|
Katowice25012 Posts
It felt weird without a banner so I googled a pic of Flash. Look at him! So triumphant!
Stat pieces always begin with a question. When you're me its often triggered by beer, but usually they begin with a question. This time the question came to me while watching OSL finals. In the fifth game Flash opened with a fast command center, and as he was getting overrun with lings I thought this is silly, doesn't Flash always lose when he does this?
I specifically remember a time when there were a lot of posts to the effect of "Flash does silly nonstandard things too often in TvZ". Apparently that same sentiment was common in the KT house, because last fall he began to open 1 rax cc every game and it was the beginning of his prolonged streak of complete TvZ dominance.
Regardless, the game finished and I forgot about it, choosing to spend time searching the forums for angry fans instead of pondering life's great mysteries (why do hot dog buns come in packs of 8 while the hot dogs come in packs of 10?). The next day it hovered in the back of my mind while I underwent a multi-hour battle with Golden Tee. I toyed with the idea of hand collecting the stats of how often he won one of his TvZ games with a double cc opening, and like any good statistician, decided it was too much work.
Then that crazy little monster did it again, only a few days later. This time was different, as not only did he use it three times in a row, but he crushed his opponent every time. As if that wasn't enough, "his opponent" wasn't just some guy, but Jaedong. He crushed Jaedong on his way to fame, fortune, glory, a cool trophy, and the envy of a million 14 year old girls. How does it make you feel to know Justin Beiber shares a fanbase as the orchestrator of a zerg genocide?
Anyway, bricks were ready, mortor being laid. Destiny. My hand was forced. Five paragraphs later here I am, saying "I made a chart and here are some numbers". Its okay because the actual chart takes about 3 seconds to digest all the possible info you look for in a thread like this.
The first step was looking at Flash's games. He has 173 TvZs in our database, but 14 of those are unusable (mostly prelims, one or two games I couldn't dig up online) so effectively 159 total. I then watched all of them long enough see what his first non-depot building was, and made a note for the games that opened with a 14CC. Here are the charts. They are pie charts, and are damn fine.
See that 3d effect? This is next gen tech.
This is the overall breakdown of how many times he used an early CC over other builds (22 instances found!). It would be pretty cool to actually break this down further into what percentage of play his 1 rax CC is vs 2 rax vs forward bunker type things, but I didn't think of it at the time. For those wondering, if you take out the MSL series vs Jaedong it drops to 12% which still seems like a signifigant chunk.
Rumble, we hardly knew ye.
This is the breakdown of what players he did it against. Unsurprisingly he does it more against people he has played a lot, generally during a longer series. Interestingly, he is 0-3 against GGPlay using fast cc (all during the same Bo5). He is 6-3 against Jaedong, 2-1 against Savior, and 2-0 against Kwanro. Of the people he has tried it against once, he has beaten them all except EffOrt.
Another note to consider are the maps, if you look at the list of games (below) you see its far more common on 4 player maps, and as well his record is far better on bigger maps. Not surprising, but its nice to have actual data that backs the things people seem to "just know". This the most fascinating cross section of the data to me, and in retrospect I wish I had gotten the numbers for how often it happens using the maps as a constant instead of a player. It seems likely that they are the single biggest factor.
This is also a chart of things that look like pacman
At 14 wins and 8 loses, he has a pretty solid history with the opener. Even if you choose to not include the series vs Jaedong, 11-8 is still pretty respectable with a build that carries more than just some risk.
In summary, Flash most certainly does not 'always lose when he does this'. My original question answered, now is the time to move on to more important things (why doesn't Donald Duck wear pants?).
Its awesome that the first super hyped TvZ finals ended in just a few minutes with 3 straight bunker rushes, and what will be remembered as one of the last super hyped starleague finishes ended with 3 fast command centers. That statement alone sums up a nearly a decade of serious Brood War strategy development.
+ Show Spoiler [Credits] +
|
At 14 wins and 8 loses, he has a pretty solid history with the opener.
That's ~64% which is worse than his average (including 14CC), so it's a slightly worse history overall.
It's still a good winrate though, much better than any non-S-class player can hope to achieve in a lifetime on the scene.
|
Baa?21243 Posts
You watched like a minute of 173 TvZs haha...props to you, did you do it all in one go or did you have to do it in chunks?
I'd imagine I'd've died of boredom doing that 
|
On June 20 2010 17:38 Sadistx wrote:That's ~64% which is worse than his average (including 14CC), so it's a slightly worse history overall. It's still a good winrate though, much better than any non-S-class player can hope to achieve in a lifetime on the scene.
yeah, but you never know what the results would be without the 14cc, maybe he is picking the best opening in an otherwise tough situation.
|
|
|
now its time to check jaedong's 5 pool
|
Very interesting. I remember you talking about this on the goat show, lol.
|
I think this goes back to what Idra said about the "calculated risks," it's also why I think Flash is so good now. He took a risky build, and went to see what kind of high pressure situations (PL, or singles tourneys) he could use it in and play well with.
|
15 of the 22 games were against zergs who Flash would have a decent chance to lose against in the late game and could use an eco advantage over (Jaedong, Savior, GGPlay), so it's really not a huge surprise. GGPlay was scary in Daum, it's not surprising Flash felt the need to 14cc.
The games vs Kwanro were of course just trying to get him to ling all-in once he saw it off a 12hatch. ^^
|
so effectively 159 total. I then watched all of them long enough see what his first non-depot building was
That's dedication and a lot of free time. O.o
Nice read. What I would really like to know is if Flash ever decides to 14cc on the spot due to impulse.
Anybody know Flash's "general" record with 14cc in other matchups? I wasn't there during Flash's tragic period of 14cc love affair.
|
this should be front page^_^
|
So the last ten games with 14cc is 9-1
It's a good build when used right 
|
Canada2480 Posts
I actually LOVE 14CC and 3 hatch before pool and always defend players who do that
seriously... how often is your opponent not 12 hatching as T or not FE as Z
|
Awesome article and an awesome read, thanks
And i love the picture of flash when he owned stork :D
|
On June 20 2010 22:23 swanized wrote:
I actually LOVE 14CC and 3 hatch before pool and always defend players who do that
seriously... how often is your opponent not 12 hatching as T or not FE as Z
The problem with 3 hatch before pool is if they scout you in a timely fashion they just get nexus first so it is about even either way.
|
CA10828 Posts
lol nice writeup
i'm a fan of pie charts too!
|
Yeah! Take that moletrap! (Usually critiques Flash's 14 CC)
|
On June 20 2010 22:23 swanized wrote:
I actually LOVE 14CC and 3 hatch before pool and always defend players who do that
seriously... how often is your opponent not 12 hatching as T or not FE as Z
LOL?
|
that was a great write-up.
thanks heyoka!
|
Roffles
Pitcairn19291 Posts
Pie Charts look like Pacman. =D
Great write up. Always can count on heyoka to deliver them stats.
|
FREEAGLELAND26781 Posts
|
On June 20 2010 17:38 Sadistx wrote:That's ~64% which is worse than his average (including 14CC), so it's a slightly worse history overall. It's still a good winrate though, much better than any non-S-class player can hope to achieve in a lifetime on the scene.
That's how good Flash is
Even that's considered bad
What a great player
|
Great read, love the charts. You should do more articles like this, it's nice to see someone choose a specific niche in the game and analyze it so deeply. Most people (including me) focus to much on general things in the game and make assumptions instead of doing research.
Front page worthy IMO
|
On June 20 2010 17:54 Carnivorous Sheep wrote:You watched like a minute of 173 TvZs haha...props to you, did you do it all in one go or did you have to do it in chunks? I'd imagine I'd've died of boredom doing that
Awesome - puts to rest a lot of theorizing.
|
On June 20 2010 17:38 Sadistx wrote:That's ~64% which is worse than his average (including 14CC), so it's a slightly worse history overall. It's still a good winrate though, much better than any non-S-class player can hope to achieve in a lifetime on the scene.
But let's not forget, some of these games are "old", so it might not be reflective of his current skills with a 14 CC build.
|
On June 20 2010 17:38 Sadistx wrote:That's ~64% which is worse than his average (including 14CC), so it's a slightly worse history overall. It's still a good winrate though, much better than any non-S-class player can hope to achieve in a lifetime on the scene.
I dare his 14cc stats, despite being below his average, have in fact pushed his other stats higher. What I'm trying to say is that having this strat in his reportoire makes it tougher for his opponents to prepare against him, so it's not as easy as saying he'd have higher overall % if he never went 14cc.
|
On June 20 2010 17:38 Sadistx wrote:That's ~64% which is worse than his average (including 14CC), so it's a slightly worse history overall. It's still a good winrate though, much better than any non-S-class player can hope to achieve in a lifetime on the scene.
take a look at his last ten 14cc games. 9-1. he doesn't just blindly try 14cc like he used to when he lost many times becouse of it.
|
<3 Pacman. There has to be a reason why LYH continually uses 14CC. I wonder what his stats are with 1rax expo.
|
Considering that his overall winrate is (or rather was if it's dropped now after all of the proleague ace losses) above 70% then a 64% winrate is obviously detracting from his overall ability to win the game.
Although he has much better results now with it since the fact that he's been switching up his builds more as noted by the 9-1 in past 10 games. So strategic use is a good thing it looks like.
|
On June 21 2010 05:50 eshlow wrote:
Considering that his overall winrate is (or rather was if it's dropped now after all of the proleague ace losses) above 70% then a 64% winrate is obviously detracting from his overall ability to win the game.
Although he has much better results now with it since the fact that he's been switching up his builds more as noted by the 9-1 in past 10 games. So strategic use is a good thing it looks like.
Wait what?
Did you check vs what players he used it?
64% is way more than the expected winrate vs most of those players.
|
<3 Charts!
I really think 14CC is a good opening, tobad it failed so much vs Effort, but awesome anyhow!
|
Thanks for the pie charts!! Makes everything better!
|
On June 21 2010 06:14 shalafi wrote:Show nested quote +On June 21 2010 05:50 eshlow wrote:
Considering that his overall winrate is (or rather was if it's dropped now after all of the proleague ace losses) above 70% then a 64% winrate is obviously detracting from his overall ability to win the game.
Although he has much better results now with it since the fact that he's been switching up his builds more as noted by the 9-1 in past 10 games. So strategic use is a good thing it looks like. Wait what? Did you check vs what players he used it? 64% is way more than the expected winrate vs most of those players.
Was about to say this.
He doesn't just randomly do a 14cc build at random intervals between all leagues.
|
On June 20 2010 22:47 LosingID8 wrote:
lol nice writeup
i'm a fan of pie charts too!
Who isn't a fan of pie charts? Although then again whenever I look at one I see an actually pie...
Anyways, nice writeup! Now if someone could tell me why this isn't in the News section, I would be thankful.
I like pie,
3FFA(who else?!?)
|
On June 21 2010 06:14 shalafi wrote:Show nested quote +On June 21 2010 05:50 eshlow wrote:
Considering that his overall winrate is (or rather was if it's dropped now after all of the proleague ace losses) above 70% then a 64% winrate is obviously detracting from his overall ability to win the game.
Although he has much better results now with it since the fact that he's been switching up his builds more as noted by the 9-1 in past 10 games. So strategic use is a good thing it looks like. Wait what? Did you check vs what players he used it? 64% is way more than the expected winrate vs most of those players.
I didn't see any analysis of expected wins vs those players.
I do agree with that the player list is harder though, but that doesn't necessarily mean much in some cases.
Conversely, it's over such a long period of time that it's going to be really hard to compare expect win rates vs those certain players.
Basically, it is what it is. Flash tends to use it more when he is less confident he can win. Sometimes it backfires against early pressure (such as against Effort in OSL..., and protoss in general for a while there), and sometimes it works magnificently.
I wonder how many times he actually wins in management games when he goes 14cc. That may be an interesting stat to look at.
|
On June 21 2010 05:19 kemoryan wrote:Show nested quote +On June 20 2010 17:38 Sadistx wrote:At 14 wins and 8 loses, he has a pretty solid history with the opener. That's ~64% which is worse than his average (including 14CC), so it's a slightly worse history overall. It's still a good winrate though, much better than any non-S-class player can hope to achieve in a lifetime on the scene. I dare his 14cc stats, despite being below his average, have in fact pushed his other stats higher. What I'm trying to say is that having this strat in his reportoire makes it tougher for his opponents to prepare against him, so it's not as easy as saying he'd have higher overall % if he never went 14cc.
Please don't forget this. If you don't work unconventional strats and rushes into your general gameplay, your opponents get to cut corners and run optimal builds against you 100% of the time. Conversely, if you are known for enjoying unconventional/risky builds, your opponents will often run less optimal builds to compensate, giving you an advantage the rest of the time when you play standard.
|
great read, thumbs up!
but the 9-1 stats in his last 10 14cc games are kinda deluding as this one loss has cost him the osl title and the golden mouse.
but with flashs monster macro, if he can pull off a 14cc and his opponent fails to successfully counter it with either an early attack or some kind of eco cheese on his own, then its basically a won game for flash. does any1 know a game where flash went 14 cc and his opponent did not go 14cc, 3 hatch or 12nex on his own and still won?
|
Really nice writeup, thanks for puttin your time and effort to do this.
Love the Pie charts <3.
Thanks a lot for doing this.
|
On June 21 2010 09:21 Black Gun wrote:
great read, thumbs up!
but the 9-1 stats in his last 10 14cc games are kinda deluding as this one loss has cost him the osl title and the golden mouse.
but with flashs monster macro, if he can pull off a 14cc and his opponent fails to successfully counter it with either an early attack or some kind of eco cheese on his own, then its basically a won game for flash. does any1 know a game where flash went 14 cc and his opponent did not go 14cc, 3 hatch or 12nex on his own and still won?
I'm assuming you're saying 3hatch no pool because 12hatch 11pool 13hatch starts gets pretty behind 14cc. Jaedong won against Flash on Loki 2 with a normal 3hatch build.
|
Very entertaining intro & writeup haha.
But gotta give you props for watching every. single. one. of Flash's TvZs holy smokes. I would have died of boredom. Thank you for the work - so apparently Flash is pretty good with 14cc?
|
If he's playing his strategies perfectly unpredictably at the right ratios he'll have the same chance of winning no matter which he actually uses on any given game
|
On June 21 2010 10:48 Muirhead wrote:
If he's playing his strategies perfectly unpredictably at the right ratios he'll have the same chance of winning no matter which he actually uses on any given game
This only applies if his opponents are playing perfectly as well.
|
|
|
On June 21 2010 11:19 Tianx wrote:Show nested quote +On June 21 2010 10:48 Muirhead wrote:
If he's playing his strategies perfectly unpredictably at the right ratios he'll have the same chance of winning no matter which he actually uses on any given game This only applies if his opponents are playing perfectly as well.
Nah it only assumes he perfectly understands his opponents' imperfections 
|
|
|
On June 21 2010 12:25 Muirhead wrote:Show nested quote +On June 21 2010 11:19 Tianx wrote:On June 21 2010 10:48 Muirhead wrote:
If he's playing his strategies perfectly unpredictably at the right ratios he'll have the same chance of winning no matter which he actually uses on any given game This only applies if his opponents are playing perfectly as well. Nah it only assumes he perfectly understands his opponents' imperfections
If you're exploiting your opponents' mistakes, your play is by definition no longer unpredictable. An optimal balancing of strategies does lead to equal winning percentages for each strategy, but optimal means that you've maximized your winning percentage against perfect play. If your opponents make mistakes, then tweaking the frequency of your strategies will increase your winning percentage.
|
This was interesting. Maybe I should practice this build-order.
The graphs were nice but descriptive labels would have been helpful. It's still an informative post. Thanks!
|
On June 20 2010 17:35 heyoka wrote:
Its awesome that the first super hyped TvZ finals ended in just a few minutes with 3 straight bunker rushes, and what will be remembered as one of the last super hyped starleague finishes ended with 3 fast command centers. That statement alone sums up a nearly a decade of serious Brood War strategy development.
Some nitpicking: the triple-bunker rush you're referring to wasn't in the finals, but the semi-finals of the '04 EVER.
|
On June 21 2010 12:41 Tianx wrote:Show nested quote +On June 21 2010 12:25 Muirhead wrote:On June 21 2010 11:19 Tianx wrote:On June 21 2010 10:48 Muirhead wrote:
If he's playing his strategies perfectly unpredictably at the right ratios he'll have the same chance of winning no matter which he actually uses on any given game This only applies if his opponents are playing perfectly as well. Nah it only assumes he perfectly understands his opponents' imperfections If you're exploiting your opponents' mistakes, your play is by definition no longer unpredictable. An optimal balancing of strategies does lead to equal winning percentages for each strategy, but optimal means that you've maximized your winning percentage against perfect play. If your opponents make mistakes, then tweaking the frequency of your strategies will increase your winning percentage.
Nah you can optimize the frequency of your strategies against known imperfect play too, also leading to equal winning percentages for each strategy.
|
On June 21 2010 22:12 Muirhead wrote:Show nested quote +On June 21 2010 12:41 Tianx wrote:On June 21 2010 12:25 Muirhead wrote:On June 21 2010 11:19 Tianx wrote:On June 21 2010 10:48 Muirhead wrote:
If he's playing his strategies perfectly unpredictably at the right ratios he'll have the same chance of winning no matter which he actually uses on any given game This only applies if his opponents are playing perfectly as well. Nah it only assumes he perfectly understands his opponents' imperfections If you're exploiting your opponents' mistakes, your play is by definition no longer unpredictable. An optimal balancing of strategies does lead to equal winning percentages for each strategy, but optimal means that you've maximized your winning percentage against perfect play. If your opponents make mistakes, then tweaking the frequency of your strategies will increase your winning percentage. Nah you can optimize the frequency of your strategies against known imperfect play too, also leading to equal winning percentages for each strategy.
not necessarily. it depends on how exactly your opponent is playing imperfectly. if he is particularly weak against early aggression, you should increase the amount of early pressure builds to obtain the maximum win rate against this particular opponent.
|
Pacman is throwing up 
|
On June 21 2010 22:26 Black Gun wrote:Show nested quote +On June 21 2010 22:12 Muirhead wrote:On June 21 2010 12:41 Tianx wrote:On June 21 2010 12:25 Muirhead wrote:On June 21 2010 11:19 Tianx wrote:On June 21 2010 10:48 Muirhead wrote:
If he's playing his strategies perfectly unpredictably at the right ratios he'll have the same chance of winning no matter which he actually uses on any given game This only applies if his opponents are playing perfectly as well. Nah it only assumes he perfectly understands his opponents' imperfections If you're exploiting your opponents' mistakes, your play is by definition no longer unpredictable. An optimal balancing of strategies does lead to equal winning percentages for each strategy, but optimal means that you've maximized your winning percentage against perfect play. If your opponents make mistakes, then tweaking the frequency of your strategies will increase your winning percentage. Nah you can optimize the frequency of your strategies against known imperfect play too, also leading to equal winning percentages for each strategy. not necessarily. it depends on how exactly your opponent is playing imperfectly. if he is particularly weak against early aggression, you should increase the amount of early pressure builds to obtain the maximum win rate against this particular opponent.
Yeah of course... the ratios with which you use each strategy should vary by opponent. A statistically interesting study would be to take two players and have them play 100 games in a very short time frame. You should find the winning rate to be nearly identical no matter what strategies either uses in any given game. I have no doubt that Flash and his coaches are playing near perfect random-based strategies. Human intuition about such things has been proven to be surprisingly good.
|
good gawd... 159 replays. t-up for effort. (your hard work, not that lousy zerg)
that being said, he sure doesnt 14cc as much as i thought he does.
|
thanks for taking the time to inform us of these interesting stats. Im sure one of his teams coaches has all these stats and more too.
|
On June 20 2010 22:12 Piste wrote:So the last ten games with 14cc is 9-1It's a good build when used right
This is an important point. Weren't his 14cc losses to GGPlay awhile ago?
|
|
|
|
|
|
EPT![[image loading]](/staff/heyoka/depth/flash0.jpg)
![[image loading]](/staff/heyoka/depth/flash1.png)
![[image loading]](/staff/heyoka/depth/flash2.png)
![[image loading]](/staff/heyoka/depth/flash3.png)
![[image loading]](/staff/heyoka/depth/flash4.png)
