|
I decided to take all the data from TLPD of maps played since the start of 2005 in pro BW (177 maps), and create an algorithm that would rank them based on number of games played, number of tournaments used in, when they were in their prime, their winrates in all-match ups, the amount of mirror match-ups played (essentially all data that can be extracted).
My intention here is to compile a list of maps that tournament organizers should consider, what maps are worth practicing on, etc.
The result is here:
![[image loading]](http://i.imgur.com/3XzE80K.png)
All-time FiWiFaKi SC:BW Map Balance Power Ranking
1 Blue Storm 20.670
2 Tau Cross 18.142
3 Fighting Spirit 14.826
4 Circuit Breaker 11.952
5 Fantasy II 9.920
6 Destination 9.897
7 Neo Medusa 9.396
8 Colosseum II 8.781
9 Sin Peaks of Baekdu 7.545
10 Match Point 7.285
11 Luna 6.992
12 Katrina 6.942
13 La Mancha 6.623
14 Eye of the Storm 6.513
15 Heartbreak Ridge 6.474
16 Python 6.442
17 Arcadia 6.258
18 Neo Requiem 6.015
19 Nostalgia 6.008
20 Longinus 2 5.980
21 Troy 5.874
22 Athena 5.839
23 Zodiac 5.765
24 Blitz 5.759
25 Hitchhiker 5.707
26 Neo Bifrost 5.090
27 New Bloody Ridge 5.023
28 Monty Hall 4.882
29 Return of the King 4.867
30 Byzantium 2 4.688
31 Sin Chupung-Ryeong 4.670
32 Into the Darkness 2 4.640
33 Polaris Rhapsody 4.443
34 Rush Hour 3 4.409
35 Odd-Eye 2 4.292
36 Triathlon 4.262
37 R-Point 4.245
38 Neo Forte 4.208
39 Grand Line SE 4.059
40 Plasma 4.048
41 Benzene 3.889
42 815 III 3.841
43 Great Barrier Reef 3.798
44 Outsider 3.713
45 Monte Cristo 3.694
46 Neo Guillotine 3.681
47 Fortress 3.640
48 Ride of Valkyries 3.534
49 Ground Zero 3.427
50 New Sniper Ridge 3.329
51 Gladiator 3.303
52 Roadrunner 3.281
53 Sin Pioneer Period 3.144
54 Aztec 3.094
55 Andromeda 3.085
56 Reverse Temple 3.083
57 Detonation F 2.984
58 God's Garden 2.932
59 Neo Electric Circuit 2.927
60 Neo Arkanoid 2.775
61 Desperado 2.725
62 Empire of the Sun 2.680
63 Lost Temple 2.498
64 Hunters 2.201
65 Arizona 2.196
66 Baekmagoji 2.184
67 Sin Gaema Gowon 2.141
68 Wuthering Heights 2.109
69 Icarus 1.977
70 Holy World 1.942
71 Judgment Day 1.923
72 Xeno Sky 1.893
73 The Eye 1.824
74 Neo Harmony 1.815
75 Jade 1.814
76 Othello 1.781
77 Raid Assault 1.761
78 Neo Forbidden Zone 1.744
79 Martian Cross 1.744
80 Parallel Lines 3 1.732
Let me know what guys think of the outcome, whether you agree or disagree I will post the algorithm I used, how I went about deriving it, an explanation, and so on - tomorrow.
Maps that were not played since 2005 were not considered for the reason that the game played a lot differently back then, and hence it's too far of a stretch to imply their current balance from such ancient performance (also the year multiplier decreases quite significantly once you go back far enough, so none of them would be near the top).
A score of zero is only achievable if one of the 6 MU's has not been played a single time, as you get further down the list, there's a bit more uncertainty in the numbers, as one game can be the difference between a 67% win rate and a 50% winrate, so if you see something very unpopular somewhere along the line, keep that in mind.
Lastly (for now), what I want to mention is that if there were two different versions of a map that were reasonably balanced (Medusa is a good example), I used an algorithm to combine their rankings into the more popular one (adequately compensated with diminishing returns), and the less popular version was omitted - the aim was to only have one version of each map in the ranking.
Interest fact #1: Interestingly, Acro is the only map that has a 50-50 winrate in all non-mirror match-ups, but due to not having all mirrors occur on the map, it receives a score of zero.
Interest fact #2: Bluestorm receives the award as the most balanced map, for having 50.4% 49.5% and 46.9% win rates in the 3 match-ups, having a very balanced mirror match-up distribution, hosting over 900 games, and being featured in 27 korean tournaments, including many recent ones (relative speaking)
Interesting fact #3: Lost temple has been featured in over 100 tournaments (102), by far and away more than any other maps, even still, it was only able to achieve rank 63, just one higher over everyone's favorite Racewars and FFA maps, Hunters.
Interesting fact #4: Hunters is the only competitive non-two, three, or four player map used since 2005, you'd have to go back to Christmas eve of 2004 to find the last used 5 player map,Evolution Predators 2 (as well as it's two predecessors beforehand), and back to early 2002 for either of Deep Purple or Winter Conquest, as the only two 6 player competitive BW maps (which also happened to be the first map that Grrrr... won in the first OSL finals). There has thus far been no professional BW 7 player map.
Hope someone found this interesting, helpful, fun, or whatever 
|
From what I've read and heard from progamers, Blue Storm is known as a heavily T favored map in both matchups. Python seems to be lower than what I expected and Nostalgia is definitely too high just because how imbalanced it is TvZ. Not having gas at the natural basically made it impossible for zergs to win. Rush hour also seems really low.
|
Oh hell yeah. My fav map hits #9 and my second favorite map hits #1. What is up fellas?
|
United States10091 Posts
Blue Storm was definitely imbalanced, Tau Cross also. FS has been showing it's balance problems.
Imo, circuit breakers is probs the most balanced map right now.
|
On October 07 2016 16:46 FlaShFTW wrote:
Blue Storm was definitely imbalanced, Tau Cross also. FS has been showing it's balance problems.
Imo, circuit breakers is probs the most balanced map right now.
I'm really curious as to what the imbalances are. Apparently Blue Storm is Terran favored. What about Tau Cross and Fighting Spirit? Also I was under the impression Circuit Breakers was quite Zerg favored.
|
Interesting to hear this from you guys, as it's what the numbers say, and not my opinion.
You're right that there's some things that are almost impossible to consider, such as:
1) If one player of one race is way better than anyone else, it can skew the results, for example Flash might make the map look fairly balanced even though it's zerg favored.
2) An imbalance might have been newly found, and it was pulled out of the map pool before it could properly reflect in the results
3) This doesn't take into account the "fun level" of any of the maps.
4) And if you want to go there, match fixing either.
5) Oh, and the last big assumption was that an equal number games were played by all races, it'd be too difficult to figure out exactly how many games were played in what time frame to normalize the data like that, especially since I collected my information by going through map profiles, and not having a comprehensive match list. Hence it doesn't take into account the era of protoss, the era of zerg, etc.
When I performed this I was quite happy with the results, as a lot of what I expected lined-up where in reasonable places (I'm not one to intentionally fudge data for something like this). Honestly, I personally would agree that FS, Circuit Breaker, Bluestorm, and Tau Cross are the most balanced maps in pro BW, in no particular order.
|
The problem with balance is that the meta-game shifted and still shifts a lot in BW. New maps help to adjust to imbalances due to metagame shifts. On the other hand, some maps that were quite balanced for a while are now imbalanced because the meta changed. Fighting Spirit is a good example. In it's heydays it was probably the most balanced map ever with less than 3 % deviation from 50 % winrate in all matchups. However, in this era of BW it shows some problems, although some of the problem might be caused by the lack of enough strong zerg players.
Blue Storm was always broken, even though an interesting map. The problem was mostly the late game because of it's nature, map splitting was very easy and the latest expos were hard to assault. T>P>Z for late game. For midgame I think it was quite balanced.
|
in my opinion as a protoss player, i think the most balanced maps were Destination and Heartbreak rdige.
Fighting spirit is really hard vs turtly zergs.
Blue storm is somewhat balanced in PvZ, but kinda hard to play mid-late game PvT.
Tau Cross is really really good for protoss.
and i feel like Circuit Breaker is not that good for PvZ, even tho i haven't played it that much as the others.
+ Show Spoiler +
|
On October 07 2016 17:32 XenOsky- wrote:in my opinion as a protoss player, i think the most balanced maps were Destination and Heartbreak rdige. + Show Spoiler +
I thought protoss players hated Destination in both match ups. I personally liked both of those maps but Destination certainly seemed to be a bit imbalanced, at least in TvP (terran favoured).
|
On October 07 2016 17:26 Miragee wrote:
The problem with balance is that the meta-game shifted and still shifts a lot in BW. New maps help to adjust to imbalances due to metagame shifts. On the other hand, some maps that were quite balanced for a while are now imbalanced because the meta changed. Fighting Spirit is a good example. In it's heydays it was probably the most balanced map ever with less than 3 % deviation from 50 % winrate in all matchups. However, in this era of BW it shows some problems, although some of the problem might be caused by the lack of enough strong zerg players.
And in reverse, reimplementing some maps after a series of meta shifts might lead to different and possibly better balanced resluts.
Also I really wouldn't lump anything pre-2008 with modern BW. The game as it was played before doesn't even compare. If we want to be more precise 2010-2012 was the pinnacle of the game, as post-KeSPA BW shows the lack of organization and practice infrastructure.
On October 07 2016 17:38 Miragee wrote:Show nested quote +On October 07 2016 17:32 XenOsky- wrote:in my opinion as a protoss player, i think the most balanced maps were Destination and Heartbreak rdige. + Show Spoiler + I thought protoss players hated Destination in both match ups. I personally liked both of those maps but Destination certainly seemed to be a bit imbalanced, at least in TvP (terran favoured).
Destination was fine for the races that mattered.
|
On October 07 2016 17:38 Miragee wrote:Show nested quote +On October 07 2016 17:32 XenOsky- wrote:in my opinion as a protoss player, i think the most balanced maps were Destination and Heartbreak rdige. + Show Spoiler + I thought protoss players hated Destination in both match ups. I personally liked both of those maps but Destination certainly seemed to be a bit imbalanced, at least in TvP (terran favoured).
Desti was not that good vs terran if terran player exploited the bridges, but it was really good for recalls and reaver play, also protoss has a lot of space to reposition his army vs mech, which can buy a lot of time to find opennings or weak spots for counter attacks or good flanks.
|
On October 07 2016 16:57 SCC-Faust wrote:Show nested quote +On October 07 2016 16:46 FlaShFTW wrote:
Blue Storm was definitely imbalanced, Tau Cross also. FS has been showing it's balance problems.
Imo, circuit breakers is probs the most balanced map right now. I'm really curious as to what the imbalances are. Apparently Blue Storm is Terran favored. What about Tau Cross and Fighting Spirit? Also I was under the impression Circuit Breakers was quite Zerg favored.
Tau Cross was generally considered well balanced considering that it's a 3 person map purely because of how open it was. No tight choke points to make it terran favored in TvP and relatively easy for zergs to get 3 gas. This led to 200 vs 200 macro wars which weren't terribly exciting. FS I consider a balanced map but very vanilla.
|
On October 07 2016 17:42 r33k wrote:Show nested quote +On October 07 2016 17:26 Miragee wrote:
The problem with balance is that the meta-game shifted and still shifts a lot in BW. New maps help to adjust to imbalances due to metagame shifts. On the other hand, some maps that were quite balanced for a while are now imbalanced because the meta changed. Fighting Spirit is a good example. In it's heydays it was probably the most balanced map ever with less than 3 % deviation from 50 % winrate in all matchups. However, in this era of BW it shows some problems, although some of the problem might be caused by the lack of enough strong zerg players.
And in reverse, reimplementing some maps after a series of meta shifts might lead to different and possibly better balanced resluts. Also I really wouldn't lump anything pre-2008 with modern BW. The game as it was played before doesn't even compare. If we want to be more precise 2010-2012 was the pinnacle of the game, as post-KeSPA BW shows the lack of organization and practice infrastructure. Show nested quote +On October 07 2016 17:38 Miragee wrote:On October 07 2016 17:32 XenOsky- wrote:in my opinion as a protoss player, i think the most balanced maps were Destination and Heartbreak rdige. + Show Spoiler + I thought protoss players hated Destination in both match ups. I personally liked both of those maps but Destination certainly seemed to be a bit imbalanced, at least in TvP (terran favoured). Destination was fine for the races that mattered.
And my model takes that into account, and treats 2010-2012 as the most important time. Otherwise maps like LT would be way higher up.
|
On October 07 2016 17:56 FiWiFaKi wrote:Show nested quote +On October 07 2016 17:42 r33k wrote:On October 07 2016 17:26 Miragee wrote:
The problem with balance is that the meta-game shifted and still shifts a lot in BW. New maps help to adjust to imbalances due to metagame shifts. On the other hand, some maps that were quite balanced for a while are now imbalanced because the meta changed. Fighting Spirit is a good example. In it's heydays it was probably the most balanced map ever with less than 3 % deviation from 50 % winrate in all matchups. However, in this era of BW it shows some problems, although some of the problem might be caused by the lack of enough strong zerg players.
And in reverse, reimplementing some maps after a series of meta shifts might lead to different and possibly better balanced resluts. Also I really wouldn't lump anything pre-2008 with modern BW. The game as it was played before doesn't even compare. If we want to be more precise 2010-2012 was the pinnacle of the game, as post-KeSPA BW shows the lack of organization and practice infrastructure. On October 07 2016 17:38 Miragee wrote:On October 07 2016 17:32 XenOsky- wrote:in my opinion as a protoss player, i think the most balanced maps were Destination and Heartbreak rdige. + Show Spoiler + I thought protoss players hated Destination in both match ups. I personally liked both of those maps but Destination certainly seemed to be a bit imbalanced, at least in TvP (terran favoured). Destination was fine for the races that mattered. And my model takes that into account, and treats 2010-2012 as the most important time. Otherwise maps like LT would be way higher up.
The meta shifted compared to 2012
|
On October 07 2016 18:07 duke91 wrote:Show nested quote +On October 07 2016 17:56 FiWiFaKi wrote:On October 07 2016 17:42 r33k wrote:On October 07 2016 17:26 Miragee wrote:
The problem with balance is that the meta-game shifted and still shifts a lot in BW. New maps help to adjust to imbalances due to metagame shifts. On the other hand, some maps that were quite balanced for a while are now imbalanced because the meta changed. Fighting Spirit is a good example. In it's heydays it was probably the most balanced map ever with less than 3 % deviation from 50 % winrate in all matchups. However, in this era of BW it shows some problems, although some of the problem might be caused by the lack of enough strong zerg players.
And in reverse, reimplementing some maps after a series of meta shifts might lead to different and possibly better balanced resluts. Also I really wouldn't lump anything pre-2008 with modern BW. The game as it was played before doesn't even compare. If we want to be more precise 2010-2012 was the pinnacle of the game, as post-KeSPA BW shows the lack of organization and practice infrastructure. On October 07 2016 17:38 Miragee wrote:On October 07 2016 17:32 XenOsky- wrote:in my opinion as a protoss player, i think the most balanced maps were Destination and Heartbreak rdige. + Show Spoiler + I thought protoss players hated Destination in both match ups. I personally liked both of those maps but Destination certainly seemed to be a bit imbalanced, at least in TvP (terran favoured). Destination was fine for the races that mattered. And my model takes that into account, and treats 2010-2012 as the most important time. Otherwise maps like LT would be way higher up. The meta shifted compared to 2012
But 2010-2012 was the pinnacle of BW, the time it was played at the highest level.
Of course it's difficult to treat the data (and some things that data cannot quantify), and I made the most reasonable assumptions I could to arrive at answers.
I'm not here to argue whether x map is #6 or #7, but mostly as a general guideline of how the maps stack up. Particularly at the very very top and the very very bottom, there can be some issues, since as you get down a lot, your uncertainty goes up due to small sample sizes, and as you reach the top, the difference between a 51% winrate and a 52% winrate becomes very significant (the perfectly balanced map results wise, as well as even mirror match up distribution would results in a map score of infinity).
edit: I could in theory compile the list for amateur broodwar as well, it's just there fewer maps (60?~), with far smaller sample sizes (minus CB and FS). Straight up, going through the game list, it seems like 2/3rds of the games are playing on FS or Circuit Breaker. It's the difference between 35k games, and 6.6k games.
edit2: Also, my bad FS and CB "only" take up 40% of games combined, though from scanning through the list, it appears that percentage has been far higher in 2015 and 2016.
|
On October 07 2016 18:12 FiWiFaKi wrote:Show nested quote +On October 07 2016 18:07 duke91 wrote:On October 07 2016 17:56 FiWiFaKi wrote:On October 07 2016 17:42 r33k wrote:On October 07 2016 17:26 Miragee wrote:
The problem with balance is that the meta-game shifted and still shifts a lot in BW. New maps help to adjust to imbalances due to metagame shifts. On the other hand, some maps that were quite balanced for a while are now imbalanced because the meta changed. Fighting Spirit is a good example. In it's heydays it was probably the most balanced map ever with less than 3 % deviation from 50 % winrate in all matchups. However, in this era of BW it shows some problems, although some of the problem might be caused by the lack of enough strong zerg players.
And in reverse, reimplementing some maps after a series of meta shifts might lead to different and possibly better balanced resluts. Also I really wouldn't lump anything pre-2008 with modern BW. The game as it was played before doesn't even compare. If we want to be more precise 2010-2012 was the pinnacle of the game, as post-KeSPA BW shows the lack of organization and practice infrastructure. On October 07 2016 17:38 Miragee wrote:On October 07 2016 17:32 XenOsky- wrote:in my opinion as a protoss player, i think the most balanced maps were Destination and Heartbreak rdige. + Show Spoiler + I thought protoss players hated Destination in both match ups. I personally liked both of those maps but Destination certainly seemed to be a bit imbalanced, at least in TvP (terran favoured). Destination was fine for the races that mattered. And my model takes that into account, and treats 2010-2012 as the most important time. Otherwise maps like LT would be way higher up. The meta shifted compared to 2012 But 2010-2012 was the pinnacle of BW, the time it was played at the highest level. Of course it's difficult to treat the data (and some things that data cannot quantify), and I made the most reasonable assumptions I could to arrive at answers. I'm not here to argue whether x map is #6 or #7, but mostly as a general guideline of how the maps stack up. Particularly at the very very top and the very very bottom, there can be some issues, since as you get down a lot, your uncertainty goes up due to small sample sizes, and as you reach the top, the difference between a 51% winrate and a 52% winrate becomes very significant (the perfectly balanced map results wise, as well as even mirror match up distribution would results in a map score of infinity).
The pinnacle probably started before 2010. Irrc I would at least include 2009 as well.
|
On October 07 2016 18:27 Miragee wrote:Show nested quote +On October 07 2016 18:12 FiWiFaKi wrote:On October 07 2016 18:07 duke91 wrote:On October 07 2016 17:56 FiWiFaKi wrote:On October 07 2016 17:42 r33k wrote:On October 07 2016 17:26 Miragee wrote:
The problem with balance is that the meta-game shifted and still shifts a lot in BW. New maps help to adjust to imbalances due to metagame shifts. On the other hand, some maps that were quite balanced for a while are now imbalanced because the meta changed. Fighting Spirit is a good example. In it's heydays it was probably the most balanced map ever with less than 3 % deviation from 50 % winrate in all matchups. However, in this era of BW it shows some problems, although some of the problem might be caused by the lack of enough strong zerg players.
And in reverse, reimplementing some maps after a series of meta shifts might lead to different and possibly better balanced resluts. Also I really wouldn't lump anything pre-2008 with modern BW. The game as it was played before doesn't even compare. If we want to be more precise 2010-2012 was the pinnacle of the game, as post-KeSPA BW shows the lack of organization and practice infrastructure. On October 07 2016 17:38 Miragee wrote:On October 07 2016 17:32 XenOsky- wrote:in my opinion as a protoss player, i think the most balanced maps were Destination and Heartbreak rdige. + Show Spoiler + I thought protoss players hated Destination in both match ups. I personally liked both of those maps but Destination certainly seemed to be a bit imbalanced, at least in TvP (terran favoured). Destination was fine for the races that mattered. And my model takes that into account, and treats 2010-2012 as the most important time. Otherwise maps like LT would be way higher up. The meta shifted compared to 2012 But 2010-2012 was the pinnacle of BW, the time it was played at the highest level. Of course it's difficult to treat the data (and some things that data cannot quantify), and I made the most reasonable assumptions I could to arrive at answers. I'm not here to argue whether x map is #6 or #7, but mostly as a general guideline of how the maps stack up. Particularly at the very very top and the very very bottom, there can be some issues, since as you get down a lot, your uncertainty goes up due to small sample sizes, and as you reach the top, the difference between a 51% winrate and a 52% winrate becomes very significant (the perfectly balanced map results wise, as well as even mirror match up distribution would results in a map score of infinity). The pinnacle probably started before 2010. Irrc I would at least include 2009 as well.
This will make a lot of controversy, but the coefficients I used were:
2012: 1
2011: 1
2010: 1
2009: 0.95
2008: 0.9
2007: 0.85
2006: 0.8
2005: 0.7
2004: 0.55
2003: 0.4
2002: 0.2
2001: 0.1
2000: 0.05
After this number multiplied the previous terms, it was also square rooted, so the difference isn't as large as it appears. I took the year for the map when the map was in its prime. Anyway, I think it's reasonable, and no, I will not change it S:
|
On October 07 2016 16:04 FiWiFaKi wrote:
44 Outsider 3.713
48 Ride of Valkyries 3.534
67 Sin Gaema Gowon 2.141
69 Icarus 1.977
78 Neo Forbidden Zone 1.744
All of my favorite Maps hardly suck in this ranking  (
Beside that, I find it hard to believe that" Enter the Dragon" and "Ragnarok" werent played since 2005?! I loved those Maps as well, allthough Ragnarok wasnt balanced at all...!
|
Ragnarok had a rating of zero, 2 mirror MU never played, and its win rates were god awful
Enter the Dragon 2004 1.402 .... In 110th place, once combining any duplicate maps that came before it, it's probably around 90th~. At that point all the maps were so shitty or so hardly played, I didn't think adding them in would add anything meaningful.
Just look at maps like Battle Royal... 0 TvT 71 ZvZ 1PvP, these maps don't even deserve a mention.
|
75 Jade 1.814
poses a lot of questions to the ranking. A very well balanced map is far at the end of the ranking.
|
On October 07 2016 18:57 kogeT wrote:
75 Jade 1.814
poses a lot of questions to the ranking. A very well balanced map is far at the end of the ranking.
Yeah, when I was entering in the numbers, I was fairly surprised too. But it was a relatively new map for pro BW standards, without a huge sample size, and I believe only 1 ZvZ played between the two versions of it.
Remember, it's a purely pro BW ranking, and doesn't include amateur BW at all. Next step is to maybe try and incorporate both (as well as international), but its not easy to decide the best way to do it.
I'm just performing a treatment of the numbers given to me, and if you look on TLPD, it definitely doesn't look like a good map. I can't just introduce arbitrary this map is good coefficients.
|
How does this take into effect elite players? (Fantasy, Flash, Jaedong, Stork, Jangbi)
Example: Fantasy/Flash have an 80% win rate against zerg on map X - but the TvZ matchup on Map X is 50/50 - which includes the best players 80% win rate. Do you by any chance eliminate extremes on your Standard Deviations?
|
Russian Federation4235 Posts
|
5 Fantasy II 9.920
LOOOOOOOOL~! The weirdest map ever considered as top 5 balanced maps.
12 Katrina 6.942
I really wanna see Katrina playing at this meta, seems decent map for starleagues.
|
Japan11285 Posts
Interesting results, I somehow predicted Blue Storm will be number one but only that. The compiled stats from the pro era make it into a 'balanced' map statswise. It's quite surprising and funny how 815 III is more balanced than Outsider lol. Or how Monty Hall is more balanced than Rush Hour and Grand Line SE.
Can you say umm, do a bi-annual rankings of maps (to account for meta changes and maybe even outliers in S-class games)? ^_^
|
Where is paradoxxx? not even listed ^^
|
Kau
Canada3500 Posts
On October 07 2016 16:04 FiWiFaKi wrote:
25 Hitchhiker 5.707
28 Monty Hall 4.882
44 Outsider 3.713
Pretty good ranks for my 3 favorite maps :D
I wonder where Waiting to Panic would fall in this ranking haha.
|
Maybe instead of one giant comparison you could split it into 2-3 year intervals to get a more accurate idea since meta, talent, changes over time?
|
On October 07 2016 22:18 catabowl wrote:
How does this take into effect elite players? (Fantasy, Flash, Jaedong, Stork, Jangbi)
Example: Fantasy/Flash have an 80% win rate against zerg on map X - but the TvZ matchup on Map X is 50/50 - which includes the best players 80% win rate. Do you by any chance eliminate extremes on your Standard Deviations?
It doesn't, I mentioned it one of my posts on the first page, just to tough to do with the way I extracted my data, and it still took 4~ hours just to transfer it all over to excel.
On October 07 2016 23:11 c3rberUs wrote:
Interesting results, I somehow predicted Blue Storm will be number one but only that. The compiled stats from the pro era make it into a 'balanced' map statswise. It's quite surprising and funny how 815 III is more balanced than Outsider lol. Or how Monty Hall is more balanced than Rush Hour and Grand Line SE.
Can you say umm, do a bi-annual rankings of maps (to account for meta changes and maybe even outliers in S-class games)? ^_^
Not sure how a bi-annual ranking would take into meta changes in its current form. Each map currently gets a value assigned in one year value based on its year of maximum popularity from a tournament standpoint. I think that that the intervals are reasonably close together that it would make almost no difference at all. What I could do in theory is extract every game played on that map individually, and give each game a different weight. It's the same thing with S-class players, I'd have to have to calculate elite players in every time period, which would involve calculating everyone's winrate, and it's just a lot and a lot of things to consider, if I was the owner of 538, maybe I'd go into such depth, but it's really difficult for something like this.
Also keep in mind that for all maps, I have roughly 36,000 data points, unlike presidential polling, the aggregate sample size is a lot smaller here. From just going through every map, I'd say you need somewhere along the lines of 150~ maps to get a good indication of balance, even in a 35k data point poll of 177 maps, there will be plenty that don't reach that. Breaking it into more subgroups to perform statistical analysis will imo enter more uncertainty than good.
|
Yeah, I didn't think Outsider was so rough initially either (I'm a fan of the map), after all the worst match up on it is ZvP at 61.2%... It's just there's also the implicit imbalance of having 68 ZvZ played, and only 13 PvP, fairly indicative that zergs were very happy to play this map.
On October 07 2016 23:58 ppp87 wrote:
Where is paradoxxx? not even listed ^^
Rank 135, score of 0.757, only 8 maps with lower, but non-zero score. Small sample size of only 30 games, and a win rate of zero in one MU send it into the ground.
On October 08 2016 02:20 Kau wrote:Show nested quote +On October 07 2016 16:04 FiWiFaKi wrote:
25 Hitchhiker 5.707
28 Monty Hall 4.882
44 Outsider 3.713
Pretty good ranks for my 3 favorite maps :D I wonder where Waiting to Panic would fall in this ranking haha.
Should've gotten it to a Korean tournament, looking at that TL thread, looks silly, yeah
|
Statistics...
6 Destination 9.897 xDDDDDDDDDDDD
Being a map where zerg simply should not win vs any terran at all ^^
_
xD
|
I wonder where Korhal of Ceres would end up
|
On October 07 2016 23:06 outscar wrote:
5 Fantasy II 9.920
LOOOOOOOOL~! The weirdest map ever considered as top 5 balanced maps.
12 Katrina 6.942
I really wanna see Katrina playing at this meta, seems decent map for starleagues.
Katrina is way too vanilla, you'd barely notice any difference in how games are played except for TvPs going longer.
The hitchhiker/monty hall ratios are hilarious if you consider that they were made specifically to stunt MJY's record and he still went nearly even in ZvT.
|
What I would actually like to see is more detailed statistics of maps that are not only broken down by racial matchups but also by (relative) starting positions of the players. In theory a map that has any sort of strong positional advantage (be it through rotational design like FS or Aztec, through differences in horizontal vs. vertical matchups, like on CB, or through outright asymmetrical design, like Fantasy, which scored suspiciously high on this list, or just through significant disparity in 3-worker gas mining rates) could seem perfectly balanced in race statistics (due to the positional advantage overcompensating for racial disadvantages and starting positions being about equally distributed over sufficiently large sample sizes) but be in fact any amount of badly balanced (from an overall fair play point of view).
Another interesting inquiry would be to look at racial winning rates over game duration. This would also yield some interesting results, for example when looking at easily split two player maps with sparse expos, such as Matchpoint or Blue Storm.
|
I totally agree with tau cross. Very fast and open map. The fact that bases are very exposed with no cliff and 3 different paths makes it a favourite for offensive and creative gameplay. No wonder it would be one of the most balanced.
|
Seeing the place of Longinus II makes me shiver about the ones that are below this.
|
There are different methods for rating a map's balance within the professional scene (not how it plays out in the current era), and maps such as Tau Cross, Blue Storm, Fighting Spirit, and Circuit Breaker always tend to top the list no matter which algorithm you use. There's been discussions similar to this thread on Korean communities as well.
I wonder if you would be interested in starting a discussion over rating the various seasons of the ProLeagues according to their map pool, and how balanced they were. Which season of the ProLeague were the most skewed towards specific races, which players were the outliers, and such. Of course, the same sort of discussion could be made for the individual leagues, but the limited number of games, and the vastly different outcomes depending on the performances of a select few players would make the numbers more difficult to dissect in my opinion.
|
|
|
|
|
On October 10 2016 22:22 Endymion wrote:
blue storm lmfao
1000% perfect balance 
|
Talking about balance using stats makes no sense. Top dominating players skew results, balance shifts month to month and often isn't total. A map can be imba in regards to a few timings, but be difficult for that race in other ways, positional matchups, etc. A map could be balanced at S level but skewed heavily for the avg pro gamer. Not many of these maps have significant amounts of games on them, especially in regards to certain time frames. All of these issues cloud the use of stats to the point of being completely unhelpful. Hence the bizarre outcomes for this map list.
|
On October 12 2016 22:44 Dazed_Spy wrote:
Talking about balance using stats makes no sense. Top dominating players skew results, balance shifts month to month and often isn't total. A map can be imba in regards to a few timings, but be difficult for that race in other ways, positional matchups, etc. A map could be balanced at S level but skewed heavily for the avg pro gamer. Not many of these maps have significant amounts of games on them, especially in regards to certain time frames. All of these issues cloud the use of stats to the point of being completely unhelpful. Hence the bizarre outcomes for this map list.
Bizarre is subjective, as I've never seen a list compiled for say 50 maps or more than was ever "better" than this, and even now, people like to point out what feels wrong and out of place, but nobody has took an honest attempt at ever ranking them properly, and people choose to pick out what they don't like instead. Often times people only have a recollection of a small number of maps, so seeing one map so high might make them think "ooh, no this map was awful to play on", without even considering the other stuff.
I made it clear that there are plenty of issues with this method, but my opinion is that the person bias when trying to implement a map system is even worse than the one created by this (at least example has shown thus far), and hence I've chosen to go down this route.
|
Monty hall was tough for zerg
|
On October 13 2016 14:15 FiWiFaKi wrote:Show nested quote +On October 12 2016 22:44 Dazed_Spy wrote:
Talking about balance using stats makes no sense. Top dominating players skew results, balance shifts month to month and often isn't total. A map can be imba in regards to a few timings, but be difficult for that race in other ways, positional matchups, etc. A map could be balanced at S level but skewed heavily for the avg pro gamer. Not many of these maps have significant amounts of games on them, especially in regards to certain time frames. All of these issues cloud the use of stats to the point of being completely unhelpful. Hence the bizarre outcomes for this map list. Bizarre is subjective, as I've never seen a list compiled for say 50 maps or more than was ever "better" than this, and even now, people like to point out what feels wrong and out of place, but nobody has took an honest attempt at ever ranking them properly, and people choose to pick out what they don't like instead. Often times people only have a recollection of a small number of maps, so seeing one map so high might make them think "ooh, no this map was awful to play on", without even considering the other stuff. I made it clear that there are plenty of issues with this method, but my opinion is that the person bias when trying to implement a map system is even worse than the one created by this (at least example has shown thus far), and hence I've chosen to go down this route. No ones going to create a list because the entire notion isnt reasonable (not trying to sound dickish). What do you get out of this, even if you found a way around all the difficulties in interpreting the stats? You have a list of maps that were balanced when played. Which isnt important information. What matters is what features, combination of features, and actual maps, are balanced right this second. In any case, there is no way to actually create a methodology that will 'properly' show what maps are balanced, for the reasons I already gave.
And bizarre is not that subjective, we have genuine, objective understandings of what constitutes balance, at least in some respects. I.e gas at natural for protoss, fe has to be possible, certain designs of naturals to deal with mutalisk harass, symmetry etc. Your list contains maps with asymmetry, with giant naturals, and so on. You have maps that are considered balanced that we knew then, and certainly know now, are not actually balanced. One of the most imbalanced maps ever designed, fantasy 2, is considered one of the most balanced by your system. Thats a bizarre result.
edit: and I would put my trust in people who understood the meta game and what features of a map are beneficial in x context than someone who used this list, in an attempt to decide balanced maps. We can only determine balance through reasoned strategical discussions, not numbers.
|
What on Earth even is this? You fail to give an account of the algorithm that you created and decided to apply to these maps, and then claim that it's "objective" or that "the numbers are saying this." Do you not realize how absurd that is? I'm not even angry. I don't have anything at stake here. This is just so far down the rabbit hole into bizarro-land that I'm not sure how to react to it. Do you seriously think that a secret algorithm that you created without making any sort of argument as to its ability to properly display actual balance statistics could possibly hold even the weight of a single person's reasoned speculation on various maps? I really, really hope not. This is perhaps the purest form of the bizarre fixation on mathematical justification that keeps on cropping up that I've ever seen. News flash: just because something has been mathematized doesn't make it good. Mathematical models are only used because they map very well to what we already know, and are then used to extrapolate. Newtonian physics wasn't created in a vacuum; it was carefully put together so as to model the movement of planets. What you're doing here might be getting to the very heart of what balance is, but we have no way of knowing whether it does because we don't know why it is the way it is and have no way to verify or critique its results. As you have laid it out so far, your algorithm and your list mean absolutely nothing.
|
On October 14 2016 00:29 Dazed_Spy wrote:Show nested quote +On October 13 2016 14:15 FiWiFaKi wrote:On October 12 2016 22:44 Dazed_Spy wrote:
Talking about balance using stats makes no sense. Top dominating players skew results, balance shifts month to month and often isn't total. A map can be imba in regards to a few timings, but be difficult for that race in other ways, positional matchups, etc. A map could be balanced at S level but skewed heavily for the avg pro gamer. Not many of these maps have significant amounts of games on them, especially in regards to certain time frames. All of these issues cloud the use of stats to the point of being completely unhelpful. Hence the bizarre outcomes for this map list. Bizarre is subjective, as I've never seen a list compiled for say 50 maps or more than was ever "better" than this, and even now, people like to point out what feels wrong and out of place, but nobody has took an honest attempt at ever ranking them properly, and people choose to pick out what they don't like instead. Often times people only have a recollection of a small number of maps, so seeing one map so high might make them think "ooh, no this map was awful to play on", without even considering the other stuff. I made it clear that there are plenty of issues with this method, but my opinion is that the person bias when trying to implement a map system is even worse than the one created by this (at least example has shown thus far), and hence I've chosen to go down this route. No ones going to create a list because the entire notion isnt reasonable (not trying to sound dickish). What do you get out of this, even if you found a way around all the difficulties in interpreting the stats? You have a list of maps that were balanced when played. Which isnt important information. What matters is what features, combination of features, and actual maps, are balanced right this second. In any case, there is no way to actually create a methodology that will 'properly' show what maps are balanced, for the reasons I already gave. And bizarre is not that subjective, we have genuine, objective understandings of what constitutes balance, at least in some respects. I.e gas at natural for protoss, fe has to be possible, certain designs of naturals to deal with mutalisk harass, symmetry etc. Your list contains maps with asymmetry, with giant naturals, and so on. You have maps that are considered balanced that we knew then, and certainly know now, are not actually balanced. One of the most imbalanced maps ever designed, fantasy 2, is considered one of the most balanced by your system. Thats a bizarre result. edit: and I would put my trust in people who understood the meta game and what features of a map are beneficial in x context than someone who used this list, in an attempt to decide balanced maps. We can only determine balance through reasoned strategical discussions, not numbers.
Yeah, sorry, I disagree with you 100%.
Your argument just invalidates using numbers for everything, win rate for players? Completely meaningless according to you.
If we went for your idea of achieving balance, ie. every map has to be symmetrical, every map has to be possible to forge fe on, etc... Then you know what would happen? We would never even learn about maps that allow Forge FE. The only effective way we are able to accurately assess balance is through watching games on these maps and seeing how they go, not through theorycrafting, because there's too many variables for us to accurately model. That's the fundamental thing we disagree on, you say the "balance when played doesn't matter", you claim that features matter, and hence to me, your argument is laughable to me.
Constructing this was not meaningless from my viewpoint, as it allows easy sorting for which maps should and should not be considered for smaller tournaments (also for my own personal curiousity). We're using statistics here, so naturally, not all our data points are at their exact true value, there are outliers that are ranked higher or lower than they should be by some margins, but not completely unreasonable margins. Also, I can gladly name 200 maps less balanced than Fantasy II used in professional Korean BW, you are free to disprove every case on an individual basis. I'm not claiming that Fantasy is the 5th best balanced map ever, I agree that it's likely somewhat higher than what it should be. When looking at the numbers closely, the "issue" likely comes from most games played on it being from prelims, and I gave the same weighting to all the games (as taking this into account would take a tremendous amount of time).
|
On October 14 2016 00:49 Acritter wrote:
What on Earth even is this? You fail to give an account of the algorithm that you created and decided to apply to these maps, and then claim that it's "objective" or that "the numbers are saying this." Do you not realize how absurd that is? I'm not even angry. I don't have anything at stake here. This is just so far down the rabbit hole into bizarro-land that I'm not sure how to react to it. Do you seriously think that a secret algorithm that you created without making any sort of argument as to its ability to properly display actual balance statistics could possibly hold even the weight of a single person's reasoned speculation on various maps? I really, really hope not. This is perhaps the purest form of the bizarre fixation on mathematical justification that keeps on cropping up that I've ever seen. News flash: just because something has been mathematized doesn't make it good. Mathematical models are only used because they map very well to what we already know, and are then used to extrapolate. Newtonian physics wasn't created in a vacuum; it was carefully put together so as to model the movement of planets. What you're doing here might be getting to the very heart of what balance is, but we have no way of knowing whether it does because we don't know why it is the way it is and have no way to verify or critique its results. As you have laid it out so far, your algorithm and your list mean absolutely nothing.
It's objective in the sense that every map is held to the same criteria, and there is no internal bias produced. Apologies, I should have listed the algorithm sooner, it is:
Map Imbalance = (%TvZ winrate - 50)^2+(%PvT winrate - 50)^2+(%PvZ winrate - 50)^2 (standard Sum of squares error)
Mirror Imbalance = 50[[(#TvT mirrors - #ZvZ mirrors)/#PvP mirrors]^2+[(#PvP mirrors - #TvT mirrors)/#ZvZ mirrors]^2+[(#ZvZ mirrors - #PvP mirrors)/#TvT mirrors]^2]
Year multiplier:
2012 1
2011 1
2010 1
2009 0.95
2008 0.9
2007 0.85
2006 0.8
2005 0.7
2004 0.55
2003 0.4
2002 0.2
2001 0.1
2000 0.05
GamesPlayed: To give more value to the certainty of it being a good map, since played a lot makes the statistics have lower deviations, and implies people like to play the map
KoreanTourneysUsed: This multiplier is a small effect as GamesPlayed is closely correlated, but tourneys used adds a bit of a bonus by being embraced by many various parties, and indicative of universal likeness.
YearMultiplier: Explained on a previous page in this thread, older games bear less weight.
Map Score = sqrt(1000*ln(GamesPlayed)*1.05^ln(KoreanTourneysUsed)*YearMultiplier/(Map Imbalance + Mirror Imbalance))
|
Wasn't TC Protoss favored in PvT due to how you had to slow push along the edges and Protoss could very easily run across the middle and hit Terran anywhere?
|
On October 14 2016 04:50 FiWiFaKi wrote:Show nested quote +On October 14 2016 00:29 Dazed_Spy wrote:On October 13 2016 14:15 FiWiFaKi wrote:On October 12 2016 22:44 Dazed_Spy wrote:
Talking about balance using stats makes no sense. Top dominating players skew results, balance shifts month to month and often isn't total. A map can be imba in regards to a few timings, but be difficult for that race in other ways, positional matchups, etc. A map could be balanced at S level but skewed heavily for the avg pro gamer. Not many of these maps have significant amounts of games on them, especially in regards to certain time frames. All of these issues cloud the use of stats to the point of being completely unhelpful. Hence the bizarre outcomes for this map list. Bizarre is subjective, as I've never seen a list compiled for say 50 maps or more than was ever "better" than this, and even now, people like to point out what feels wrong and out of place, but nobody has took an honest attempt at ever ranking them properly, and people choose to pick out what they don't like instead. Often times people only have a recollection of a small number of maps, so seeing one map so high might make them think "ooh, no this map was awful to play on", without even considering the other stuff. I made it clear that there are plenty of issues with this method, but my opinion is that the person bias when trying to implement a map system is even worse than the one created by this (at least example has shown thus far), and hence I've chosen to go down this route. No ones going to create a list because the entire notion isnt reasonable (not trying to sound dickish). What do you get out of this, even if you found a way around all the difficulties in interpreting the stats? You have a list of maps that were balanced when played. Which isnt important information. What matters is what features, combination of features, and actual maps, are balanced right this second. In any case, there is no way to actually create a methodology that will 'properly' show what maps are balanced, for the reasons I already gave. And bizarre is not that subjective, we have genuine, objective understandings of what constitutes balance, at least in some respects. I.e gas at natural for protoss, fe has to be possible, certain designs of naturals to deal with mutalisk harass, symmetry etc. Your list contains maps with asymmetry, with giant naturals, and so on. You have maps that are considered balanced that we knew then, and certainly know now, are not actually balanced. One of the most imbalanced maps ever designed, fantasy 2, is considered one of the most balanced by your system. Thats a bizarre result. edit: and I would put my trust in people who understood the meta game and what features of a map are beneficial in x context than someone who used this list, in an attempt to decide balanced maps. We can only determine balance through reasoned strategical discussions, not numbers. Yeah, sorry, I disagree with you 100%. Your argument just invalidates using numbers for everything, win rate for players? Completely meaningless according to you. They arent all that meaningful, actually. You have to look at the numbers, the maps, the players they went against, and the events that actually unfolded-- in other words you'd have to understand the game and their career before you made random assumptions about the quality of their play. All you'd get by looking at their numbers is an aggregate across years, it gives you no comprehension greater than that.
If we went for your idea of achieving balance, ie. every map has to be symmetrical, every map has to be possible to forge fe on, etc
Thats essentially how maps have been balanced for years now... Then you know what would happen? We would never even learn about maps that allow Forge FE. Hm? Are you trying to say that if we never deviated from what was assumed to be balanced we would of never experimented, and therefore never understood the game better? Thats not what you wrote but its my best guess. Well, trying to design maps based on our actual understanding of what constitutes a balanced map in theory, does not rule out experimentation, or the increase of knowledge (which itself would spur different map developments). It just means we design based on knowledge, and not just randomly pooling numbers together. The only effective way we are able to accurately assess balance is through watching games on these maps and seeing how they go, not through theorycrafting, because there's too many variables for us to accurately model. Yes...you would look at the games to see how they go, but you would also theorycraft. There clearly is not "too many variables" to theorycraft, because I can give you (and have) concrete features of maps that we know are important for balance, that we can know through theory, without observation. Observation confirms it, but I dont need to watch progamers play on a map with the original blue storm natural before I say 'actually, a natural like that will make 2 hatch mutalisk nearly impossible to stop, making it very ZVT favoured'. And we as a community didnt come to that conclusion because we pooled up thousands of games played on that map and looked at the total winrate, no, we simply discussed how the game plays, what advantages are accrued though x feature, and so on. Its theorycrafting and experience that brings about knowledge. Your just...grabbing numbers. Meaningless. Shallow.
That's the fundamental thing we disagree on, you say the "balance when played doesn't matter", you claim that features matter, and hence to me, your argument is laughable to me. I honestly have no idea at all how you got that interpretation, I think there is a language barrier here. This simply isnt close to what I said.
Constructing this was not meaningless from my viewpoint, as it allows easy sorting for which maps should and should not be considered for smaller tournaments
If people use your map list as a guide for balance, I can tell you right now they will be disappointed. Many of your balanced maps are unplayable due to imbalance.
|
Just last page people were saying that Bluestorm is heavily T favored, and now you're claiming heavily Z favored?
Have you ever thought that maybe there's many ways to play this game, and having an idea about how the game should be played is automatically not objective? Maybe it is a tougher map to play against 2 hatch muta, but you have stronger follow-ups, it might be a good map to do one base plays on, etc... Many variables balance out to even out.
As an extreme example for the sake of an example, imagine you have a natural without a gas, but the map is an island map, do your initial statements hold, no. You can't isolate each map element individually, and there are too many elements to treat with relationship to one another. People cannot just look at a map and know "this will be a good map", unless they want the exact same playstyle they've seen on another map that is essentially the same.
My argument is that there are some 250~ maps used in pro broodwar, and nobody in the world has the knowledge of all of these maps to be able to rank them all with any meaningful understanding. Heck, just watching ONE game that is 15 minutes long from each non-mirror match up would take 188 hours... And hence having a system to measure roughly how they stack up is meaningful.
I agree that it's not a perfect list, but it provides useful guidance on which maps should be automatically disqualified due to imbalance, until you narrow your list to say 30 maps (from all the maps in existence, no errors due to omission or just not being familiar with a map)... And then do a more careful analysis such as:
-What maps do the players like to play
-What maps make for interesting games
-Which maps will be most fair to all the races
-Having enough variety: 2/3/4p maps, different tilesets, playstyles, etc
Also keep in mind this list stops at 2012, so saying things have been like that for years is because all the Afreeca streamers just play the same 2-3 maps over and over again.
Hm? Are you trying to say that if we never deviated from what was assumed to be balanced we would of never experimented, and therefore never understood the game better? Thats not what you wrote but its my best guess. Well, trying to design maps based on our actual understanding of what constitutes a balanced map in theory, does not rule out experimentation, or the increase of knowledge (which itself would spur different map developments). It just means we design based on knowledge, and not just randomly pooling numbers together.
Yes, that is what I'm trying to say. In fact, most businesses operate by using numbers to analyze their decisions. For better or worse, the entire SC2 map-pool and entire race balance was done this way from the beginning of beta to up until recently, and we are talking about one of the largest video game companies and the most successful RTS of the recent generation. Not to mention, investors look at numbers when deciding things, companies look at numbers when deciding what projects to take on... Often times, the itty gritty details that the numbers don't show are too difficult to put into a data structure, and hence the numbers is the best we have.
This is the case I believe we have with BW, players who play only play one race, and imo there's plenty of inherent bias with opinions on these kinds of things. Secondly, I play BW... And really, how many different maps do you play on a regular basis that you could comment on the balance of? As for watching, we have such a limited map pool now that 100 maps would be completely ignored as nobody thinks about them anymore, even though they are very good. And then there's people who have preconceived notions about what a map "should be".
This list doesn't discriminate. It takes a list of the use and results of all the maps, and tells you which are playable and which are not (within a margin of error, because some factors are not internalized, and usually they are not significant, but they can be - but at least it prevents any human bias). The rest is up for you to decide.
This is what I'm here for to provide.
edit: Lastly, obviously this is for pro-level players, balance achieved across all levels is more or less impossible, so that should be a given, results may vary for your typical C- iccup player.
|
On October 10 2016 17:47 b0lt wrote:I'm curious how low Central Plains ranked.
A score of zero ^_^
0 TvT, 1 ZvZ, 27 PvP... What more needs to be said, almost on the level of Battle Royal.
|
On October 14 2016 08:10 FiWiFaKi wrote:
Just last page people were saying that Bluestorm is heavily T favored, and now you're claiming heavily Z favored? No, I said the original natural construction of Bluestorms natural was heavily Z favoured. Not modern bluestorm. Which is another problem of your list; why would you combine earlier versions of maps with newer ones? The whole point your trying to get at is balance, the difference between different versions of maps are...balance oriented.
Have you ever thought that maybe there's many ways to play this game, and having an idea about how the game should be played is automatically not objective? There are many ways...and the best ways change week by week or month by month, but there are optimal ways to play, that players can be aware of, principles that extend year to year, consistent features of the metagame we can speak of with certainity. There is knowledge in this community, and it is not improved one whit by throwing numbers at it. Maybe it is a tougher map to play against 2 hatch muta, but you have stronger follow-ups, it might be a good map to do one base plays on, etc... Many variables balance out to even out. That was my original point in saying that this list is nonsensical...a map may have skewed results in favour of one race, while in reality its actually only hard to stop 'x' push on that map. Five months down the line that build may be totally unviable and the map actually quite bad for the race that once dominated it. The results of this rapid balance shift could give close to a 50% result. When you look at numbers all you see are results, not nuance.
As an extreme example for the sake of an example, imagine you have a natural without a gas, but the map is an island map, do your initial statements hold, no. You can't isolate each map element individually, and there are too many elements to treat with relationship to one another. People cannot just look at a map and know "this will be a good map", unless they want the exact same playstyle they've seen on another map that is essentially the same.
There are going to be certain constancies in playstyle if your looking for consistent and balanced results. Period. One base builds are not balanced, we've had years of experience to understand that. Protoss without access to gas late game can not compete with zerg late game, we've had years of experience to understand that. But island maps! Is not an argument against either of these features, its simply a different conditional scenario. In order words your confusing your own pedantry with an argument.
My argument is that there are some 250~ maps used in pro broodwar, and nobody in the world has the knowledge of all of these maps to be able to rank them all with any meaningful understanding. Heck, just watching ONE game that is 15 minutes long from each non-mirror match up would take 188 hours... And hence having a system to measure roughly how they stack up is meaningful. Again...people dont need a list. Lists are not helpful. Your map has sin baekdu as one of the most balanced maps, but this map is a decade old, and with meta game changes is in fact ridiculously bad.
I agree that it's not a perfect list, but it provides useful guidance on which maps should be automatically disqualified due to imbalance, No. It. Doesnt. What it shows are RESULTS, not balance. If you cant give any understanding for any of the myriad factors that could impact results, then simply looking at the data is shallow and meaningless. out of a 100 games that go 50-50 for either side, how many were between players of skill imbalance? Was there equal preparation between the players? What was the metagame like during that particular period, because overall racial metagame balance is obviously going to impact balance on particular maps as well, yes? How many of these games were cheese? How many of these cheeses were done due to the balance of the map, and how many succeeded or failed because of the balance of the map? What features are imbalanced, and why? In what way were they, and has the metagame since then already found a solution? and on and on and on and on. You have no answers for any of this, just the inane reply of "data!". Sorry, not good enough. Heres you top 20 "balanced maps"
1 Blue Storm 20.670 <<< historically was very rigged z> t, then became very t>z favoured
2 Tau Cross 18.142 <<<< historically was alright for tvp, but when it came back a few years later was P >>> t
3 Fighting Spirit 14.826 <<<< horrible zvt map in the current metagame, was previously more balanced
4 Circuit Breaker 11.952
5 Fantasy II 9.920 <<<< positionally and racially imbalanced all over the place, unplayably bad.
6 Destination 9.897 <<< extremely abusive for p at the start, dominating for lategame for a terran turtle, was a good pure mech map in tvz but thats fallen out of favour, pure mech would likely get curb stomped now-- and its not a good map to open bio, so mech switch is going to be poor.
7 Neo Medusa 9.396
8 Colosseum II 8.781
9 Sin Peaks of Baekdu 7.545 <<<< very very imbalanced in the modern meta game, simply an out of date map
10 Match Point 7.285
11 Luna 6.992 <<<< very very imbalanced in the modern meta game, simply an out of date map
12 Katrina 6.942 <<< carrier hell intially that then got countered heavily by the 3/3 flash meta, not sure where how this map would end up in the modern meta game.
13 La Mancha 6.623
14 Eye of the Storm 6.513
15 Heartbreak Ridge 6.474
16 Python 6.442 <<<< very very imbalanced in the modern meta game, simply an out of date map
17 Arcadia 6.258
18 Neo Requiem 6.015 <<<< very very imbalanced in the modern meta game, simply an out of date map, also imba just generally p> t
19 Nostalgia 6.008 <<<< very very imbalanced in the modern meta game, simply an out of date map
20 Longinus 2 5.980 <<<< very very imbalanced in the modern meta game, and was never a balanced map, actually.
Anyway, im done with this conversation. I've clearly repeated myself for three posts in a row, I dont think this is productive anymore.
|
I understand what you're saying, but once again, I think you're wrong.
Your entire argument is that not enough factors are internalized by the model, and hence the model has no value. I made some assumptions that I stated that will allow balance to predicted from results, and that's what I did. For example, TLPD uses an ELO system to rank player skill levels, once again, this doesn't internalize everything, but it's a good indication of how good a player is (does a better job than my model for maps anyway). People don't care about some subjective opinion ranking, and it's not useful or practical, unless it's for a very small sample size, such as what the BW PR was. All sports, whether its tennis, chess, golf, hockey... They all have a ranking system for how good a player or team has done throughout the season, and it's what goes. Yes, someone might have gotten hella lucky that their opponent got injured or something, and hence they got further they would otherwise, but it's how it is (people remember the outcome, not the performance).
All your arguments are theorycrafting about whether they will work or not, even though you haven't seen these maps played in 5 years, that's like a civil engineer using only numbers on his paper before he constructs a bridge.
You value the "current meta" too much, it's completely silly. The game is still the same as it's been for more than a decade, the game and its understanding really didn't change much compared to 2008-2009, where most of these maps were heavily used. If people like Jaedong, Flash, and Bisu were playing on these maps in 2008-2012, and after dozens or hundreds of games, if some of these things couldn't be figured out, chances are they aren't completely game breaking, or are very slight imbalances that can be overcome by a slightly superior player. Remember, ranking the most balanced map is the same as ranking the least imbalanced map, maybe that's a way to help you think about it. It's cool that you want to criticize all these maps, because apparently in your view all maps are bad (or you are welcome to go create your own list)... Anyway, easy to say what sucks, yes offer an alternative, so I'll take most of that with a grain of salt.
I made this ranking because I've never seen something done like this before, and it had my curiosity. So I invested a bit of time and thinking in making this up, to give back just a little bit. I don't know of any other quantitative analysis of maps done before, so I thought that some people might appreciate it.
Either way, nice talk, goodbye.
|
On October 07 2016 18:27 Miragee wrote:Show nested quote +On October 07 2016 18:12 FiWiFaKi wrote:On October 07 2016 18:07 duke91 wrote:On October 07 2016 17:56 FiWiFaKi wrote:On October 07 2016 17:42 r33k wrote:On October 07 2016 17:26 Miragee wrote:
The problem with balance is that the meta-game shifted and still shifts a lot in BW. New maps help to adjust to imbalances due to metagame shifts. On the other hand, some maps that were quite balanced for a while are now imbalanced because the meta changed. Fighting Spirit is a good example. In it's heydays it was probably the most balanced map ever with less than 3 % deviation from 50 % winrate in all matchups. However, in this era of BW it shows some problems, although some of the problem might be caused by the lack of enough strong zerg players.
And in reverse, reimplementing some maps after a series of meta shifts might lead to different and possibly better balanced resluts. Also I really wouldn't lump anything pre-2008 with modern BW. The game as it was played before doesn't even compare. If we want to be more precise 2010-2012 was the pinnacle of the game, as post-KeSPA BW shows the lack of organization and practice infrastructure. On October 07 2016 17:38 Miragee wrote:On October 07 2016 17:32 XenOsky- wrote:in my opinion as a protoss player, i think the most balanced maps were Destination and Heartbreak rdige. + Show Spoiler + I thought protoss players hated Destination in both match ups. I personally liked both of those maps but Destination certainly seemed to be a bit imbalanced, at least in TvP (terran favoured). Destination was fine for the races that mattered. And my model takes that into account, and treats 2010-2012 as the most important time. Otherwise maps like LT would be way higher up. The meta shifted compared to 2012 But 2010-2012 was the pinnacle of BW, the time it was played at the highest level. Of course it's difficult to treat the data (and some things that data cannot quantify), and I made the most reasonable assumptions I could to arrive at answers. I'm not here to argue whether x map is #6 or #7, but mostly as a general guideline of how the maps stack up. Particularly at the very very top and the very very bottom, there can be some issues, since as you get down a lot, your uncertainty goes up due to small sample sizes, and as you reach the top, the difference between a 51% winrate and a 52% winrate becomes very significant (the perfectly balanced map results wise, as well as even mirror match up distribution would results in a map score of infinity). The pinnacle probably started before 2010. Irrc I would at least include 2009 as well.
Lost Saga (?) MSL in 2009 was the end of the old era, it was the last time that old school players (Savior and Nada in this case) got to the Ro8 of a major tournament. But the trend of old players being unable to keep up with a bunch of new macro monsters who all practiced 12-14 hours a day began in 2007 or so.
|
On October 14 2016 05:00 FiWiFaKi wrote:Show nested quote +On October 14 2016 00:49 Acritter wrote:
What on Earth even is this? You fail to give an account of the algorithm that you created and decided to apply to these maps, and then claim that it's "objective" or that "the numbers are saying this." Do you not realize how absurd that is? I'm not even angry. I don't have anything at stake here. This is just so far down the rabbit hole into bizarro-land that I'm not sure how to react to it. Do you seriously think that a secret algorithm that you created without making any sort of argument as to its ability to properly display actual balance statistics could possibly hold even the weight of a single person's reasoned speculation on various maps? I really, really hope not. This is perhaps the purest form of the bizarre fixation on mathematical justification that keeps on cropping up that I've ever seen. News flash: just because something has been mathematized doesn't make it good. Mathematical models are only used because they map very well to what we already know, and are then used to extrapolate. Newtonian physics wasn't created in a vacuum; it was carefully put together so as to model the movement of planets. What you're doing here might be getting to the very heart of what balance is, but we have no way of knowing whether it does because we don't know why it is the way it is and have no way to verify or critique its results. As you have laid it out so far, your algorithm and your list mean absolutely nothing. It's objective in the sense that every map is held to the same criteria, and there is no internal bias produced. Apologies, I should have listed the algorithm sooner, it is: Map Imbalance = (%TvZ winrate - 50)^2+(%PvT winrate - 50)^2+(%PvZ winrate - 50)^2 (standard Sum of squares error) Mirror Imbalance = 50[[(#TvT mirrors - #ZvZ mirrors)/#PvP mirrors]^2+[(#PvP mirrors - #TvT mirrors)/#ZvZ mirrors]^2+[(#ZvZ mirrors - #PvP mirrors)/#TvT mirrors]^2] Year multiplier: 2012 1 2011 1 2010 1 2009 0.95 2008 0.9 2007 0.85 2006 0.8 2005 0.7 2004 0.55 2003 0.4 2002 0.2 2001 0.1 2000 0.05 GamesPlayed: To give more value to the certainty of it being a good map, since played a lot makes the statistics have lower deviations, and implies people like to play the map KoreanTourneysUsed: This multiplier is a small effect as GamesPlayed is closely correlated, but tourneys used adds a bit of a bonus by being embraced by many various parties, and indicative of universal likeness. YearMultiplier: Explained on a previous page in this thread, older games bear less weight. Map Score = sqrt(1000*ln(GamesPlayed)*1.05^ln(KoreanTourneysUsed)*YearMultiplier/(Map Imbalance + Mirror Imbalance))
I greatly appreciate that you posted this. I'm not prepared to give a full critique of the exact algorithm at this point, so all I'll do is suggest that in situations with few points of concrete data and significant confounding factors, some of which are out of our ability to even measure, a statistical analysis is going to yield less valuable results than a well-considered expert analysis. This sort of follows (but is not exactly the same as) the whole black-swan issue that, as an economics major, I'm sure you've heard about. With Broodwar maps and balancing, we are confronted with a very small number of recorded games that aren't necessarily reflective of the balance of those maps, but rather reflect the individual skill and preparation of the similarly small number of humans engaging in them. Your system does a good job of generally weeding out the astonishingly terrible maps and nudging better ones up to the top, but it fails to account for many of the nuances that show up in actual play. Take, for example, a theoretical map that starts out as apparently overpowered for one race over another race by some small margin, with play lingering around there for a while (for whichever reason, the map stays in the pool for a long time). Then, all of a sudden, an innovation reveals a deep yet hidden imbalance in the map which sets the previously disadvantaged race at a massive advantage. A fairly small number of games are played here, and then the map is removed from the available pools because of its age and the imbalance. Your system might incorrectly rank this map as being quite balanced, when the reality is that it was removed from the pool for being so incredibly imbalanced. An expert observer would spot this instantly, being familiar with the history of the map. Your algorithm, then, is only as strong as your own observations and efforts to correct it in favor of what we already know - making it already dependent on the account of an expert observer for its own accuracy. This is a problem every time someone attempts a mathematical solution to a problem in the social sciences. In that way, it is literally impossible for your algorithm to be worthwhile, unless you somehow got access to a ton of ICCUP or Fish records and got around the data pool problem. That's what I wanted to point out.
|
On October 13 2016 19:26 iPlaY.NettleS wrote:
Monty hall was tough for zerg
Monty Hall, like most maps from that period of time was designed to make MJY lose.
|
There was a TLPD page that had BW map balance using an algorithm that took player skill into consideration, similar to what people are suggesting here. You had to have the URL though, there was no link to it. I think it's gone now.
If P4ndemik or PoP still post they might know what I'm talking about.
edit: It was at http://www.teamliquid.net/tlpd/korean/maps/balance_table.php and it's gone now. 
|
On October 15 2016 05:46 r33k wrote:Monty Hall, like most maps from that period of time was designed to make MJY lose.
This post made me wonder if it's possible to create a map that guarantees victory in certain matchups.
You can make a map that has an exactly 50% win rate for every matchup pretty easily, but it's not very interesting (just give one of the starting locations no minerals).
Terran can be given a 100% win rate by giving none of the starting locations minerals, and adding an island with minerals.
You can give zerg a 100% win rate by giving both starting points exclusive access to a cliff over the other starting point's minerals, with creep from neutral hatcheries blocking :
![[image loading]](http://i.imgur.com/Un3Esnj.png)
Is it possible to give protoss a 100% win rate on a map?
|
That's interesting, I didn't know maps were created for MJY to lose. This makes his record even more impressive.
OP I think the stats were interesting but as people have said, there are many imperfections in this method of analysis, akin to the problems faced in social 'sciences'.
For that matter what you're analysing cannot be compared to sports because they only have one "map". So in BW, maps are a different dimension, which is absent in other sports.... except maybe rainy weather vs a sunny day or a football pitch at a high altitude.
On October 15 2016 12:50 b0lt wrote:Show nested quote +On October 15 2016 05:46 r33k wrote:On October 13 2016 19:26 iPlaY.NettleS wrote:
Monty hall was tough for zerg Monty Hall, like most maps from that period of time was designed to make MJY lose. This post made me wonder if it's possible to create a map that guarantees victory in certain matchups. You can make a map that has an exactly 50% win rate for every matchup pretty easily, but it's not very interesting (just give one of the starting locations no minerals). Terran can be given a 100% win rate by giving none of the starting locations minerals, and adding an island with minerals. You can give zerg a 100% win rate by giving both starting points exclusive access to a cliff over the other starting point's minerals, with creep from neutral hatcheries blocking : Is it possible to give protoss a 100% win rate on a map?
Very interesting thought experiment. I can't think of anything for protoss either.
|
On October 15 2016 12:50 b0lt wrote:Show nested quote +On October 15 2016 05:46 r33k wrote:On October 13 2016 19:26 iPlaY.NettleS wrote:
Monty hall was tough for zerg Monty Hall, like most maps from that period of time was designed to make MJY lose. This post made me wonder if it's possible to create a map that guarantees victory in certain matchups. You can make a map that has an exactly 50% win rate for every matchup pretty easily, but it's not very interesting (just give one of the starting locations no minerals). Terran can be given a 100% win rate by giving none of the starting locations minerals, and adding an island with minerals. You can give zerg a 100% win rate by giving both starting points exclusive access to a cliff over the other starting point's minerals, with creep from neutral hatcheries blocking : Is it possible to give protoss a 100% win rate on a map?
This stuff was already discussed in Korean communities.
Have a map that's similar to the one that gives zergs a 100% win rate, have the gas already depleted so that any other build paths are near obsolete, and remove the creeps on the cliffs, and design the cliff so that the minerals are out of range by marines but in range of photon cannons, and use the smaller size of the photon cannons so that the "sweet spot" that is in range of the minerals have the exact size that fits a photon cannon.
In order to make defenses from the lower ground useless, have random objects all around the mineral field so that you cannot place any defensive towers.
This way only protoss can harass the enemy through the cliffs, and any other build paths would be taken advantage of any capable player playing protoss.
|
On October 15 2016 05:46 jalstar wrote:There was a TLPD page that had BW map balance using an algorithm that took player skill into consideration, similar to what people are suggesting here. You had to have the URL though, there was no link to it. I think it's gone now. If P4ndemik or PoP still post they might know what I'm talking about. edit: It was at http://www.teamliquid.net/tlpd/korean/maps/balance_table.php and it's gone now.
Doing it through TLPD would be ideal, because that was you'd have all the data at your disposal immediately, particularly things it could remedy:
1) Chess ELO works that for a 400 point difference, you have a 10/11 chance of winning. I don't know the exact Liquipedia formula, but you could take the instantaneous ELO at every point and combine it into win statistics for every game. This would give a weighted value of win rate, the ELO used could be some combination of vs match-up ELO and total ELO. This would remove the assumption that players of all races are equal in skill level, and it'd take into account good players being bad players on an imbalanced map (I think with sample sizes this large, this effect would be almost negligible, but many of your disagree).
2) It could take into account the number of games played by each race in different timeframes. Currently my assumption was that equal number of games by all races on all maps (I think when looking at large enough time frames, it's reasonably accurate)... But the issue this causes is that if there's 80 TvT's played and 40 PvP's, my algorithm reads this as hey, it's Terran favoured, as Protoss are trying to dodge playing on this map (particularly in PL). However it's possible that there are only half as many protoss as terran, and hence that's not the case. It'd fix that bias, I think this could/would definitely provide a benefit to the current system, particularly at the upper end of balance
3) Instead of giving the map one single multiplier for a year of maximum popularity, it's balance could be calculated at a few different periods (or give a slightly larger weighting to the later on areas to calculate the winrate). For example, games in 2012 would count as 1.5 games, so if it goes 30-30 in 2012, but 30-20 in 2008 (which would have an arbitrary value of 1), the weighted winrate would be 53.57% vs 54.55%.
4) Potentially give a different weighting to different tournaments depending on what VETO system they have, since no-vetoes make it that the amount of mirrors played isn't as significant as in say PL, where teams will never put out a terran on a map where Terran does poorly.
Some things could never be taken into account, such as the manner in which the game was won, among other things like how good a player really is, for example iloveoov's dominance in numbers never looked all that impressive to me, same thing about say Conor McGregor in UFC, that's more about style and aura that's given off.
Anyway, a statistical analysis is always imperfect, the world has too many variables and too much chaos, but often it's the best we can do, so making the best model possible and analysing it might give the most meaningful results for the discussion.
edit: Not enough data in TLPD to gather data for winrate vs time, but it'd be neat to see. I suppose you could add it into the balance model, but not necessarily sure what's more balanced - 100% winrate in early game and 0% winrate in lategame, versus 50-50 in both (assuming same number of games ending in early and lategame). I quite like the dynamic that you just need to weather the storm and then it's smooth sailing (zerg trying to get out defilers out in time for example), particularly common in Dota. Don't want to make too many subgroups of the data though, as outside of the 20~ big maps, the sample sizes for the rest aren't large.
|
I would like to see what are the most balanced maps by matchup.
Even more, I would love to have map designers make maps for specific matchup only, and turnaments that would use maps dynamically, depending on matchup.
|
Blood Bath is still the best map.
|
On October 16 2016 07:57 arbiter_md wrote:
I would like to see what are the most balanced maps by matchup.
Even more, I would love to have map designers make maps for specific matchup only, and turnaments that would use maps dynamically, depending on matchup.
name them "tears of the random player", "picker's outrage" and "zerg suck hard".
|
|
|
|
|
|
EPT
