 22-11 favour P vs Z TvZ: 11-3 favour T vs Z TvP: 4-0 favour T vs P TvZ: 14-29 favour Z vs T PvZ: 14-24 favour Z vs P PvZ: 3-7 favour Z vs P This tell me that P is quite favoured on T while Z is quite favoured on P. Nothing else. Considering that the analysis was carried on 4 different maps, so different playground, i COULD state that: Zerg have a balance mu (ZvT) and an easy mu (ZvP) Protoss have an easy mu (PvT) and an imbalanced mu (PvZ) Terran have a balaced mu (TvZ) and an imbalanced mu (TvP) Map issue or race balance issue? Beside this, great article as usual. Very good JOB. It's a map issue and that should be obvious. | ||
|
3 Lions
United States3705 Posts
| ||
|
axel
France385 Posts
Thanks man, i'm still very surprised that people spend time to make so good , intresting and detailed articles about broodwar. Very nice. | ||
|
Lemonwalrus
United States5465 Posts
| ||
|
GeneralStan
United States4789 Posts
| ||
micronesia
United States24776 Posts
On January 21 2008 22:00 dybydx wrote: I like this idea and I think it can be applied to other areas as well. Although, you might want to propose it in a more suitable place.i propose a standardized method to measure which map is more imba than the other. let x = min(TvP, PvT) y = min(TvZ , ZvT) z = min(PvZ, ZvP) q = min(x , y , z) balance meter = x * y * z * 8 the lowest score is 0, and the highest score is 1. lower score = imba we can also calculate the "hotness of a specific race" hot meter for terran = [sum over all map of] chance of t win in a map^2 / number of maps x 2 lowest score is 0, highest score is 1. 1 being red hot. hence, according to my imba meter, (using latest statistics right now on TLPD map info) Baekmagoji = 0.189841968 Blue Storm = 0.48798288 Most balanced Katrina = 0.328012272 Un'Goro Crater = 0 Most imba This means Un'Goro Crater (it the TvP imba continues) has to be removed. It would make sense too i mean if the toss player has no chance of winning this map, it would be very unfair to play a T player in the finals. (assuming other maps are balanced) | ||
|
Tigron
United States13 Posts
I've often thought about the map maker's catch-22. The only real way to tell if a map is balanced or not is to put it through its paces in a tournament. Any kind of prior "beta testing" that could be done would be ineffective, as the best people to help with that are the players, who have a conflict of interest. But if the map turns out to be unbalanced, it defeats the purpose of having a tournament. Are we testing the maps here, or the players? | ||
Daigomi
South Africa4316 Posts
| ||
|
JensOfSweden
Cameroon1767 Posts
Although I gotta say that people read too much into map imbalance sometimes. If for example PvT is 10-5 on a map this is too small of a sample to mean anything at all. When people talk about maps, the players skill is unfortunately often ignored. Anyhow, fucking impressive write-up | ||
|
pandas
31 Posts
On January 21 2008 22:00 dybydx wrote: i propose a standardized method to measure which map is more imba than the other. let x = min(TvP, PvT) y = min(TvZ , ZvT) z = min(PvZ, ZvP) q = min(x , y , z) balance meter = x * y * z * 8 the lowest score is 0, and the highest score is 1. lower score = imba we can also calculate the "hotness of a specific race" hot meter for terran = [sum over all map of] chance of t win in a map^2 / number of maps x 2 lowest score is 0, highest score is 1. 1 being red hot. hence, according to my imba meter, (using latest statistics right now on TLPD map info) Baekmagoji = 0.189841968 Blue Storm = 0.48798288 Most balanced Katrina = 0.328012272 Un'Goro Crater = 0 Most imba This means Un'Goro Crater (it the TvP imba continues) has to be removed. It would make sense too i mean if the toss player has no chance of winning this map, it would be very unfair to play a T player in the finals. (assuming other maps are balanced) lol? sample size? | ||
Xeofreestyler
Belgium6776 Posts
thanks! | ||
|
Cascade
Australia5405 Posts
On January 21 2008 22:00 dybydx wrote: i propose a standardized method to measure which map is more imba than the other. let x = min(TvP, PvT) y = min(TvZ , ZvT) z = min(PvZ, ZvP) q = min(x , y , z) balance meter = x * y * z * 8 the lowest score is 0, and the highest score is 1. lower score = imba we can also calculate the "hotness of a specific race" hot meter for terran = [sum over all map of] chance of t win in a map^2 / number of maps x 2 lowest score is 0, highest score is 1. 1 being red hot. hence, according to my imba meter, (using latest statistics right now on TLPD map info) Baekmagoji = 0.189841968 Blue Storm = 0.48798288 Most balanced Katrina = 0.328012272 Un'Goro Crater = 0 Most imba This means Un'Goro Crater (it the TvP imba continues) has to be removed. It would make sense too i mean if the toss player has no chance of winning this map, it would be very unfair to play a T player in the finals. (assuming other maps are balanced) Nice, I would suggest a different measure though.  I think we both agree that a x = y = z = 0.1 map is more imbalanced than a x = 0, y = z = 0.5 map. Your measure can't tell the difference of a 0,0,0 map and a 0,50,50 map, which I think is a flaw.maybe ( x + y + z)? Or if you want to discourage heavy imbalances more you could take sqrt( (.5-x)^2 + (.5-y)^2 + (.5-z)^2 ) or even higher powers for more weight to worst matchup on the map. Also normalise them if you prefer from 0 to 1. Well, that was just a few ideas... Good article btw! I'll watch a few of the vods and think about what you wrote. | ||
|
dybydx
Canada1764 Posts
lol? sample size? obviously when we take statistics we are assuming the sample is representative of the population. it may not be in this case but it is the most accurate information we have. this is why i stated "if the TvP imba continues" it would help the analysis alot if we can get a hold of qualification round information as well, but hey, i tried. | ||
|
dybydx
Canada1764 Posts
Nice, I would suggest a different measure though. I think we both agree that a x = y = z = 0.1 map is more imbalanced than a x = 0, y = z = 0.5 map. Your measure can't tell the difference of a 0,0,0 map and a 0,50,50 map, which I think is a flaw. it was intended. i realized this problem when i formulated the equation. but then i realized 3 things 1. starleague do not offer TvZ only or ZvP only maps. all match ups are possible on any given map. balance must therefore imply across all 3 races, not 2 races. 2. in addition, suppose we are at superfight 5, T,Z,P race wars. on a 0,0,0 map, it is a deadlock, on a 0,5,5 map means 1 race immediately get taken out without contest. thus 0,0,0 is actually less imba than 0,5,5 3. finally, i feel that any map that has 0 in it should be taken out because there is no real contest. | ||
|
pyrogenetix
China5098 Posts
i just showed this to a friend who said that sc was no skill and it totally changed his mind. he said he had no idea the distances, layout, chokes etc played such a big role. | ||
|
Fermata
5 Posts
I liked that map T.T it looked interesting and seemed like it could bring about some new and exciting plays... | ||
|
vhallee
899 Posts
| ||
|
ineverwin
United States93 Posts
Zergs often expand to the expo behind their opponent's nat, anyways. edit: Thanks for writing this up, but I do think that the sample size is wayyyyyy to small for us to gain any knowledge from it, and I don't think we can be accurate by trying to theorycraft imblances from so few games. Look at temple: it's balanced, but the first 30 or so games would have never told us that (and you are using sets sometimes much smaller than that). | ||
| ||
