The State of the Swarm Report

That the new mech build is a TvZ cure-all is pure fantasy.

In case you're wondering, the TLPDized fantasy is intended, and is a dumb play on.

Based on an aggregate of all games since the start of BroodWar to some time Last year, which is on a spreadsheet on my other computer, and based on conventional wisdom (Z>P, T>Z, P>T), the background rates of win percentages for the non-mirror matchups for Zergs are: ~60% ZvP, and ~44% ZvT. The ZvT figure is based on an aggregate of 36 entries in the individual leagues of TLPD.

So, compare this to the win percentages for the year 2008 through the games of January 10th, 2009, from all leagues: ZvP ~51.1%; ~ZvT 42%.

Compared to the background rates, this is dramatic. Particularly, the ZvP is startling, but consistent with the arrival of the Golden age of Protoss. The ZvT is interesting, in that the introduction of the mech build (while admittedly late) to the game in 2008, has seemed to manifest itself in a short amount of time in ZvT. Let me examine this further by taking a look at the ZvT win percentages for leagues leading up to the introduction of the mech build, to see if the mech build in fact is in the cause of the discrepancy in the TvZ rate:



* 2008 Proleague R2 - 6-3
* 2008 Minor League - 12-12
* 2008 Proleague - 21-35
* STX Masters - 4-2
* 2008-'09 Proleague (Only games pre-10-17-08) - 5-9
* ODT 2007 (2008 games) - 6-2
* GOM MSL S4 - 12-13
* Bacchus OSL - 6-7
* OSC 2008 - 3-7
* 14th Survivor - 7-13
* Star Invitational - 5-5
* EVER OSL - 6-4
* Arena MSL - 6-15
* WCG Korea - 4-4
* Incruit OSL - 7-11
* 15th Survivor - 5-13
* Club Day MSL - 7-5
* Classic Season 2 - 6-12


This comes out to a total ZvT record for all 2008 games pre-mech build to be: 128-172, or ~42.6%. So, it is clear that 2008 was already a year of decline in ZvT win rates before the introduction of the mech build. Now, let us juxtapose this with the win percentages of games after the mech build. Here are the figures for all games played in TvZ from 10-17-08 (the date of the fantasy v GGPlay series) onward to January 10th:

* 2008-'09 Proleague (post 10-17-08) - 41-55
* Classic Season 2 - 12-16
* 16th Survivor - 8-16
* Batoo OSL - 5-8


The total ZvT record for all games post 10-17-08 is, then: 66-95, or ~41%. Now, I have not bothered to sift through every TvZ VOD post-10-17-08 to sort out which ones used a mech build or some variant thereof, and then count the wins and losses. Pragmatically speaking, I haven't that kind of time, nor the inclination, for as much as I love the matchup, I think the mech build has ruined its beauty and made it too similar to ZvP. I also find sifting through the TLPD and playing with numbers much more appealing than watching 161 VODs.

All of that is to say, I cannot claim for sure that the mech build has resulted in the ~1.6% dip in the Zerg win rate. I can claim that with even less certainty because I have no idea what caused the Zerg win rate to drop those few points from its background rate in the first place. There are any number of explanations, because the win rate was already falling. If I was asked what the most likely cause of the the post 10-17-08 dip was, I would say that it was pure chance. This, however, is a point of debate.

I understand that I have neglected to establish what kind of a move is statistically significant in BroodWar between years/events/tournaments. This is most pragmatically measured by looking at the number of games affected. In this case, the 1.6% translates to ~2.5 games. Clearly, this is insignificant. The forced conclusion, of this, then, is that the mech build has not had a dramatic affect on Zerg win rates, and thus has not had a dramatic affect on the matchup. Now, again, you can cry fowl on me because I have admitted myself that I have not looked through every VOD, I cannot know for sure if every single player who has gone mech has not lost (or, for that matter, won). While this may be true, my goal was to establish if the mech build has had a dramatic affect on the matchup as a whole, which it provably has not.

With that having been said, it is probably a better conclusion to say that there are too few data to know for sure. If Zerg win rates continue to decline, and the mech build gains in popularity and use, perhaps it will be possible to claim that the mech build has resulted in fewer games won for Zerg. At present, however, it is safe to claim either no affect, or too few data.

Last Stand

What toll has the Golden Age of Protoss taken on the Zerg swarm? First, look at the background rate vs. the present rate: 60% v 51%. That's over the year that this change has happened. First, I'll do the short answer. A difference of approximately 9%, over the course of 426 games played translates to 38.34 more games going in favor of Protoss than Zerg. This is quite significant. There is perhaps no better example for this than the Classic Season 2, where below average TvZ (59.6%) more than combined and compensated with ZvP (43.1%) to completely eliminate the Zerg from the league.

With that question easily and quickly answered, my curiosity has turned to the dragons, free, Kal, JangBi, Stork, Bisu, and BeSt. Is is possible that the rampaging affects of these giants has combined for a good part of the change? It doesn't seem so far fetched. These players are staples of individual leagues and see considerable time in Proleague. The Club Day MSL fielded 5 of them in the round of 8, and one became the champion. The Incruit OSL saw 3 of them (and only one Zerg) in the round of 8. All of them made it to the Ro16 in the Classic S2, and 3 of them still remain.

Here are some facts:

* Bisu is 7-1 in Proleague vs. Zergs, steamrolling YellOw[ArnC], by.hero, Luxury, Calm, and EffOrt. His only loss was to Jaedong (and there's no shame in that).
* Stork is 3-2 in Proleague vs. Zergs, defeating GGPlay, ZerO, and Luxury and losing to Calm and type-b.
* BeSt is 4-3 in Proleague vs. Zergs, winning over Shark, ZerO, RorO, and sAviOr. BeSt is also the weakest dragon against Zerg.
* Kal is 7-2 in Proleague vs. Zergs, most notably defeating Luxury and Jaedong.
* JangBi is 2-3, losing to EffOrt, Shine[kaL] and hyvaa.
* free is 6-1, only losing to Saint.


Combined, the dragons are 29-12 in Proleague against Zerg. That's an astonishing 70%. Looking down the list, however, it does not seem so good to combine these win rates. After all, they range from an astonishing 87.5% for Bisu to a mediocre 40% for JangBi. With such a range, to combine them seems somewhat dishonest. But let's also look at their individual league results vZerg starting with the Incruit OSL, Club Day MSL, the Classic S2, Lost Saga MSL and Batoo OSL. Essentially, the leagues start with and since they have been called dragons.

* Bisu is 8-5.
* Stork is 5-1.
* BeSt is 4-2.
* Kal is 4-2.
* JangBi is 6-0.
* free is 4-2.


Look at these against the Proleague stats. The performances of the dragons appear complementary between individual and team leagues. The dismal JangBi in Proleague is at 100% winrate in individual leagues. Stork's middling 60% is now ~83%. The least changed is BeSt, whose PvZ continues to be his worst matchup.

Once again, combining them, the dragons are 31-12, an even more astonishing 72%. Okay, so the dragons are ripping and tearing through Zergs like paper. That much is true. But it's not enough to claim for sure that the dragons are the exclusive cause, or even the leading cause, for a 9% drop in Zerg win rates. What needs to be established for this to be claimed with accuracy is that Protoss who are not the dragons are struggling just as much (or hopefully worse) against Zerg.

The trouble here is that I could easily doctor this to make my point. I could pick a bunch of rookies who, in particular, have sucked against Zerg and say 'Well, damn, looks like I'm right again!' I'm going to resist that urge, and say that I'm going to pick as many Protoss players as I can who have played in at least two of the leagues that the dragons have, and compare their results to the dragons. The idea is that the exact competition is somewhat controlled, and the circumstances (maps, pressure situations) are also controlled. Also, that would establish that the players are not bad players, as they would have had to qualify for at least two of the leagues. It turns out there are a couple of players for whom these standards are true: BackHo, Much, Tester, and HerO.

* BackHo is 3-4.
* Much is 2-2.
* Tester is 2-3.
* HerO is 2-3.


I don't really think that I need to enumerate the differences in the record and percentages (for the record, it's 42.8%). So how about players who aren't spectacular in the individual leagues? Like, Say, Pusan and SangHo? Well, Pusan is 4-0 this ProLeague, but among his victims are two ACE players, the Zerg problem child Kwanro, and hyvaa. SangHo is 8-5 in his last 13 vZs, a 61% win rate. I could spend my time disparaging SangHo's victims too, but I don't feel so much of a need to. How about Anytime and Rock, two other staples of Proleague? Admittedly, they are both slumping (although Rock's committment to mediocrity has always been life-long). Anytime is sitting at 1-3 this Proleague and 7-3 through the year, and Rock is a dismal 4-8 since January of '08.

The only stone left unturned, then, are the Protoss rookies. These I would very much like to investigate, but being that they are rookies, they have played far too few games (a) for their win percentages to be accurate representations of their skill in the matchup and (b) for their wins to contribute to an overview of the matchup.

So I'm left to explain the success of Anytime and SangHo in 2008 if I want to say that it is the dragons in the end that have caused the dip in Zerg win rates, and then one more thing to explain after that, which I'll get to. Now, explaining SangHo: 8-5, defeating acknowledged ZvP travesties MuMyung, RorO (twice), and Study. He also defeated JoJoBa in JoJoBa's first recorded games. SangHo's impressive victories are numbered 2, defeating Jaedong and ZerO. SangHo also had the benefits of Plasma, Byzantium, Colosseum II, and Medusa. SangHo also managed to lose to OversKy, newbie JoJoBa, Yarnc's less than impressive ZvP, and great. Now I'm not going to try to and establish SangHo's ZvP success as pure circumstantial, lucky, or purely attributed to favored maps and poor opponents. To claim that would be entirely untrue. However, it is important to note the disparity between SangHo's winrate and the dragon's, 61% to ~71%.

Fortunately, Anytime's success is more easily explained. His 6 game PvZ winstreak to start 2008 starred a dramatically slumping sAviOr, complete newbie besga, ZerO, and a vP failure in Kwanro as his victims. Twice he slaughtered sAviOr, once on Baekmagoji, once on the horribly imbalanced Katrina. Complete newbie besga fell victim as much to Anytime's pushes as he did to Loki II's sickening 4-9 ZvP record. Anytime did beat oDin fair and square though.

Now, the tour de force of this explanation is the following. I've counted the number of games won and lost by the dragons pre- and through 2008. It turns out that the dragons brought home exactly 28 more games for Protoss in 2008 than they had from years before. Recall that 38.34 is the number of games more brought in for Protoss in 2008 than in years before. You'll notice that 28 is not 38.34. However, I never pretended to establish that the dragons are the singular cause for the drop in zerg winrates. Given that other players, with some exceptions, have been returning rather middling results, and the dragons have increased their PvZ winrates by approximately 10%, it is not outrageous to say that 6.5% (28's relation to 38.34 in respect to the 9% difference) of the 9% drop in win rates could be exclusively the result of the Protoss dragons. Now, I cannot claim this with perfect accuracy. Doubtless, the results of other Protoss players pre- and through 2008, in sufficient number, could possibly be made to show an increase of 38 wins. I understand that I have nothing concrete to connect these 28 wins with the requisite 38 extra in question, but it is my strong suspicion that this is the case. However, I am open to other opinions on this matter.

Tangentially related, I want to introduce the hope of the Swarm to end this era of tyrants. I present to you: The Dragon Slayers: July: 5-2 (3-0 v Best, 2-2 v Bisu) in the last year, and Jaedong 17-7 (damn). Also tangentially related, to end the era of vicious acorns, maGma has blessedly volunteered for the task. maGma isØ-Ø lifetime against acorns.

The Insight of Casey Stengel

Casey Stengel was a baseball player, famous for saying this: "See that kid over there? He's 20 years old. In 10 years, he's got a chance to be a star. Now, see that other kid over there? He's 20 too. In 10 years, he's got a chance to be 30."

If there's one thing that's struck me during the Proleague over 2008, it's the sheer number of new faces for the Zerg race. Names like by.hero, type-b, EffOrt, by.great, Shine[kaL], hyvaa, HoeJJa, Juni, ZerO, RorO, and s2 have all made their debuts recently and are quickly becoming staples of the rosters of their respective teams. Even more players, like Clay, Robbie, Modesty, and others have played their first games in 2008. This may not be unique among the new blood entering the scene, but it does have many implications for the future of the Swarm. Each one of the players named have had success, many of them have had more failures in 2008 than successes, but almost every career upon entering the scene becomes that way.

So who should you Liquibet on? Who should you cheer for? Who will make your eyes bleed as you watch? Unlike the rest of this article, the following section is mostly my opinion. But, before I get to the newbies, let's look over the veterans. There are some things that may surprise you. Let me start with this one.

1. GGPlay was miserable in 2008. His sour record (25-26 since the start of the year) was clearly one of the reasons why CJ dumped him (although he has improved somewhat since joining the Stars).

2. July still kicks butt. His 10-4 Proleague record is one of the more impressive records for Zerg right now, even if perceptions are that Calm has him trumped. For the record, July went 33-24 in '08. Calm went 30-25. How about that?

3. OversKy is perplexing. With everyone on ACE constantly failing around him, this guy has managed to go 16-18 in the year. He's trumped SangHo, ZerO, Jaedong, Canata, and FrOzean twice. I hope he keeps playing after his Airforce days, because I really like him. Note: I called GGPlay miserable for a better win percentage than OversKy. I know this. It's because OversKy is on a crappy team with reduced practice everything, and GGPlay is on a good team with excellent practice everything. It's the factor of expectations. It doesn't surprise anyone that MuMyung sucks.

4. Jaedong went (I only counted once, and there were a lot of games. I may be off by one or two) 81-41 in 2008. Most valuable Zerg? You bet.

5. It breaks my heart to say it, but the twins are over-hyped. Luxury went 49-39 in '08, YellOw[ArnC] a sad 28-28. YellOw[ArnC] is carrying OGN on his back? Try Leta.

6. There's absolutely nothing to say about sAviOr in 2008. That story belongs to 2009. I'm still holding on to this despite two losses...

That takes care of the old guys, basically the only Zergs who've seen more than a few Proleagues. I'm being honest about that. Most of the Zergs seeing a lot of games these days are newbies. Some are good. Some will be good. Some are bad, but might be good. But most are very, very bad. Let's start with the good, the five I'm calling up.

1. EffOrt. CJ's new workhorse. They're putting everything they have into this kid. His record in 2008 was very promising, 22-15. This guy has not been given an easy workload, and he's improving rapidly. Look at his games from Classic 1 to Classic 2. He's only going to get better.

2. ZerO. This kid is the reaL deal, and I'm calling it--he's going to be the Zerg Leta soon. What he really needs is mentoring, and I don't think GGPlay is the guy to do it. In a bit, a couple months maybe, he's going to explode.

3. by.hero. His games against Bisu were impressive, but he's quietly been doing work for some time now. He's in both leagues, and more important he has the guidance of July and Calm. He won't go much farther this year, I don't think. Next year, though, look out.

4. Saint. Saint isn't exactly new, but winning is certainly new to him. This is my hunch bet. He's going to get better, and he's going to gain confidence. MBCGame is putting a lot into him. All that playing time is going to translate into considerable improvement, I think. Of course, there's plenty of disagreement.

5. type-b. Last Starleague go-round, I didn't believe it. He seemed like a one-hit wonder (circa Thezerg), but his Proleague performance has been...unexpected. Unlike Thezerg, who decided to drink a giant can of how to suck at StarCraft, type-b took his lucky Starleague runs and won some Proleague games against good players, like Stork, Hwasin, Light, Really, and YellOw[ArnC]. He may not be great now, but he'll get there.

Now for the five down, down, down.

1. s2. I just hate it when this guy plays. It's like being repeatedly propositioned by Richard Simmons for 'cuddling lessons.' SKT rolls over and accepts a loss in the Zerg slot whenever s2 is in the lineup. Of course, then SKT buys Thezerg, which means they accept a more expensive loss in the Zerg slot.

2. Thezerg. Did you see how excited he was when he beat LuCifer? It's like he couldn't believe it himself that he'd actually won a game. To tell the truth, neither did I. Don't forget the 1-9 he went 8-19-08 and 12-19-08. Wins against that sack of medicority Shark and Mr. Minor League LuCifer aren't going to save him from his Destiny at the bottom of the SKT Zerg lineup. What I don't get is why they never play GoRush. He actually made it out of the qualifiers!

3. RorO. 2-11 between 11-16-08 and 12-24-08. But I can understand WeMade's incredible desire to try someone else out when keke is their only other option.

4. keke. See a pattern? The guy who stinks, and the guy brought in to replace him. While keke isn't as bad as the others on this list, since he has recieved more playing time, he's done nothing with it, and still continued his relentless march to being nothing exceptional. His main selling points were his Neo Requiem ZvT wins. Whoop. Dee. Doo.

5. FireFist. 10-11 isn't going to get you a spot in the lineup on a team with the arguably better Haran, 815 and HoeJJa, and then the definitely better Luxury. FireFist isn't going to suffer from getting worse, but is rather going to suffer from not being played for being worse than Haran and 815 and not as young and promising as HoeJJa.

Don't sue me if I'm wrong, but feel free to disagree.

So that's that. The State of the Swarm report. The mech build hasn't decimated the Zerg, but the Golden Age of Protoss has. The Zerg old guard are putting up unconvincing results, excepting a few, and as such, I expect that the new promising Zerg players are going to lead the Swarm out of this troubling time in StarCraft history. Cross your fingers for EffOrt for OSL champion. I am.