Map making is an important part in Brood War balance. Be it cool map features like mineral walls or lurker egg blocks, or just plain and simple standard maps, they always played a role in how each race played out a map. However, one map stood head and shoulders above the rest in terms of balance, and has been featured in nearly every league and tournament since its debut. It is no doubt this map is Fighting Spirit.
Since the map first came out, there was nothing but praise for the map. A map that all three races felt was beneficial to their playstyles. A standard natural, relatively close and easy third, and a wide open center with a few walls. It is no question why this map led the way for future map designs and changed the very definition of standard map making in Brood War. Maps like Jade, Sniper Ridge, La Mancha, and Icarus all share that classic similar safe third.
Fighting Spirit
TvZ
ZvP
PvT
KeSPA era
148-139 (51.6%)
134-117 (53.4%)
143-137 (51.1%)
Recently, Fighting Spirit has been eliminated from the map pool in SSL11, after about 5 ½ years leading tournament map pools. However, the community welcomed the change, saying it was a breath of fresh air and that the map had its weak points and flaws. Therefore, we must dive in to the truth behind Fighting Spirit and maps that follow its lead.
Hypothesis #1: The map balance is wrong.
In the current post-KeSPA era, the map has had very a skewed balance towards Terran with Zerg and Protoss getting the short end of the stick.
Fighting Spirit
TvZ
ZvP
PvT
KeSPA era
148-139 (51.6%)
134-117 (53.4%)
143-137 (51.1%)
post-KeSPA era
253-143 (63.9%)
187-184 (50.4%)
194-225 (46.3%)
So why is this? Why was the KeSPA era stat line so evenly balanced while in the current era it's so unbalanced? The map itself did not change during the transition so statistics should remain the same. Some may suggest that with more time, the pros are able to abuse every small advantage they could, and Terran players ultimately came out on top. However, I am very skeptical. This argument suggests that both players and coaches could not solve the map in its reign of over 5 years, which seems extremely doubtful, even laughable.
But is it? If we look at similar map statistics using the maps I listed above that use a similar safe third and optional third due to spawn imbalance, would we find a trend? Here are the stats for each map: (Sniper Ridge and Jade both had slight remakes, so I will be combining their games played to formulate a better conclusion)
Fighting Spirit and Similars: KeSPA Era
TvZ
ZvP
PvT
Fighting Spirit
148-139 (51.6%)
134-117 (53.4%)
143-137 (51.1%)
Sniper Ridge
17-14 (54.8%)
10-9 (52.6%)
19-21 (47.5%)
Jade
6-3 (66.7%)
10-12 (45.5%)
25-18 (58.1%)
Icarus
53-57 (48.2%)
41-29 (58.6%)
25-29 (46.3%)
La Mancha
63-57 (52.5%)
76-53 (58.9%)
87-62 (58.4%)
Looking at each individual map, each map does show some blatant imbalances. Sniper Ridge favors Terran, Jade favors Protoss, Icarus favors Zerg, and La Mancha is equal for all races, but skewed towards each imbalanced matchup (T>Z>P>T). Jade and Sniper Ridge both have relatively small sample sizes to pull from, especially with Jade in the TvZ matchup. So to help with those lacking in sample size, let's pull up the post-KeSPA stats:
Fighting Spirit and Similars: KeSPA and post-KeSPA
TvZ
ZvP
PvT
Fighting Spirit (KeSPA)
148-139 (51.6%)
134-117 (53.4%)
143-137 (51.1%)
Fighting Spirit (post-KeSPA)
253-143 (63.9%)
187-184 (50.4%)
194-225 (46.3%)
Sniper Ridge (KeSPA)
17-14 (54.8%)
10-9 (52.6%)
19-21 (47.5%)
Sniper Ridge (post-KeSPA)
39-39 (50.0%)
40-28 (58.8%)
27-24% (52.9%)
Jade (KeSPA)
6-3 (66.7%)
10-12 (45.5%)
25-18 (58.1%)
Jade (post-KeSPA)
58-45 (56.3%)
54-42 (56.3%)
70-62 (53%)
La Mancha (KeSPA)
63-57 (52.5%)
76-53 (58.9%)
87-62 (58.4%)
La Mancha (post-KeSPA)
10-9 (52.6%)
17-9 (65.4%)
9-9 (50.0%)
(Icarus not included since only 4 games were played on it).
If we look at this much more prominent stat line, you'll notice every map is skewed towards the standard imbalance. So while no map distinctively favors one race, they are all imbalanced towards the same matchups. (Perhaps another article about why there is race imbalance another time).
But is Fighting Spirit entirely balanced? In a discussion with kogeT, he mentions a few minor aspects of the map that add up to give Terran a favorable advantage.
There are no cliffs for muta micro or lurker abuse.
It enables easy map splitting late game which favors Terran.
The middle expansion inherently favors tanks and Terran.
Main mineral lines are protected by the boundaries of the map (in the corners).
Simcity/building placement concept enables a safe natural (maybe a little too safe).
Here’s my response to this. While these features are definitely worth noting, there are some flaws to this theory. Map splitting rarely comes into play, as much TvZ games end well before the map splits. However, it is noted that with tanks, map splitting becomes incredibly advantageous for Terran:
Continuing on, the middle expansion almost never comes up in TvZ games, but more commonly comes up in TvT games. Therefore, this is a non-factor. The main mineral lines being protected by the map does hinder mutalisk harass; however, most maps have adopted the corner base layout for years now, so this is also relatively unconvincing. Building placement structures have existed for a while, but it should be noted that the chokes of naturals have gotten noticeably smaller and smaller. This hinders the ability for zergling run-bys and surrounds.
However, I still believe these things in total do not put Zerg players at such a disadvantage to warrant the 64% win rate we see on FS. Maybe something closer to 55% would be more reasonable, but the current win rate still stands to be a little more exaggerated. But let’s say they did, and that Fighting Spirit really is a Terran favored map against Zerg for the above reasons. Why did Terran players not dominate during the KeSPA-era then? 51.6% is near even, so there is no evidence that these map imbalances played a role then.
Ultimately, there is no clear evidence that Fighting Spirit is imbalanced due to map design. While koget brings up potential ideas for why Fighting Spirit is Terran favored, the current extreme win rate we see is far from what they should be. Which brings me to my second theory...
Hypothesis #2: Player quality has changed.
If we look back to the KeSPA era, each race had their fair share of S and A class players.
Protoss was led by their well-recognized heroes, Bisu, Stork, and JangBi, the most popular and skilled of the 6 Dragons. The rest of Protoss who followed in their footsteps involved Stats, Movie, Horang2, Snow, Shuttle, BeSt, and free. (10)
So in terms of player quality, each race had their fair share of players to hold the torch and pull their own weight. So how about now? Above all, Terran players have dominated. And it's no surprise. The Terran have been making up over 50% of top8s consistently (Kongdoo Starz League only featured 2 Terran in the top8 most likely due to the lack of sSak and Last).
Zerg, while in numbers, have failed to ultimately impress and display their true potential. Big names like ZerO, EffOrt, Killer, Kwanro, and hero have shown up, but while quality Zerg players are close to that of Terran, they still do not stand a chance against them. (5)
Protoss has also failed to impress me, obviously aside from our lord and savior Bisu. free, Snow, Movie, Jaehoon have all made a noteworthy appearances, but they also stand no chance against Terran players. (5)
In the top 10 Elo ranks alone, Terran players often make up half of the slots with only 1 Protoss in the top (Bisu). In fact, ZerO and especially Killer, may be ranked too high considering their current performances.
So given all of this data and evidence, does this really demonstrate that player quality affects map statistics like Fighting Spirit? While I was previously an advocate of this hypothesis, the more research I conducted, the more I realize that this is simply not the case as much as it seems. If Terran power was really the reason for Fighting Spirit's imbalance, then every other map would also show similar statistics, if not completely mirrored. But as we have seen before, the maps have shown relatively stable matchup statistics aside from the natural race imbalance. So what can the reason be? Let’s delve deeper into why Fighting Spirit shows these imbalances.
Zerg and Terran Win-rates on Fighting Spirit
vT
vZ
EffOrt
6-2 (75%)
sSak
14-8 (63.6%)
ZerO
9-7 (56.3%)
Last
2-1 (66.7%)
Killer
16-17 (48.5%)
Sea
21-9 (70.0%)
herO
14-20 (41.2%)
Mong
29-13 (69.1%)
Larva
7-20 (25.9%)
PianO
8-1 (88.9%)
Total
52-66 (44.1%)
Mind
10-1 (90.9%)
HiyA
28-11 (71.8%)
Total
112-44 (71.8%)
Using the 7 Terran and 5 Zerg I previously listed, I took the data from the TvZ matchups on Fighting Spirit and this is what I came up with. Astonishingly, Terran have an almost 72% win rate, numbers so high that they cannot be fathomable in Starcraft terms. Zerg players on the other hand, have a dismal 44% win rate. However, skeptics of these statistics may say I picked 7 Terran who are notoriously good at TvZ, and 5 Zerg who are notoriously bad at ZvT. But did I? You might be surprised at the data.
ZerO, hero, and Kwanro all have about 45% vT win rates, while Killer and EffOrt have above 50%. In total, their combined win rate is 209-245, or a bit above 46%. So they see a -2% drop in win rate on Fighting Spirit, but this isn’t very conducive of firm evidence. But the Terran data was extremely surprising and caught me off guard. The Terran group only has a win rate of 256-214 combines, or 54.5%. Their Fighting Spirit win rate is a +17% increase compared to their average, which means if you remove Fighting Spirit from the map pool, Terran actually only have a 144-170 win rate, or 45.9%, against Zerg. This translates to nearly +26% in win rate between Fighting Spirit and other maps, representing a dramatic 44% difference. This measure, by itself, reveals to us the underlying reason why Fighting Spirit has such an obscene win rate. Terran have simply figured out Zerg on this map, while Zerg players have been unable to cope with it.
Conclusion
Ultimately, Fighting Spirit’s current massive Terran favored disparity can be answered through one simple answer, Terran players perform and prepare so much for this map compared to any other map in the current map pool. When there is a 44% difference in win rates for the top Terran players between FS and other maps, it means three possible things.
That Terran players just prepare so much better than Zerg players for this specific map.
That there are features that inherently favor Terran in the TvZ matchup. However, there is no evidence to this measure, since Terran players go basically even with Zerg on every other similar map as Fighting Spirit, leaving no conclusive data to support this theory.
The other maps in the map pool are imbalanced in favor of Zerg, which is why these Terran players have a hard time dealing with them. This claim is also not well supported, since once again, there is little proof of map imbalances with the rest of the map pool.
One thing I look forward to in the coming SSL and other tournaments that remove Fighting Spirit, will Terran players still reign supreme over Zerg? Considering their 45.9% win rate on every other map, Terran seem to still be confused with the rest of the map pool. Will they adapt quickly enough, or will Terran finally be brought down to normal levels? All of this will be revealed to us at the conclusion of SSL 11.
Thanks, FlashFTW. Definitely some food for thought
Note to all our readers, if anyone has an interesting idea or topic s/he'd like to write about or analyse, do contact TL BW staff. We'd love to help in the presentation, content, or just language editing, to give it the best platform possible. Do let us know and we'd be glad to help make it happen!
If we look at this much more prominent stat line, you'll notice every map is skewed towards the standard imbalance. So while no map distinctively favors one race, they are all imbalanced towards the same matchups.
Interesting point! If such imbalance is naturally expected to a degree, then it implies there's a breaking point at which the map exceeds that acceptability threshold...
There's no especially good reason to think that balance is equal at all levels of skill. Maybe this is what balance looks like a notch under the highly structured environment of progaming teams.
StarCraft has always been imbalanced. terran has always been the underpowered one among protoss and Terran. The problem is no one knows how to play Protoss or Zerg which is why there's so much imbalance in Terran's favor. Bisu is the closest one to using Protoss correctly and has a similar playstyle to mine.
I personally think that Terran is the 'strongest' race in Broodwar if the player can take advantage of all its cost-effective units. (Think widow mine with vultures, science vessel, marine and medic etc) However, you need to be able to micro well and be able to multi-task well in order to take advantage of the race. When you watch Flash play Terran at the peak of his game, it just seems so impossible to break the terran defense.
map and matchup statistics saved from iccup from around 2009, by sas.Pez this was somewhere mid-season, but with close to 400k games, this is the largest sample that i know.
I counted the most played maps' statistics for you:
The overall you can see on the photo: TvZ: 50%; ZvP: 51,4%, PvT: 52,9%
Iccup shows this stat for FS right now (i don't know if it's working or not):
and this for Python:
I just posted this here to have stats for the lower levels too. It would be interesting to see how the stats change with rank, but we don't have data about that. Or at least i don't have. If any of you have any kind of large match up balance stats, send me please.
Thanks for all this! I always had un-researched suspicions as to imbalances in Fighting Spirit, so I'm glad to see this discussed more in-depth. In the realm of pure opinion, I can definitely say that I've grown bored of FS as its era's one map.
The fact that you left out KAL out in the Protoss Heroes section while including Horong2 in your top 10, and the fact that you even mentioned the six dragons, is utterly inexcusable
Map gets older, defense gets better, Terran timings get figured out. FS is probably the most exhausted map in history so if Bisu speaks the truth there's no map more likely to feel the Terran-creeps-ahead effect that he describes.
What's more, there can be little doubt that the overall mechanical skill level has dropped and the fishlords that get into these broadcasted events are ultimately a tiny group of amateurs with only roughly-equal skill levels, meaning 1) it's more likely for all the good players to randomly be one race and 2) the aforementioned passive, unaffected-by-mechanics advantage that Terran builds on old and overplayed maps is even more significant.
On July 28 2015 12:25 EngrishTeacher wrote: The fact that you left out KAL out in the Protoss Heroes section while including Horong2 in your top 10, and the fact that you even mentioned the six dragons, is utterly inexcusable
Nice write-up otherwise, enjoy reading thanks.
i have no excuse for this. i knew i felt like i was missing out on stuff. but i included horang2 because he has some good proleague moments. i consider him like a kt.violet player. good in proleague, no real results in individual leagues.
I want to thank everyone for supporting me in this long journey for this article. It means a lot to me how much people have helped me and talked to me about this beautiful game. Ever since I joined this site, I wanted to become a writer here. And 5 years later, it's finally happened. Thank you. Seriously.
Everyone here makes excellent points in addition but overall, I hope this article has been written well for all of you, it being my first real article on this site besides the preview articles I've written up. Ultimately, this topic kind of narrowed down to a TvZ debate, which I hope I cover in future articles so stay tuned
Shoutouts to prech, BR, BD, koget, and 2pac for all the help and advice you've all given me.
I enjoyed seeing the breakdown and the thoughts, but am not sold on the conclusion you reached at the end.
It seems unlikely that terran would prepare dramatically better on FS than zerg would. I can't think of a reason of the top of my head that would explain why terrans prepare better than zergs on FS, and can think of several arguments against that conclusion.
I guess to me the gist of the conclusion is that if it's not A or B, then C is the only leftover possibility. To me, there are many more options than just C, and it does not follow that if A and B are false, then C must be true.
On July 28 2015 13:26 L_Master wrote: I enjoyed seeing the breakdown and the thoughts, but am not sold on the conclusion you reached at the end.
It seems unlikely that terran would prepare dramatically better on FS than zerg would. I can't think of a reason of the top of my head that would explain why terrans prepare better than zergs on FS, and can think of several arguments against that conclusion.
I guess to me the gist of the conclusion is that if it's not A or B, then C is the only leftover possibility. To me, there are many more options than just C, and it does not follow that if A and B are false, then C must be true.
No I agree with that statement. TBH I don't know the reason for it. It's very... unknown to me still. I said it's a MAYBE, because I mean, Terrans play normal on every other map. And FS isn't super TvZ favored. :/
am I just blind or are spawn locations mentioned almost nowhere? I think it's pretty important to keep that in mind with rotational symmetry maps like FS.
On July 28 2015 15:26 Meavis wrote: am I just blind or are spawn locations mentioned almost nowhere? I think it's pretty important to keep that in mind with rotational symmetry maps like FS.
I don't see your point with this. Zergs might spawn advantageously, and Terrans might spawn advantageously. There's a 33% Terran spawn better, 33% Zerg spawn better, and 33% cross map. So I'm not sure what the point here is. Could you please elaborate?
I liked the article overall. However, you should probably point out more that you could only observe professionals. See Zimp's question below. Hence, the conclusion with the preparation part seems to make sense to me, even though the article couldn't possibly deliver a proof. Thumbs up either way.
Only small mistake I think I found was about rankings: You claim to have had a look at Elo numbers. I guess you took that from TLPD? TLPD doesn't feature Elo, but Glicko2 or at least some other form of ordinary Elo. Not that it matters much, in the circumstances it's more or less the same.
Also, -10 points for not writing ELO to summon raging itsjustatank :D
On July 28 2015 11:29 zimp wrote: I counted the most played maps' statistics for you:
The overall you can see on the photo: TvZ: 50%; ZvP: 51,4%, PvT: 52,9%
and this for Python:
I just posted this here to have stats for the lower levels too. It would be interesting to see how the stats change with rank, but we don't have data about that. Or at least i don't have. If any of you have any kind of large match up balance stats, send me please.
First part: I guess Destination, HBR and Python seem to be more skewed the more mobile race. Has not much to do with Proxy Gates, but rather the fact that it is a lot easier to simply contain a Protoss / Terran in the bases for longer. He moves out, you move in, meanwhile you take the map. Zerg is faster than both races, Protoss faster than Terran - and that's how you get to these numbers. The same effect doesn't work as crucial on maps like FS, where you can at least get the third rather easy and get 25% control of the map. Just a theory though, since I think Proxy Gates, BBS ans 5/6 Pools will neutralize each other in large number samples, regardless of the map (logic of the builds is the same).
For the second part: You could try to use the data generated by the D Rank Teamleague and the C Rank tournaments over the past three years. The sets won't be nearly as big, but good enough, I assume. Additionally, you could try to go through ICCup ladder tournaments and add equally paired players in those. However, the ICCup data base won't really allow you to filter stats, as sad as it is.
A drop from 72% to 46% is not 26% difference. It's 26 percentage points or 26pp difference. It is - however - also a 36% difference or a drop by 36%. Please update the article accordingly. :/
Otherwise - cool read but kind of predictable conclusion.
On July 28 2015 15:26 Meavis wrote: am I just blind or are spawn locations mentioned almost nowhere? I think it's pretty important to keep that in mind with rotational symmetry maps like FS.
I don't see your point with this. Zergs might spawn advantageously, and Terrans might spawn advantageously. There's a 33% Terran spawn better, 33% Zerg spawn better, and 33% cross map. So I'm not sure what the point here is. Could you please elaborate?
I see his point. spawn positions obviously have an effect winrates. its quite possible that in your sample group of games for a particular matchup that the majority of the games sampled might be, say, cross-spawn positions for example. this will obviously effect the win-loss ratio for matchup, since generally cross-spawns will favour a particular race and close-spawns another
I believe in what Bisu said. Stronger defense inherently increases the margin for error and at the level of the pros, that margin however small can be huge deciding factors in a game.
For the sake of removing fighting spirit from tournament map pool, I'll say yes, it's a fact. ^_^ Great post FlashFTW
While I like the write-up I disagree with the assumption that Terran "figured" out Zerg on FS. I would assume level of play is actually lower than it was a couple years back and if the current Terrans faced the Zergs from 2010 they wouldn't perform as well.
On July 28 2015 22:26 thezanursic wrote: While I like the write-up I disagree with the assumption that Terran "figured" out Zerg on FS. I would assume level of play is actually lower than it was a couple years back and if the current Terrans faced the Zergs from 2010 they wouldn't perform as well.
That's reasonable. Lower level play in most games tends to exacerbate balance issues. You can even see this in a game like chess, for example.
On July 28 2015 23:39 JieXian wrote: Maybe the newer strategies are more effective on FS. Was a late game mech switch that popular in the Kespa era?
FS is very good for a late game mech switch.
Mech switches were beginning to take rise in the late KeSPA era. But late game turtle with mass tanks was always in the meta (see the FlaSh vs Calm gif)
On July 28 2015 18:32 nimdil wrote: OK this is annoying.
A drop from 72% to 46% is not 26% difference. It's 26 percentage points or 26pp difference. It is - however - also a 36% difference or a drop by 36%. Please update the article accordingly. :/
Otherwise - cool read but kind of predictable conclusion.
Thanks, fixed the language to reflect this. +26% win rate and using the average as the divisor 25.9/((45.9+71.8)/2), it represents a 44% difference.
When did Fantasy actually invent the now popular Late Mech Switch and how long did it take to get popular? Compare that to the KesPA life-span of Fighting Spirit. Did Fighting Spirit's cycle end just as late mech became popular?
On July 29 2015 05:09 HyralGambit wrote: @FlashFTW Great article. Question:
When did Fantasy actually invent the now popular Late Mech Switch and how long did it take to get popular? Compare that to the KesPA life-span of Fighting Spirit. Did Fighting Spirit's cycle end just as late mech became popular?
Maybe that explains it.
To be completely honest, I don't know exactly when the mech switch began becoming popular. I think it began mid-late 2010.
Maybe the TvZ meta shifted to favor Terran? Maybe? But at the same time, I think if the meta did shift in favor of Terran, wouldn't we see similar results on every other map like La Mancha, Sniper Ridge, Neo Jade?
Amazing article mate. Id like to drop my 2 cents on this topic, apart from what the article already mentioned.... well a bit more than that:
-FS is a easy map to understand in terms of vectors.
Historically, zerg was always the race that benefited the most from multi angled atacks and flanks. Every map was easy enought to understand, but the atacks could come from at a lot of different locations, due to the map layouts (What I means is that you could hit the enemy in many adjacent areaes and still flank them rather well, if the army position was correct).
This made it quite hard for both terran and Protoss to anticipate when exactly the pincer would try to close around you, requireing more expirience in each of these situations to know where to evacuate if it was required. How to allocate your army for maximum efficiency and most importantly, how to set your army's layers.
If you take a close look at FS, very few areas allow for well coordinated pincers, and in all cases, the exit route is extremely easy to spot. This does not mean that it is easier to do it correctly, just that it is harder to screw up.
Terrans have always benefited from micro intensive battles due to their mid term stand in between zerg and protoss (more life than zerg, more dps than Protoss, less life than protoss, less dps than zerg... huge generalization, I know).
So, everytime you want to do something (for example establish a 4th) its rather clear how you want to do that in FS. If you want to push a Protoss base with mech, its rather clear where to set mines to cut off reinforcements or main armies. *What I mean in terms of vectors is that if A enemy, and his army goes X side, my army B should go this other side, etc.
-FS is a map that has been played enought for terrans to develop very clutch timings.
We all know zerg vs terrans tries to delay making sunkens for the first push as long as possible. We all know that Zergs try to use mutas to delay pushes. We all know that zergs almost MUST make lurkers to defend safely and eco-efficient the 3rd base.
Pros know it even better, and a simple things like 1 overlord kill or 1 vulture that kills 2 gas drones can be enought to block the zergs natural flow and respose to a normal game.
The main thing here is that the terrans can think about a specific way they want to get that 1 small win to cripple the zerg enought for their timing and practice it for a while. The zergs cannot themselves imagine all possible variables, and mentaly prepare for all of them, and then practice all of them. (This is also why unorthodox plays like zeros 2 base hive work so well sometimes, its something terrans dont have expirience with).
-FS is a rather small map.
Back in the Blood Bath map days, zergs were destroying everyone due to map size. FS gives rather short time, unless cross spawns, of warning before the enemy is at your door. Zerg has clearly a problem here due to larva mechanics. There is a lot more to it, like the variablility of early builds terrans can do, adding more complexity, but id say thats a rather big factor.
-Terrans VS Zerg is a VERY imbalanced match up.
This might surprise most of you.... but I always thought zerg had the shorter end of the stick here, massivly. Players have been able to "balance" the game due to their builds and skill, not due to patches (plz no sc2 comments here :D). For those of you who dont know it, pro korean scene had a sort of cycle when a certain race dominated untill it was dethroned. Boxer, ILoveoov, Saviour, Bisu, Nada, Flash, Jangbi..... Those were all players who destroyed the previous imbalances by beating players with things they had no practice against, because only they knew how their build worked.
Same happens now. I mentioned Zero before. Only he knows how to make the 2 base hive build, so it will be very hard to practice against it with your ally to get ready for a match vs Zero.
We do not have pro broodwar teams anymore, so even if a Zerg figures out a new amazing build.... who does he test it with? He cant do it with the player he will play against.... and its likely that his friends wont be as high of a caliber as his enemy in a SSL.... so tehre is no way to confirm if a new build would be good or not... This leads to rather known builds, and in this current state of effects, Terrans is in my opinion stronger than zerg, zerg slightly stronger than Protoss, and terran and protoss are about even. (My opinion of course).
Finally, Id like to point out one thing i disagree with the author:
Free is in my opinion a players who is best at PvT, with quite crappy PvP and terrible PvZ. He is one of my favorite players, and I always loved his PvTs, but almost watched praying when he was against a Z or a P.
On July 28 2015 18:32 nimdil wrote: OK this is annoying.
A drop from 72% to 46% is not 26% difference. It's 26 percentage points or 26pp difference. It is - however - also a 36% difference or a drop by 36%. Please update the article accordingly. :/
Otherwise - cool read but kind of predictable conclusion.
If you throw that few units into a ton of siege tanks, you deserve to lose. But yeah, the map is really turtley. I've said this for years. You can play super-greedy and get away with it (1-fact double expo; what?) because each of the four quadrants are very efficient to defend because they funnel units into a very small area.
On July 28 2015 11:18 Ty2 wrote: StarCraft has always been imbalanced. terran has always been the underpowered one among protoss and Terran. The problem is no one knows how to play Protoss or Zerg which is why there's so much imbalance in Terran's favor. Bisu is the closest one to using Protoss correctly and has a similar playstyle to mine.
Sometimes, I just want to slap you.
On July 29 2015 06:16 Bill Murray wrote: lost temple is still cooler than fighting spirit
On July 28 2015 18:32 nimdil wrote: OK this is annoying.
A drop from 72% to 46% is not 26% difference. It's 26 percentage points or 26pp difference. It is - however - also a 36% difference or a drop by 36%. Please update the article accordingly. :/
Otherwise - cool read but kind of predictable conclusion.
Thanks, fixed the language to reflect this. +26% win rate and using the average as the divisor 25.9/((45.9+71.8)/2), it represents a 44% difference.
OK why the average? You use - as divisor - the number from which you start the comparison.
So i.e. drop from 100 to 80 is 20% drop while increase from 80 to 100 is 25% increase. 22.5% or 20/((100+80)/2) has no place here for anything, really. Unless english has some obscure rule about it ... ?
NEW statistics. Love it!!! it would be great to see the statistics, if they are even recorded, of all the iccup games ever played on fs. Thanks for the article
A perfect map and perfect balanced game can and will never exist. If it did, the game would break down. Its the same concept as perpetual motion.
On July 29 2015 05:38 FlaShFTW wrote: To be completely honest, I don't know exactly when the mech switch began becoming popular. I think it began mid-late 2010.
Maybe the TvZ meta shifted to favor Terran? Maybe? But at the same time, I think if the meta did shift in favor of Terran, wouldn't we see similar results on every other map like La Mancha, Sniper Ridge, Neo Jade?
Isn't the 3rd base slightly harder to secure/defend on all three of those maps compared to Fighting Spirit? They seem harder to access and reinforce from the natural.
I admit most of my games are played on FS, so these are just observations from pro games. It's interesting to think about, great article.
What struck me about your initial stat-binge was the contrast between Fighting Spirit and two of the other "standard" maps you mentioned: Icarus and La Mancha. While you described Icarus as "favoring Zerg", it would be equally accurate to say it is "anti-Protoss", based on the KeSPA numbers. Which perhaps explains why it's fallen out of favor post-KeSPA. La Mancha, meanwhile, displays "normal" imbalance both KeSPA and post-KeSPA (though Protoss appear weaker).
My initial instinct on examining the maps is to say the design of the third bases is the critical factor in otherwise fairly similar and standard maps.
However I want to first touch on a couple other differences. First, the center designs are strikingly different. Fighting Spirit's center is flat but broken-up. Icarus' middle is essentially unbroken though slightly narrower. La Mancha has impassable terrain in the very middle, but the circular center pattern resulting is very wide all the way around. The ridges (like on HBR, but more so) are important in the early and mid-game "contain" stage but (relatively speaking!) lose importance in the face of late-game army sizes. The openness of both these maps would seem to favor Zerg on a first guess in ways Fighting Spirit's layout doesn't.
The thirds, however, are even more different. On Icarus, the third is "safe" once established but hard to get to, allowing a Zerg to contain the Terran much more easily. On La Mancha, the third is on low ground (removing one defender's advantage) and the access from the natural is much wider (making runbys easier).
My tentative hypothesis is that late-game mech (or at least a heavy tank count) has become much more relatively common in post-KeSPA Korean play, which has resulted in Fighting Spirit's relatively easily defended terrain becoming more Terran-friendly.
It's not definitive, of course. But it's the naturally-occurring explanation, and certainly warrants further exploration. Though perhaps watching hundreds of games and comparing builds is... time-consuming. Maybe we need to start adding build catalogs to games, analogous to classifying chess games by opening?
Wouldn't hate it if BW had one more balance patch to address how the meta has changed since 2001, much less the post-Kespa era... after all SC2 has had something like 22 balance-patches (literally).
But of course, there's no real incentive for Bliz to do that, not unless they re-released the game ('Bnet 2.0 Edition'?).
And even if they did, in the state the company is in now... would they just screw it up?
The map itself did not change during the transition so statistics should remain the same.
What? This is of course false.
Some may suggest that with more time, the pros are able to abuse every small advantage they could, and Terran players ultimately came out on top. However, I am very skeptical. This argument suggests that both players and coaches could not solve the map in its reign of over 5 years, which seems extremely doubtful, even laughable.
What? Again, this is of course false: The implication given by saying that Terrans eventually came out on top, IS the claim that the coaches and players will solve a map over the course of five years. Claiming terran is better on fs may or may not be true, but it clearly isnt the antithesis of saying players will figure out maps.
As for the rest, to summarize neatly: Map balance changes because the meta game changes, and I dont think map making is reducible simply to looking at statistics of maps with similar features, as it is the complex interplay of these features that actually create balance. And, as the metagame changes, a map will become more or less favourable over time. In the current meta game t > z, z > p, in a more substantial way than at the very end of the kespa era. Basically every map has become harder for protoss to beat zergs, and zergs to beat terrans. And these maps could remain imbalanced for the next year, two, or indefinitely. The meta game may simply evolve to necessitate features these maps are lacking...
Adding on to what I said already, there's nothing really to learn from this well-written summarized stat-crunching except that we can finally be sure FS is overplayed to the point that its competitive exhaustion is - alongside other factors unrelated to its design - reducing the quality of games played on it.
On July 29 2015 08:48 VGhost wrote: My tentative hypothesis is that late-game mech (or at least a heavy tank count) has become much more relatively common in post-KeSPA Korean play, which has resulted in Fighting Spirit's relatively easily defended terrain becoming more Terran-friendly.
It's not definitive, of course. But it's the naturally-occurring explanation, and certainly warrants further exploration. Though perhaps watching hundreds of games and comparing builds is... time-consuming. Maybe we need to start adding build catalogs to games, analogous to classifying chess games by opening?
Such more advanced game information/data would be definitely be very useful for further analysis. It'd be difficult to collect, even if logging just strategies/tactics used, as multiple approaches and variations of those approaches are used all the time, but it'd certainly enable a more sabermetric-ish data dive into what tactics are best suited for each map/match-up/etc rather than relying on, or in addition to, heuristic evaluations
Reminds me of this story in the NY Times about how League of Legends enable players/users to access intricate game data
On July 29 2015 10:41 Mirabel_ wrote: Adding on to what I said already, there's nothing really to learn from this well-written summarized stat-crunching except that we can finally be sure FS is overplayed to the point that its competitive exhaustion is - alongside other factors unrelated to its design - reducing the quality of games played on it.
True that. Just compare the average TvZ on Fighting Spirit to the dynamic TvZ on Circuit Breaker.
So happy Circuit Breaker is getting more popular these days.
On July 29 2015 10:41 Mirabel_ wrote: Adding on to what I said already, there's nothing really to learn from this well-written summarized stat-crunching except that we can finally be sure FS is overplayed to the point that its competitive exhaustion is - alongside other factors unrelated to its design - reducing the quality of games played on it.
That does feel like the most simple, unassuming conclusion.
On July 29 2015 10:41 Mirabel_ wrote: Adding on to what I said already, there's nothing really to learn from this well-written summarized stat-crunching except that we can finally be sure FS is overplayed to the point that its competitive exhaustion is - alongside other factors unrelated to its design - reducing the quality of games played on it.
True that. Just compare the average TvZ on Fighting Spirit to the dynamic TvZ on Circuit Breaker.
So happy Circuit Breaker is getting more popular these days.
On July 29 2015 10:41 Mirabel_ wrote: Adding on to what I said already, there's nothing really to learn from this well-written summarized stat-crunching except that we can finally be sure FS is overplayed to the point that its competitive exhaustion is - alongside other factors unrelated to its design - reducing the quality of games played on it.
True that. Just compare the average TvZ on Fighting Spirit to the dynamic TvZ on Circuit Breaker.
So happy Circuit Breaker is getting more popular these days.
Circuit is even worse for zerg than FS is.
I did some research on this. Since the start of April this year, TvZ is quite balance with zerg actually ahead by a game.
On July 29 2015 10:41 Mirabel_ wrote: Adding on to what I said already, there's nothing really to learn from this well-written summarized stat-crunching except that we can finally be sure FS is overplayed to the point that its competitive exhaustion is - alongside other factors unrelated to its design - reducing the quality of games played on it.
True that. Just compare the average TvZ on Fighting Spirit to the dynamic TvZ on Circuit Breaker.
So happy Circuit Breaker is getting more popular these days.
Circuit is even worse for zerg than FS is.
I did some research on this. Since the start of April this year, TvZ is quite balance with zerg actually ahead by a game.
On July 29 2015 10:41 Mirabel_ wrote: Adding on to what I said already, there's nothing really to learn from this well-written summarized stat-crunching except that we can finally be sure FS is overplayed to the point that its competitive exhaustion is - alongside other factors unrelated to its design - reducing the quality of games played on it.
True that. Just compare the average TvZ on Fighting Spirit to the dynamic TvZ on Circuit Breaker.
So happy Circuit Breaker is getting more popular these days.
Circuit is even worse for zerg than FS is.
I did some research on this. Since the start of April this year, TvZ is quite balance with zerg actually ahead by a game.
But Circuit Breaker seems to favor mutas and guardian play. Granted the double bridges are better for Terran defensively, but mutas control them well and pre-defiler timings are less deadly.
The mineral only third seems more exposed and areas to drop (Zerg drops) are numerous as well.
On July 29 2015 10:41 Mirabel_ wrote: Adding on to what I said already, there's nothing really to learn from this well-written summarized stat-crunching except that we can finally be sure FS is overplayed to the point that its competitive exhaustion is - alongside other factors unrelated to its design - reducing the quality of games played on it.
True that. Just compare the average TvZ on Fighting Spirit to the dynamic TvZ on Circuit Breaker.
So happy Circuit Breaker is getting more popular these days.
Circuit is even worse for zerg than FS is.
I did some research on this. Since the start of April this year, TvZ is quite balance with zerg actually ahead by a game.
I think, by using Artosis logic, wouldn't every map at the time be Terran favored? And if Terrans really are supreme right now, wouldn't that mean every map should be Terran favored in a TvZ? But as I pointed that out, it's simply not the case.
On July 28 2015 23:39 JieXian wrote: Maybe the newer strategies are more effective on FS. Was a late game mech switch that popular in the Kespa era?
FS is very good for a late game mech switch.
Mech switches were beginning to take rise in the late KeSPA era. But late game turtle with mass tanks was always in the meta (see the FlaSh vs Calm gif)
Yes but was it done almost every game and so early like in post-kespa?
I think an analysis is needed, though it'll take a lot of effort.
On July 29 2015 10:41 Mirabel_ wrote: Adding on to what I said already, there's nothing really to learn from this well-written summarized stat-crunching except that we can finally be sure FS is overplayed to the point that its competitive exhaustion is - alongside other factors unrelated to its design - reducing the quality of games played on it.
True that. Just compare the average TvZ on Fighting Spirit to the dynamic TvZ on Circuit Breaker.
So happy Circuit Breaker is getting more popular these days.
Circuit is even worse for zerg than FS is.
I did some research on this. Since the start of April this year, TvZ is quite balance with zerg actually ahead by a game.
About that giant FS screenshot wallpaper, I bumped the original thread, but it doesn't show up cuz it's in blogs (and I suspect most people don't browse blogs through the sub-forum menu), so, can anyone deliver, please?
On July 30 2015 02:40 bITt.mAN wrote: About that giant FS screenshot wallpaper, I bumped the original thread, but it doesn't show up cuz it's in blogs (and I suspect most people don't browse blogs through the sub-forum menu), so, can anyone deliver, please?
I'll deliver give me a few days to dig up the original! (I'm still welcoming recommendations for good hosts for truly HD pictures like this, I'm afraid imgur still wouldn't take it verbatim)
Great write-up, and thanks for the image attribution! It was quite a surprise to see it on the frontpage after all these years.
On July 30 2015 02:40 bITt.mAN wrote: About that giant FS screenshot wallpaper, I bumped the original thread, but it doesn't show up cuz it's in blogs (and I suspect most people don't browse blogs through the sub-forum menu), so, can anyone deliver, please?
I'll deliver give me a few days to dig up the original! (I'm still welcoming recommendations for good hosts for truly HD pictures like this, I'm afraid imgur still wouldn't take it verbatim)
Great write-up, and thanks for the image attribution! It was quite a surprise to see it on the frontpage after all these years.
Not sure if there's an even higher resolution version, but here's the 4096px one on Mega
Does anybody know how the ELO system for SC:BW worked? For example, if Flash had a TvZ Elo of 2350 and Zero had a ZvT Elo of 2261, what's the estimated probability that Flash beats Hero? Is it based on Normal distribution? Logistic distribution?
Also, does TL have historical ELO? Like, can I get the ELOs of JD and Flash before they played their game on FS in the Bigfile MSL final?
On July 30 2015 02:40 bITt.mAN wrote: About that giant FS screenshot wallpaper, I bumped the original thread, but it doesn't show up cuz it's in blogs (and I suspect most people don't browse blogs through the sub-forum menu), so, can anyone deliver, please?
I'll deliver give me a few days to dig up the original! (I'm still welcoming recommendations for good hosts for truly HD pictures like this, I'm afraid imgur still wouldn't take it verbatim)
Great write-up, and thanks for the image attribution! It was quite a surprise to see it on the frontpage after all these years.
Not sure if there's an even higher resolution version, but here's the 4096px one on Mega
On July 31 2015 10:06 ghrur wrote: Does anybody know how the ELO system for SC:BW worked? For example, if Flash had a TvZ Elo of 2350 and Zero had a ZvT Elo of 2261, what's the estimated probability that Flash beats Hero? Is it based on Normal distribution? Logistic distribution?
Also, does TL have historical ELO? Like, can I get the ELOs of JD and Flash before they played their game on FS in the Bigfile MSL final?
if you get the date of the series, you could click on the past elo ranking and find the date/month that you want. then just update it to the next day after the finals was played and you'll have the updated elo rank i think? I'm not entirely sure.
Also nice to know this thread has been highjacked by Fighting Spirit artwork. I find nothing wrong with that ^^ Carry on. No but seriously that art is beautiful I want a poster of it in my room.
I don't know if the definition of imbalance is the same for progamers as for us. I find zerg to be slightly disadvantaged on FS because it is very easy to proxy them. If you take a look on current zerg, they open overpool so many times just because of extreme abuse of rush bunker, and that makes the task much more difficult. Also zerg tends to have a defensive gameplay in the opening at least until they secure a third. FS path are short it is quick to get around which give little time to anticipate. It heavily helps for proxy strat and timing push. Not saying we zerg don't do timing push but I believe protoss and terran are more fit to that type of gameplay. If you compare to other map like empire of the sun or la mancha... I think i don't need to explain myself more. But again it is just from my perspective and experience of the game.
Awesome article. It feels weird when we talk about such intricate balance things because there's just so many reasons why it can get skewed.
I personally feel that Terran players attract the most hardest working and passionate pro players, and that's why there's such a skew in the Post-Kespa era.
On August 01 2015 03:02 lestye wrote: Awesome article. It feels weird when we talk about such intricate balance things because there's just so many reasons why it can get skewed.
I personally feel that Terran players attract the most hardest working and passionate pro players, and that's why there's such a skew in the Post-Kespa era.
What the hell man.. That's such an insult to other dedicated pro-gamers, I can't even believe it.
On August 01 2015 03:02 lestye wrote: Awesome article. It feels weird when we talk about such intricate balance things because there's just so many reasons why it can get skewed.
I personally feel that Terran players attract the most hardest working and passionate pro players, and that's why there's such a skew in the Post-Kespa era.
What the hell man.. That's such an insult to other dedicated pro-gamers, I can't even believe it.
+1. I mean, even limiting it to just post-Kespa era, what... Bisu doesn't work hard? Hero doesn't work hard? They and all the other pro P and Z players aren't 'passionate'? wtf?
On July 31 2015 10:06 ghrur wrote: Does anybody know how the ELO system for SC:BW worked? For example, if Flash had a TvZ Elo of 2350 and Zero had a ZvT Elo of 2261, what's the estimated probability that Flash beats Hero? Is it based on Normal distribution? Logistic distribution?
Also, does TL have historical ELO? Like, can I get the ELOs of JD and Flash before they played their game on FS in the Bigfile MSL final?
There's some good info on the TL Elo rating on Liquipedia. Basically, as far as I understand it, it's not weighted or has any statistical adjustment done on it. Players earn and lose Elo points based on the strength of his/her opponent (who similarly gains and loses the same amount of points), so there's no weighting done by strength of competition or how "major" a tournament is...
There are some reference threads listed on the page, one of which discusses a little bit of Elo inflation over time, as there are basically more Elo points to go around as more higher caliber players emerged as the progamer scene developed.
As for Flash and ZerO, they have a ~90 point difference--
A player whose rating is 100 points greater than his/her opponent's rating is expected to win 64% of the time.
Alas, the historical Elo isn't spit out/visible on TLPD, but the data's there for calculations at any time. There are and have been plans to enable this historical view, much like Aligulac has done with SC2, as BW has such a rich history of ~15 years of progaming that can be analysed...
On August 01 2015 03:02 lestye wrote: Awesome article. It feels weird when we talk about such intricate balance things because there's just so many reasons why it can get skewed.
I personally feel that Terran players attract the most hardest working and passionate pro players, and that's why there's such a skew in the Post-Kespa era.
What the hell man.. That's such an insult to other dedicated pro-gamers, I can't even believe it.
+1. I mean, even limiting it to just post-Kespa era, what... Bisu doesn't work hard? Hero doesn't work hard? They and all the other pro P and Z players aren't 'passionate'? wtf?
I guess my joke didnt come off right, I was just pretending to be a Terran above all else guy :-P
I wonder how FS and CB are both balanced when at 1st map center/middle is buildable but in 2nd map is not which will prevent terran to secure certain area with the help of depots+turrets.
On July 31 2015 10:06 ghrur wrote: Does anybody know how the ELO system for SC:BW worked? For example, if Flash had a TvZ Elo of 2350 and Zero had a ZvT Elo of 2261, what's the estimated probability that Flash beats Hero? Is it based on Normal distribution? Logistic distribution?
Also, does TL have historical ELO? Like, can I get the ELOs of JD and Flash before they played their game on FS in the Bigfile MSL final?
There's some good info on the TL Elo rating on Liquipedia. Basically, as far as I understand it, it's not weighted or has any statistical adjustment done on it. Players earn and lose Elo points based on the strength of his/her opponent (who similarly gains and loses the same amount of points), so there's no weighting done by strength of competition or how "major" a tournament is...
There are some reference threads listed on the page, one of which discusses a little bit of Elo inflation over time, as there are basically more Elo points to go around as more higher caliber players emerged as the progamer scene developed.
As for Flash and ZerO, they have a ~90 point difference--
A player whose rating is 100 points greater than his/her opponent's rating is expected to win 64% of the time.
Alas, the historical Elo isn't spit out/visible on TLPD, but the data's there for calculations at any time. There are and have been plans to enable this historical view, much like Aligulac has done with SC2, as BW has such a rich history of ~15 years of progaming that can be analysed...
Not only that, but theres more points to go around as there was a gradual inflation in the amount of games played every year. You can find a top pro from 2000 who played like, eight games the whole year.