Racial Balance in GSL (2010-2013 WoL)
As the only person(?) in the world who watched this unpopular ZvZ GSL final, let me inform all of you that, a few minutes ago,
+ Show Spoiler [GSL final spoiler] +
has become the new GSL champion. Now that the last WoL GSL is over, this must be the best time to discuss the history of racial balance in GSL. Starcraft is an asymmetric RTS with 3 distinct races. As such, balance has always been a big concern. It is ideal for every race to have exactly 50% winrate against each other, but it rarely happens. Therefore, balance discussion has never ended since Day 1. However, illogical “X race OP” arguments based solely on a few games, personal anecdotes, preconception or hatred are rampant in this community. For more legitimate discussion, statistically meaningful enough number of games need to be examined to see if X race is actually OP and by how much if so.
While GSL stats shouldn’t be the only source of balance discussion, it should be weighed heavily because GSL is THE league where the absolute best players compete, and everyone else should aspire to be like them. No one would care even if Bronze league were 100% Zerg simply because they lack skill to defend 6 pool. (just if) Similarly, results of random guys in master league who can’t play as good as GSL competitors don’t matter as much when it comes to balance. They all just have to improve. GSL, the best league on the planet, is where the balance is at. Let’s take a look at numbers.
Table & Graph
+ Show Spoiler +
Note & Disclaimer
+ Show Spoiler +
1. Balance whine at your own risk.
Not that many people care about WoL balance at this point, but there seems to be always a few people who can’t help posting aggressive hate comments, knowing an incoming ban. If you want a ban so desperately, you can always request at The Automated Ban List
2. I am more than fully aware that the data is inevitably skewed. For example,
- Maybe Open S1 Ro64 scrub vs another scrub match shouldn’t have the same weight as Code S Finals.
- There have been several format changes. (2011 tournaments are shorter with fewer games played but held more frequently etc.)
- Sample size is relatively small.
etc. etc. There must be about another dozens of factors. However, I can’t magically come up with a totally unbiased formula to adjust all these factors, either. 0% flawed comparison is impossible no matter how numbers are taken. Therefore, raw numbers from TLPD are used on “a win is a win” basis. Not a perfect statistics, but it should be good enough to find out the general trend, which is the goal here.
3. There are 18 GSL titles in WoL: 3 from Open Season, 13 from Code S, 1 from World Championship, and 1 from Super Tournament. However, World Championship is excluded in this thread because the format is completely different from the other ones. In a sense, Open Season and Super Tournament are Code S and Code A combined with 64(63) participants instead of normal 32(31). These 4 tournaments count as Code S for convenience. Therefore, 17 tournaments(13 Code A) are in the data above. Also, other GOMTV organized invitationals like Blizzard Cup etc. are excluded.
4. Up-Down matches, wild card, and Code A qualifiers are excluded to keep it simple and to save my time figuring out format details for each season. By GSL, I mean either Code S or Code S+A.
5. Calculation for winrates doesn’t include mirror match-ups. Obvious, but important to note.
6. Colors in tables = colored race has higher than 50% winrate in a match-up, tournament or overall.
7. To make up for the small sample size and fluctuating results, 3-Tournament Average is used to find out the general trend. It is basically a simple moving average used in stock market etc. 3-T Ave. = average winrate of a particular tournament, the previous tournament, and the next tournament. For Open S1 and S1 2013, it is the average of 2 adjacent tournaments because the 3rd tournament isn’t available for calculation.
8. I don’t play much ladder these days, but I am a Zerg player if that matters for anything. I’m not trying to push my “Zerg agenda,” believe it or not. I felt I needed to mention the race I play just in case. Unbiased content is hard to come by these days isn’t it?
+ Show Spoiler +
Facts only without interpretation:
- Terran is the most represented race in Code S. 297T-190Z-179P accumulated total players.
- TvT is the most played mirror match-up in Code S+A. It is played more than 3 times the number of ZvZ and PvP (631 TvT – 209 ZvZ – 208 PvP)
- TvT is the most played mirror match-up both in every Code S and Code S+A except for GSL S1 2013. (for Code S, 14 solo, tied twice)
- ZvP is the least played non-mirror match-up. (1013 TvZ – 655 ZvP – 953 PvT)
- Every race has experienced both the highest and lowest winrate multiple times in Code S+A tournaments.
- Zerg has the highest winrate in Code S. (excl. mirror)
- Terran has the highest winrate in Code A. (excl. mirror)
- Terran has the highest winrate in Code S+A. (excl. mirror)
- Protoss has less than 50% winrate in both ZvP and PvT in Code S+A, hence, less than 50% overall, too.
- Zerg 64.4% in Open S2 is the highest winrate of all time in Code S+A..
- Protoss 32.8% in Aug. is the lowest winrate of all time in Code S+A.
+ Show Spoiler +
One might disagree with my own opinion & interpretation:
Terran is the most played race in 14 out of 17 tournaments. It must be partially thanks to Terran favored balance during 2011, but I wonder how much is because of Korean Terran tradition ever since BoxeR in BW.
GOMTvT is real. When one mirror match-up is played 3 times more than the other two and is the most played mirror in 16 out of 17 tournaments, it is safe to say that GOMTvT prevailed the scene not just for a short period in late 2011 but for almost the entire WoL period. We have to be glad that TvT is arguably the most enjoyable mirror match-up for spectators. Imagine having 3 times more ZvZ or PvP…SC2 e-sports would have died a long time ago.
Winrate graph is very volatile in 2010~early 2011. No one knew how to play yet back then, and balance patches were coming frequently.
Terran had been the OP race for a long time overall. Terran occasionally choked (July for example), but in 3-Tournament Average, it kept having 50%+ winrate from 2010 all the way up to mid-2012. When all is said and done, “Wings of Liberty” seems an appropriate name for this expansion which Terran dominated for so long.
Zerg 3-T Ave. winrate oscillates like a cosine curve, alternating between good times and bad times. The last upward curve is dubbed, “Lings of Liberty.” The graph shows that it’s a real phenomenon. Zerg is OP today. There is no doubt about it. Zerg “Vengeance” came a little too early.
Protoss has been struggling. It is the least represented race with the lowest winrate. 2011 was especially bad year for Protoss. In early-mid 2012, Protoss enjoyed a short heyday, “Blinks of Liberty” if you will . Even so, it was a moderate dominance compared to Terran / Zerg at other times.
Code A is Terran’s playground in PvT. 40.6%(59.4% Terran winrate) is such a disaster for Protoss.
Both Terran and Zerg have over 50% winrate overall in Code S+A. Protoss is the only race that doesn’t. Although it’s still within 45%~55% “balanced” line, Protoss needs some more love. Not that Protoss should be OP in HotS because of this, though.
Since exactly 50% each is nearly impossible, it would have been nice to have rock-paper-scissors relationship near 50% similar to BW’s T>Z>P>T record. SC2 WoL Code S+A ended with Z>T>P & Z>P. Poor Protoss.
So-called “Queen Patch” was so powerful. On one hand, Terran had been favored before the patch in TvZ for more than 1.5 years. On the other hand, Zerg domination after the patch is as big of a problem if not bigger. Terran domination lasted longer, but Zerg domination today is more intense and even at a ridiculous level. Zerg buff itself was a reasonable move by Blizzard as ZvP was also bad for Zerg back then, but it was waaaaay too much. As a result, we have “Lings of Liberty” today. The patch is not the only reason for the current state, but it must be playing a big role considering the timely Zerg rise. Don’t name names, but “patchzerg” exists for sure just like “releaseterran” existed. Having “patch X race” is inevitable when X race is relatively OP.
Having said all these, I would say that WoL SC2 is a game that is well-balanced enough if the entire WoL days are taken as one season. One concern is that Protoss has been having the hardest time, but every race and match-up is at least within 45%~55% line. 50% is the ultimate goal, and one might say that the acceptable line is 47%~53%(achieved) or even 49%~51%, but overall, WoL is not as terrible as one might think in terms of balance. Current balance as of March 2013 itself is not fine, but overall balance in “WoL Season” looks fine enough.
2010 ~ early 2011 = volatile era
mid-2011~ late 2011 = Terran OP “Wings of Liberty” era
early 2012 ~ mid 2012 = Protoss OP “Blinks of Liberty” era
late 2012 ~ 2013 = Zerg OP “Lings of Liberty” era
Ultimate summary of WoL GSL:
5, 10, 20 years from now, people won’t care or remember the difference between 2011 and 2012. Details don’t matter when memory is so distant. Then, what is the best way to describe WoL days in a concise manner? How would future history books talk about WoL GSL? Well, when you look at the big picture, there is no remarkable steady imbalance in WoL. No race has been dominant for the entirety. Therefore, I dare say that people will remember WoL GSL with something consistent throughout:
“so many Terran players and their GOMTvTs.”
+ Show Spoiler +
First of all, special thanks to whoever is updating TLPD and Liquipedia pages. I took all stats from there. I think many people already knew this volatile – Terran – Prorotss – Zerg balance transition. Maybe Protoss one was not as obvious. Anyways, it’s always nice to have numbers to back up the claim. There are many other interesting numbers I didn’t point out in my analysis. Please post your findings and thoughts. I used the word “OP” casually in this thread, but I’m not trying to incite people or anything. Also, I avoided “There are more X race in Ro.8” line because it is not a substantial balance argument. Comparing map-by-map results is how it should be done, and I explained why at my previous thread: Balance Discussion Math(Best of N format analysis). Thank you all for reading. This is my last WoL guide/article. Once HotS is out, there will be no or fewer articles for a while because I want to first play single player campaign and Monobattle on Ladder (1v1). Then, I will focus on updating existing guides/articles with HotS changes. Your feedback is much appreciated as always.
+ Show Spoiler +
+ Show Spoiler [Article etc.] +
Monobattle on Ladder (1v1)
BitByBit Fan Club
A bit on BitByBit
IlIlIlIlIlIl or lIlIlIlIlIlI?
Optimal Creep Spread in Theory
Various Businesses in Starcraft 2
Balance Discussion Math(Best of N format analysis)
Underground Activities in Starcraft 2
Artosis pylon Art
Map Size History & Analysis
Larva disappearing Glitch in 1.5 (not about 20th larva)
+ Show Spoiler [Guide] +