|
On January 05 2009 19:29 Love.Zelduck wrote:He said it before I included rationalizations/numbers? At the time, ya I'm glad he said it lol.
Its not a terrible idea, but it was a bad execution, lol. Anyways, its an interesting way to think about map balance, but it does assume coaches always know best, which I think is a pretty commonly contested point (for example: *WHY WOULD THEY SEND HIM OUT FOR ACE?!*).
Also it doesn't take into account the types of players the coach has to work with. It may be KTF's best interest to send out a zerg or protoss for their ace match on medusa, but you can be damned sure they are going to send out Flash 9 times out of 10. Sometimes players practice specific maps and since sending them out twice before the ace match isn't possible, the coach may have to settle for a player of a different race if he doesn't have many good players of the 'favored' race, or is stuck because of the three race rule. And just in general, teams are going to have to make adjustments based on the team they are playing against, which affects who they send out because those particular players matchup stats are often more important than the maps stats.
Anyways, if you're going to look at map balance that way, you might just be better off just looking at how many mirror matches were played for each race on each map. It essentially tells the same story anyways (if there are zero ZvZs and 20 TvTs / PvPs, its probably a safe bet the coaches don't feel safe sending a zerg out on that map).
This has the added benefit of giving us the interesting statistic of how many games on that map are mirror matches (the lower the %, the more confident the coaches probably are sending out any race). Of course this suffers from some of the same pitfalls I listed above, but its still pretty interesting. I've taken the liberty of doing this for a couple maps:
+ Show Spoiler +Map Name
Mirrors out of Total Games
Percentage of mirrors
Destination
55 out of 176 games
31%
Medusa
50 out of 144
35%
Colosseum (I and II)
59 out of 179
33%
Andromeda
59 out of 201
29%
Blue Storm
89 out of 353
25%
Blue Storm overall has a very low percentage of mirror matches compared to many other maps, despite its zerg leaning in the overall matchup stats. (probably due to the update in the map; terran actually went 50/50 with zerg in the last 24 TvZs played on the map)
|
Numbers + Show Spoiler +
Map - Imba factor
T: %T's sent out
Z: %Z's sent out
P: %P's sent out
Destination - 3
T: 31
Z: 33
P: 36
Tau Cross - 4
T: 32
Z: 37
P: 30
Medusa - 6
T: 27
Z: 37
P: 36
815 III - 7
T: 35
Z: 38
P: 26
Sin Peaks of Baekdu - 8
T: 31
Z: 41
P: 28
Blue Storm - 9
T: 27
Z: 42
P: 31
Sin Chupung-Ryeong - 10
T: 38
Z: 39
P: 23
Andromeda - 11
T: 28
Z: 44
P: 28
Colosseum II - 15
T: 24
Z: 28
P: 48
Baekmagoji - 15
T: 34
Z: 19
P: 48
Python - 16
T: 49
Z: 25
P: 26
Arcadia 2 - 17
T: 31
Z: 50
P: 19
Katrina SE - 17
T: 16
Z: 34
P: 50
Neo Harmony - 18
T: 51
Z: 26
P: 22
Nemesis - 20
T: 23
Z: 23
P: 53
Neo Requiem - 20
T: 13
Z: 35
P: 53
Katrina - 22
T: 14
Z: 31
P: 55
Monty Hall - 23
T: 35
Z: 10
P: 55
Un'Goro Crater - 25
T: 40
Z: 52
P: 8
Chupung-Ryeong - 25
T: 53
Z: 8
P: 39
Monty Hall SE - 26
T: 59
Z: 16
P: 26
Colosseum - 30
T: 49
Z: 3
P: 49
Rush Hour 3 - 34
T: 67
Z: 28
P: 6
Harmony - 38
T: 71
Z: 19
P: 10
Othello - 39
T: 72
Z: 15
P: 13
Wuthering Heights - 42
T: 75
Z: 3
P: 21
Geometry - 43
T: 76
Z: 15
P: 9
Raid Assault 2 - 45
T: 22
Z: 78
P: 0
You need to double the number the for each race for mirror matches since both coaches decided to send out a T/Z/P.
I know you didn't do this at least on destination because the numbers are actually T:50 Z:53 P:54 (not sure about P but it is close to that)
I hope no one beat me to this
|
Coaches will send players out based on the map, yeah. But each decision is not made independently based on the map.
So you see all zerg on x map. Doesn't mean the map is imbalanced, since it could mean that zerg is completely hosed on the other maps in the match.
|
On January 05 2009 20:59 Shikyo wrote:Show nested quote +On January 05 2009 19:34 baubo wrote:On January 05 2009 18:55 PH wrote:
I hiiiighly doubt that Othello is that imbalanced. I also highly doubt that Medusa is that balanced.
Elaborate on your elaboration...let's see numbers and the stats you used.
EDIT
I read your edit of your elaboration...if that's the case, then the actual games played on the map don't figure into your analysis, and that you're basing this whole thing off of who got sent against who on what map?
-____-;; If you take into account records, then we're talking about player imbalance, not map imbalance. Put players like Best or Stork on terran favored maps and they can still rape. That doesn't make the map PvT imba. I don't see what's wrong with analyzing things this way. If a map, for example, 60% PvT, yet coaches send out terrans most of the time, it doesn't make the fact PvT imba. It just means that coaches will only dare putting the best PvTers on the map as snipers. Or maybe the team just happens to have good terrans and lacks Protoss? KTF?
On January 05 2009 23:57 errol1001 wrote:
Coaches will send players out based on the map, yeah. But each decision is not made independently based on the map.
So you see all zerg on x map. Doesn't mean the map is imbalanced, since it could mean that zerg is completely hosed on the other maps in the match.
In the proleague this season, coaches sent out 31% protoss, 34% zerg, and 35% terran. Suffice to say, it's not a matter of teams lacking a certain race. Interesting that everyone is screaming protoss imbalance, when it's actually the least used race in the proleagues.
Now I'm not saying the way the op used the stats is unbiased, but I currently find no reason why it's not legit. Everyone has their own opinions, but at least on the pro level, this method of judging map balance seems fairly good. At least better than "Oh protoss imba because PvZ is 6-2!"
|
On January 05 2009 23:14 wswordsmen wrote:
You need to double the number the for each race for mirror matches since both coaches decided to send out a T/Z/P.
I know you didn't do this at least on destination because the numbers are actually T:50 Z:53 P:54 (not sure about P but it is close to that)
I hope no one beat me to this
I did do this. The listed numbers are percentages (I wrote over the actual numbers when I was doing it originally). And I only included proleague games and excluded individual tournament games since obviously I'm trying to look at coach decisions not who qualifies for the osl.
|
In the proleague this season, coaches sent out 31% protoss, 34% zerg, and 35% terran. Suffice to say, it's not a matter of teams lacking a certain race. Interesting that everyone is screaming protoss imbalance, when it's actually the least used race in the proleagues.
Now I'm not saying the way the op used the stats is unbiased, but I currently find no reason why it's not legit. Everyone has their own opinions, but at least on the pro level, this method of judging map balance seems fairly good. At least better than "Oh protoss imba because PvZ is 6-2!"
You completely missed the point. The coaches send out various players/races on the maps because they obviously want to win. The OP takes this to make a list of balanced and unbalanced maps. The point is that these choices are not made in a vacuum, rather they are heavily influenced by OTHER factors.
Other people have pointed out problems too. This doesn't mean the stats aren't interesting. They are. They're not a direct representation of map balance though.
|
On January 06 2009 06:25 errol1001 wrote:
You completely missed the point. The coaches send out various players/races on the maps because they obviously want to win. The OP takes this to make a list of balanced and unbalanced maps. The point is that these choices are not made in a vacuum, rather they are heavily influenced by OTHER factors.
Other people have pointed out problems too. This doesn't mean the stats aren't interesting. They are. They're not a direct representation of map balance though.
Ya, I <3 interesting stats. It's why I did this. I don't expect people to agree with the methods, I just wanted to share what I did so others could see what it looks like and interpret the list for themselves.
:D
|
lmao katrina wasnt broken jesus
the methods apparently were not good at all looking at the list
one reason it doesnt work well is that teams have to use one of every race this season and that skews results and u didnt take that into account
|
Was it just me that read "Protoss: Most Balanced/Broken Maps of..."? :D
Also i wouldn't consider these stats the only thing to use when talking about balance. (Medusa wtf?).
|
On January 06 2009 06:25 errol1001 wrote:Show nested quote +
In the proleague this season, coaches sent out 31% protoss, 34% zerg, and 35% terran. Suffice to say, it's not a matter of teams lacking a certain race. Interesting that everyone is screaming protoss imbalance, when it's actually the least used race in the proleagues.
Now I'm not saying the way the op used the stats is unbiased, but I currently find no reason why it's not legit. Everyone has their own opinions, but at least on the pro level, this method of judging map balance seems fairly good. At least better than "Oh protoss imba because PvZ is 6-2!"
You completely missed the point. The coaches send out various players/races on the maps because they obviously want to win. The OP takes this to make a list of balanced and unbalanced maps. The point is that these choices are not made in a vacuum, rather they are heavily influenced by OTHER factors. Other people have pointed out problems too. This doesn't mean the stats aren't interesting. They are. They're not a direct representation of map balance though.
Are you just randomly guessing? Or can you point out what these factors would be and how this would affect maps?
|
you say alltime then you say from 2006 onward
|
What about Monty Hall, the zerg graveyard?
|
Katowice25012 Posts
I was under the impression that coaches weren't allowed to send the same race on a map twice in a row (so if you send a P on medusa, the next time you need to send T or Z), but I have no idea where I got that. Did I just make that up or is that how it works this season?
|
thedeadhaji
39489 Posts
I actually like OP's reasoning more than straight up stats. There would ideally be some sort of weighting and fixing based on the lineup of each team because not all teams are deep in each race, but it's really not a bad basis for judging which map was perceived by the proteams as having the most racial partiality.
|
On January 06 2009 08:37 baubo wrote:Show nested quote +On January 06 2009 06:25 errol1001 wrote:
In the proleague this season, coaches sent out 31% protoss, 34% zerg, and 35% terran. Suffice to say, it's not a matter of teams lacking a certain race. Interesting that everyone is screaming protoss imbalance, when it's actually the least used race in the proleagues.
Now I'm not saying the way the op used the stats is unbiased, but I currently find no reason why it's not legit. Everyone has their own opinions, but at least on the pro level, this method of judging map balance seems fairly good. At least better than "Oh protoss imba because PvZ is 6-2!"
You completely missed the point. The coaches send out various players/races on the maps because they obviously want to win. The OP takes this to make a list of balanced and unbalanced maps. The point is that these choices are not made in a vacuum, rather they are heavily influenced by OTHER factors. Other people have pointed out problems too. This doesn't mean the stats aren't interesting. They are. They're not a direct representation of map balance though. Are you just randomly guessing? Or can you point out what these factors would be and how this would affect maps?
check my post at the top of the page, the second paragraph. Some teams have a different set of players to work with. KTF has Flash as their best terran and maybe Tempest or Stats? as their best protoss; they are probably going to prefer to send out Flash even if P has a slight advantage on the map, because Flash's matchup advantage far exceeds the map's racial advantage. Thats just one example, but there are a lot of teams that have similar dilemmas (see SKT with Best/Bisu > s2/Thezerg, and Lecaf with Jaedong > Backho). This most affects maps I think when it comes to ace match: if the other team sends out their no-name protoss because of the maps balance, KTF will laugh and send out Flash who will wreck him. For this reason, both teams usually end up sending out their best players regardless of race, or players who have prepared a specific build rather than blindly picking their best player from a specific race.
|
On January 06 2009 09:09 thedeadhaji wrote:
I actually like OP's reasoning more than straight up stats. There would ideally be some sort of weighting and fixing based on the lineup of each team because not all teams are deep in each race, but it's really not a bad basis for judging which map was perceived by the proteams as having the most racial partiality.
Yeah it's useful reasoning, but it has flaws, just like anything else:
1. Seasonal map synthesis.
If the map pool collectively favors any one race at any point, it will likely not reflect the entirety of the skew. This is in part due to team lineups (ie if a team has a strong zerg, and there are no strong zerg maps, they will still likely send him out frequently because they do not have a better option - so even if a map disadvantages him by 10%, he may be 10% better than any other option), and because of the recent season's lineup requirements, which both limit the # of repeated races per match, and requires a minimum number per match (this definitely reduces racial variance). However, if the map pool was essentially balanced, then neither of these factors would matter, and the racial spread would be a true distribution of racial utility at the proleague level.
I'm positive "sniping" would not be a skew, because the % of snipes would probably be equal to the true racial balance, in a large enough sample.
2. Imperfect map strategy.
Map balance changes over the course of a season, either through new strategies and paradigms in the metagame, or just in the map.
3: Small sample size
One of the most important things to remember in statistics. Most people say raw statistics are useless, but it's just an incorrect statement. It's more accurate to say, "statistics only contain so much useful information," and sample size largely determines how useful the information is. I don't think there's ever been a sabermetric Starcraft study, but I'd assume you'd need at least 20 games to get a usable sample.
|
On January 06 2009 06:51 ish0wstopper wrote:
lmao katrina wasnt broken jesus
the methods apparently were not good at all looking at the list
one reason it doesnt work well is that teams have to use one of every race this season and that skews results and u didnt take that into account
Remember Flash fucked that up with his wins against Stork haha.
Oh yeah what about Troy?
|
On January 06 2009 06:51 ish0wstopper wrote:
lmao katrina wasnt broken jesus
the methods apparently were not good at all looking at the list
one reason it doesnt work well is that teams have to use one of every race this season and that skews results and u didnt take that into account
katrina not broken? im assuming broken means not imbalanced in which case i have to disagree.. by alot.....
|
On January 05 2009 19:02 Shikyo wrote:
Medusa
Race Stats (non-mirrors):
TvZ: 12-16 (42.9%) [ Games ]
ZvP: 15-20 (42.9%) [ Games ]
PvT: 21-10 (67.7%) [ Games ]
Mirrors: 15 TvT | 11 ZvZ | 24 PvP
Sounds balanced.
Also, one thing you didn't take into account is the fact that in 08-09 Proleague, you can't just send out the best race on their best map because every team must send out at least one player of each race. But whatevers.
That seems really unbalanced becuase it's clearly P favored balanced is all number about the same number so it's a trade off of T>Z>P>T
|
Did you have some formulaic way of determining the "imba factor" or did you just make it up?
|
|
|
|
|
|
EPT
