ZvZ build order statistics

So I charted some (208) pro ZvZ games.

My motivation for doing this was that people have always claimed that ZvZ is just build order rock-paper-scissors. But how true is that claim? And even if it is mostly true there are still games when players win despite a build order disadvantage; examining such games should be quite educational.

There are also lots of other questions to be answered. How frequently do various builds get used? How do these choices vary across 2/3/4 player maps? How do certain build order fare against each other? Do certain players favor certain builds? And so on.

Anyways, on to the results!

People say ZvZ is 9p>12h>12p>9p

12h v 12p:  28-17
12p v 9p: 13-11
9p v 12h: 20-7

but it seems closer to 9p>>>12h>>12p>9p. It would be nice to have a larger sample size for 12p v 9p.

What about builds other than those three?

Build frequencies:
 12h - 134
 12p - 127
 9p - 81

 overgas - 58

 overpool - 11
 10h - 3
 9h - 1
 4p - 1

Overgas stands out as something we're not considering. OK, so how does it fit into the picture? My understanding was that it hard counters 9pool but is way behind vs later pool builds

overgas vs...
  12h: 2-4
  12p: 8-11
  9p: 6-5

  overpool: 2-1
  4p: 0-1

but it seems that overgas doesn't really have such a simple characterization.
And the others?

Build matchup records:
 10h
  12h: 0-1
  12p: 1-0
  9p: 1-0
 9h
  12h: 0-1
 overpool
  12h: 1-1
  12p: 2-0
  9p: 2-0
  overgas: 1-2
 4p
  overgas: 1-0

Super small sample size!

Here is the big picture

 12h: 68-66
 12p: 59-68
 9p: 44-37
 overgas: 27-31
 other: 10-6

I have some other things but I am too lazy to talk about it so here are those things
+ Show Spoiler [other things] +

+ Show Spoiler [Map number frequencies] +

+ Show Spoiler [Build frequencies by map number] +

+ Show Spoiler [Build records by map number] +

+ Show Spoiler [Build matchup records by map number] +

+ Show Spoiler [Build matchup records by player] +

 action
  12h
   12h: 0-1
   12p: 2-0
   9p: 0-2
  12p
   12p: 1-2
  9p
   12h: 0-2
   12p: 0-1
   9p: 3-0
  other
   9p: 1-0
  overgas
   12p: 0-1
   overgas: 2-0
 alone
  12h
   12p: 0-1
  9p
   12p: 1-1
  other
   overgas: 0-1
  overgas
   12p: 1-0
   9p: 0-1
   other: 0-1
 byul
  12p
   9p: 0-1
  overgas
   9p: 1-0
   other: 1-0
 calm
  12h
   12h: 2-1
   12p: 2-1
   9p: 0-1
   other: 1-0
   overgas: 0-1
  12p
   12h: 4-1
   12p: 2-1
   9p: 0-1
   overgas: 1-0
  9p
   12p: 0-3
   9p: 0-2
   overgas: 0-1
  overgas
   12p: 0-1
   9p: 0-1
 chavi
  12h
   12p: 1-0
 crazy-hydra
  12h
   12h: 2-1
   12p: 2-0
  12p
   12h: 1-3
   12p: 2-0
   9p: 1-0
   other: 0-1
   overgas: 0-1
  9p
   12h: 1-0
   9p: 1-0
   overgas: 0-1
  overgas
   12p: 1-1
 dream.t)check
  12p
   overgas: 0-1
  overgas
   12h: 0-1
   overgas: 0-1
 effort
  12h
   12p: 1-0
  12p
   12h: 0-1
   12p: 1-0
  9p
   12h: 1-0
   9p: 0-1
  other
   other: 0-1
  overgas
   12p: 1-0
 ggaemo
  12h
   12h: 1-0
   12p: 0-3
   9p: 0-1
  12p
   12p: 0-1
   9p: 1-1
  9p
   12h: 0-1
   overgas: 0-1
  other
   overgas: 1-0
  overgas
   12p: 0-1
   overgas: 0-2
 great
  12h
   12h: 1-0
   9p: 1-0
  12p
   12p: 0-1
   9p: 0-1
  9p
   12h: 0-1
   12p: 0-1
   9p: 0-1
  overgas
   12p: 0-2
   overgas: 1-1
 hero
  12h
   12p: 1-0
  12p
   12p: 0-2
   9p: 0-1
   overgas: 0-1
  overgas
   overgas: 0-1
 hero[join]
  12h
   12p: 2-1
  12p
   9p: 1-0
  other
   12p: 1-0
 hoejja
  12h
   12h: 1-1
   12p: 1-2
   9p: 0-1
   other: 1-0
   overgas: 1-0
  12p
   12h: 1-0
   12p: 1-0
   9p: 1-0
  9p
   12h: 1-0
  other
   other: 1-0
  overgas
   9p: 1-0
 hope
  12p
   overgas: 0-1
 hydra
  12h
   12h: 2-1
   12p: 0-2
  12p
   12h: 2-0
   12p: 4-2
   9p: 1-0
   overgas: 3-0
  9p
   12h: 2-0
   12p: 1-1
   other: 0-1
   overgas: 0-1
  overgas
   12p: 1-1
   9p: 0-1
   overgas: 1-0
 hyuk
  12h
   12h: 0-1
   12p: 1-1
   9p: 1-0
  12p
   12h: 1-1
   12p: 1-2
   9p: 1-0
   other: 0-1
   overgas: 1-0
  overgas
   9p: 1-0
 hyun
  12h
   12h: 2-1
   12p: 3-0
   9p: 0-2
  12p
   12h: 0-1
   overgas: 1-1
  other
   12h: 0-1
  overgas
   12p: 0-1
   overgas: 0-1
 hyvaa
  12h
   12h: 0-1
   12p: 1-0
   9p: 0-1
   overgas: 1-0
  12p
   12h: 0-1
   12p: 0-1
   overgas: 0-2
  9p
   12h: 2-1
   12p: 1-0
   overgas: 0-1
  overgas
   other: 0-1
 jaedong
  12h
   12h: 2-2
   12p: 1-0
   9p: 2-3
  12p
   12h: 0-4
   12p: 2-0
   9p: 1-0
   overgas: 1-0
  9p
   12h: 1-1
   12p: 1-2
   9p: 2-0
  other
   12p: 1-0
  overgas
   12p: 1-0
   overgas: 0-1
 juni
  12h
   12h: 0-3
   9p: 0-1
  12p
   12h: 0-3
  9p
   overgas: 1-0
  overgas
   9p: 1-0
 killer
  12h
   12h: 1-1
   12p: 3-1
   9p: 2-1
  12p
   12h: 4-3
   9p: 0-1
  9p
   12p: 1-1
   9p: 1-1
  other
   9p: 1-0
  overgas
   12h: 2-0
   12p: 1-1
   9p: 1-0
   overgas: 1-0
 modesty
  12h
   12h: 1-3
   12p: 1-1
  12p
   12h: 1-1
   12p: 0-1
   other: 0-1
   overgas: 2-0
  9p
   12h: 1-0
   12p: 0-1
   9p: 1-0
  other
   12p: 1-0
 orion
  12p
   12h: 0-1
   9p: 0-1
 peace
  12h
   12p: 1-1
   other: 0-1
  12p
   9p: 1-0
   overgas: 1-0
  9p
   12h: 1-0
   overgas: 0-1
 perfective
  12h
   12h: 0-1
   12p: 1-0
   9p: 0-1
  12p
   overgas: 0-1
  9p
   12h: 1-0
  overgas
   9p: 1-1
 roro
  12h
   12h: 2-2
   12p: 0-1
   overgas: 1-0
  12p
   12p: 0-1
   9p: 0-1
  9p
   12h: 1-1
   12p: 1-1
   9p: 0-1
   overgas: 1-0
  overgas
   12h: 0-1
   12p: 0-1
   overgas: 0-1
 s2
  12h
   12h: 1-1
   12p: 1-0
  12p
   9p: 1-0
  9p
   other: 0-1
   overgas: 1-0
  overgas
   12h: 0-2
   12p: 1-0
   other: 1-0
   overgas: 2-0
 sacsri
  12h
   12h: 0-1
 saint
  12h
   12h: 0-1
  12p
   12p: 0-1
  9p
   9p: 0-1
 sascri
  12h
   9p: 0-1
 shine
  12h
   9p: 0-1
  12p
   12h: 0-1
   12p: 1-1
   9p: 1-0
  9p
   overgas: 1-0
  other
   12h: 0-1
   9p: 1-0
 soo
  12h
   12p: 1-1
  12p
   12h: 0-1
   9p: 0-1
  9p
   12h: 1-0
   9p: 0-1
  other
   12h: 1-0
   overgas: 1-0
  overgas
   overgas: 1-0
 soulkey
  12h
   12h: 4-1
   12p: 2-0
  12p
   12h: 1-1
   9p: 2-1
   overgas: 1-0
  9p
   12p: 1-1
  other
   12h: 0-1
  overgas
   12p: 1-0
 where
  12h
   12h: 1-0
   9p: 0-1
   other: 1-0
  9p
   12h: 1-0
 woon
  12h
   12h: 0-1
   overgas: 0-1
  9p
   12h: 1-0
 wow
  overgas
   12p: 0-1
 zerg[kal]
  12h
   9p: 0-1
   overgas: 1-0
  12p
   12h: 0-1
   12p: 0-1
  other
   overgas: 0-1
 zero
  12h
   12h: 3-1
   12p: 0-1
   9p: 1-2
  12p
   12h: 2-4
   12p: 3-1
   9p: 1-1
  9p
   12h: 5-0
   12p: 4-0
   other: 0-1
   overgas: 1-0
  overgas
   9p: 0-1
   overgas: 1-1

If you want to search for specific games, here is the place to do it.

The format is

number of player map|winning player|losing player|winning build|losing build

+ Show Spoiler [data] +
+ Show Spoiler [some notes about the data] +

So yeah, I hope this was interesting and useful

brood war for life, brood war forever

nbaker United States. June 28 2012 10:31. Posts 928

Profile #

Hmm, this is very interesting. I think recently 12 pool has been doing better vs. 12 hatch, but it's definitely still a disadvantage. 9 pool vs 12 pool feels the most imbalanced to me when I play, but with progamer ling and muta micro, players have kept it pretty close.

I'm curious what good players think of 11pool vs 12hatch. Sometimes I win flat out with lings and sometimes it's not even close and I'm way behind.

endy China. June 28 2012 10:43. Posts 6073

Profile Blog #

Very interesting post !
9 pool seems to be the best build statistically, at least recently.

Congrats for 10000th !

★ SK Telecom T1 ★ MSL Forever ★ SonicBJ is my hero ★ Kim Taek Yong ♥ I took a photo with Bisu before it was mainstream

ninazerg United States. June 28 2012 13:01. Posts 1832

Profile Blog #

On June 28 2012 10:31 nbaker wrote:
Hmm, this is very interesting. I think recently 12 pool has been doing better vs. 12 hatch, but it's definitely still a disadvantage. 9 pool vs 12 pool feels the most imbalanced to me when I play, but with progamer ling and muta micro, players have kept it pretty close.

I'm curious what good players think of 11pool vs 12hatch. Sometimes I win flat out with lings and sometimes it's not even close and I'm way behind.

I'm not particularly good (C zerg), and I tried 11pool as an actual opening, and although it's technically "safe", it doesn't allow you to attack a person going for 12 hatch, and doesn't let you match the aggression of a 9pool build, which is particularly tricky if you're going 11pool as means of expanding.

Instead of an expand build, you could be going for a fast tech build, but it seems like Overpool and 12pool are better suited for 1-base 2-hatchery play.

Don't know entirely for sure, though! Just giving some thought to it based on the games I've played with it so far.

Holder of Excalibur, savior of Eria.

FlaShFTW United States. June 28 2012 13:26. Posts 4955

Profile Blog #

9 pool FTW!!! :D

congratz on 10000!!! ill make it there in like a decade xD

Never Surrender. Never GG. Always Fighting. BW4LYF. Woo Jung Ho Never Forget.

TheShimmy United States. June 28 2012 13:38. Posts 1723

Profile #

This is really awesome!

A truly intriguing take for any Zerg.

Hyvaa #1 Fan. Gogo STX, Dear, Bogus, Classic, and Mini! Always a BW fan!

hacklebeast United States. June 28 2012 16:32. Posts 4630

Profile Blog #

i'm surprised that there were that many overgases. Do the stats include mirrors?

Protoss: Best, Paralyze, Jangbi, Nal_Ra || Terran: Oov, Boxer, Fantasy, Hiya|| Zerg: Yellow, Zero

Crunchums United States. June 28 2012 16:54. Posts 10372

Profile Blog #

On June 28 2012 10:43 endy wrote:
Congrats for 10000th !

On June 28 2012 13:26 FlaShFTW wrote:
congratz on 10000!!!

thanks : D

On June 28 2012 16:32 hacklebeast wrote:
i'm surprised that there were that many overgases. Do the stats include mirrors?

I was too! My intuition is that overgas is fairly map dependant (used 22% of the time on 2 player maps, 9.5% of the time on 3 player maps, 12% on the time on 4 player maps) and since most of these games are from proleague where ZvZ's are more likely to be on certain (zerg favored) maps, maps where overgas is relatively strong loom large in the data. Anecdotally, while collecting the data I felt like pros favored overgas on Alternative (where 29 of 94 matches were ZvZ).

Mirrors are in there but I took them out when presenting matchup data above because it looked silly to say "hey, did you know that overgas is 9-9 vs overgas? fascinating!" Anyways here are the mirrors:

12h - 26
12p - 18
overgas - 9
9p - 8
overpool - 1

brood war for life, brood war forever

corumjhaelen France. June 28 2012 21:32. Posts 4782

Profile Blog #

As someone who loves ZvZ, thank you very much for this, it's really interesting. Not really sure what to conclude from it, apart from the fact that the rock paper scissors affirmation seems really overblown =)

‎numquam se plus agere quam nihil cum ageret, numquam minus solum esse quam cum solus esset

Shiladie Canada. June 29 2012 01:07. Posts 1618

Profile Blog #

Awesome work, thank you!

I'd also be curious to find out how long each of those games are, so we can eliminate games that go long enough that the opening BO matters less. I think we'll see even stronger support for the RPS nature of that stage in the game.
Also it would be interesting to see Jaedong's stats during his JvZ reign of terror and how much those games may have affected these results (if this goes back that far)

Last edit: 2012-06-29 01:07:45

The evolution of SC2: T or P comes up with something abusive and Z goes "I gotta deal with THIS shit too?"

Sayle United Kingdom. June 29 2012 01:55. Posts 2695

Profile #

Nice work. Makes me wonder why nobody bothered to do it before (especially the proponents of ZvZ not being terrible aka Harem).

http://www.teamliquid.net/forum/viewmessage.php?topic_id=390936

BisuDagger United States. June 29 2012 02:15. Posts 5373

Profile Blog #

Grats on 10,000. This tells me to 9pool everytime haha!

~BD, not just for BisuDagger but for BroodWar Doctor

Bakuryu Germany. June 29 2012 02:17. Posts 268

Profile Blog #

thx for the data, makes me wonder why they play 9p so rarely (could be alternative...)

Crunchums United States. June 29 2012 04:21. Posts 10372

Profile Blog #

On June 29 2012 01:55 Sayle wrote:
Nice work. Makes me wonder why nobody bothered to do it before.

Well there is some work associated with setting yourself up to use data usefully (if I could do it again I would use a database instead of writing a bunch of silly python <__<;; ), but I think the biggest thing is that collecting data is pretty dull. 208 games felt like a ton, but it still wasn't enough to give the sample sizes you'd like.

On June 29 2012 01:07 Shiladie wrote:
I'd also be curious to find out how long each of those games are, so we can eliminate games that go long enough that the opening BO matters less.

Although I think build orders can still matter in longer games, tracking game length definitely seems like something worth doing. Practically it might be difficult though - iirc when I started watching BW they didn't have the ellapsed time displayed like they do now.

I think we'll see even stronger support for the RPS nature of that stage in the game.

That would be entirely unsurprising given the way certain matchups tend to play out (eg 9p laughing at 12h, 12h overpowering 12p with lings right before mutas).

Also it would be interesting to see Jaedong's stats during his JvZ reign of terror and how much those games may have affected these results (if this goes back that far)

My guess is that my data doesn't go back that far, but that would definitely be interesting.

On June 29 2012 02:17 Bakuryu wrote:
thx for the data, makes me wonder why they play 9p so rarely (could be alternative...)

9p was chosen 16.4% / 22.6% / 20.3 % of the time on 2 / 3 / 4 player maps respectively so... I wish I had done this better so we could look at splits by individual maps rather than just number of players for the map :<

brood war for life, brood war forever

sheaRZerg United States. June 29 2012 15:02. Posts 512

Profile Blog #

I think the large number of 12 h mirrors (compared to total 12 h v. total 12p) indicates there are particular maps where it is viewed as better (electric circuit, for instance, comes to mind).

"Dude, just don't listen to what I say; listen to what I mean." -Sean Plott

Miwyfe England. June 30 2012 08:55. Posts 76

Profile #

Awesome work nice one!
The data from 2 player maps is much more valuable than 3 and 4 player maps because overlord scouting is not recorded in the data and players know where to send lings.

OK so on to the data.
In the Other things spoiler tag you have:

Map Number Frequencies
2 - 70
So 70 games were played on 2 player maps in your sample.

Build frequencies by map number
+ Show Spoiler +

So 12pool was played 39 times out of 140 builds.

Build records by map number
+ Show Spoiler +

So 12pool won 22 times in total vs other builds and lost 17 times.

Build matchup records by map number
+ Show Spoiler +
So 12 hatch beat 12 pool 8-5
12 pool beat 9 pool 7-1
9 pool beat 12 hatch 6-2

What does all this mean. Well unfortunately it means nothing. The 12 hatch might have beat the 12 pool everytime on one map and lost 5 times on another map.

Maps and spawn positions dictate build orders and these dictate the winner in broodwar! When doing statistics on broodwar, always separate data by the map then by spawn position. Only then can you start to analyze the data.

So as you can see I would personally group the data primarily by the map it was played on. Also I would only use 2 player maps and I would record where each player spawned (yes even on 2 player maps, and especially in zvz). If I were to do 3 or 4 player maps, I would only use games where both players scouted first time with overlords (or both players scouted second etc), and I would indicate which starting positions each player was in.

How about you though Crunchums? You said you would also do things differently if you were to do it again.

Last edit: 2012-06-30 10:24:42

GRCJH Canada. June 30 2012 11:09. Posts 76

Profile #

whoa, am I missing something? I never ever see 12 hatch

you were born too soon, you'll never explore the galaxy

FlaShFTW United States. June 30 2012 13:21. Posts 4955

Profile Blog #

On June 30 2012 11:09 GRCJH wrote:
whoa, am I missing something? I never ever see 12 hatch

wut. 12 hatch is one of the most popular. especially on the 4 player maps.

Never Surrender. Never GG. Always Fighting. BW4LYF. Woo Jung Ho Never Forget.

Mortality United States. July 01 2012 00:24. Posts 4790

Profile Blog #

Well...

The classical rock-paper-scissors element to ZvZ started existed prior to modern muta micro and was most prominent with early maps like Lost Temple, which has relatively short rush timings compared to most modern maps. With shorter rush distances, 9p > 12h becomes even more pronounced, meanwhile without modern muta micro, 12h > 12p and especially 12p > 9p become more pronounced.

And I always thought overgas was more of a counter to overpool anyway. Lately it seems overpool is out of fashion, but for a long time it was a staple of ZvZ, and probably the most common build.

Even though this Proleague bullshit has been completely bogus, I really, really, really do not see how Khan can lose this. I swear I will kill myself if they do. - nesix before KHAN lost to eNature

Chef July 01 2012 20:56. Posts 9708

Profile Blog #

I keep wanting to make a post in this thread but then have to delete it haha.

I think it's undeniable that in all matchups in BW build orders can give player an advantage. Furthermore, a player who is better than his or her opponent will make up for a disadvantage in build order the longer the game goes on, the more decisions he or she is able to make that are better than his or her opponent's. Since ZvZ is short, you might conclude that that means that the advantage is going to be stronger since there is less time to overcome it, HOWEVER, I believe that within those short ZvZs players have to make more decisions over a shorter amount of time. That takes focus and precision which is not in every players skillset (at least, not their strongest skill). A player like Jaedong or Hydra or any other ZvZer with a 60% + record in ZvZ achieves this with superior ling micro and decisions and knowledge about ZvZ. And that record isn't particularly less pronounced than any other record in any other matchup. So what I am trying to say is that the better player is still winning a fair amount of the time in ZvZ, and perhaps that better player is what we'd call the 33% (although admitedly it is very easy to lose a ZvZ just from one or two mistakes, which is one reason people begin to feel the matchup is volitile—but that has nothing to do with the build orders).

ty for the stats. I think you should have included weighting for when a player was simply favoured to win in ZvZ. I also don't know if you chose the best season to gather stats, since by this time so many teams were disbanding and there were a lot of things to hurt players skills. It's difficult to read into this data but I'm still glad you made the effort to collect it. IMO rock paper sissors proof would look closer to 90% + wins in build order 'counters' Where this isn't much more extreme than someone getting away with 14 nex vs a cautious player.

I think since ZvZ is so much about being 'in the zone' it would be interesting to look at the streaks players go on. When I just look briefly over a few profiles, I often see either a lot of wins in a row or a lot of losses. Especially noticable with Hydra, whose record is actually only 55% even tho there are times when we have thought his ZvZ was nearly unbeatable.

Last edit: 2012-07-01 21:00:55

LEGEND!! LEGEND!!

1 2 3 Next All

Please log in or register to reply.