|
Here are some Code S Week 1 stats featured in my Code S Recap over on ESFI:
Zergs have sucked at counter-picking maps. They've won only 3/13 games on maps they have picked against the other two races. Protoss are 7/10 and Terran are 9/16. List of picked maps. (Feel free to check my work on this one as I was really tired when I compiled it)
There was a nice separation on how different tiers of players did against the others. Ro8/Code A/UpAndDown/Special Invite went 62%/54%/36%/11% in win percentage against the other tiers. And it went 75%/67%/20%/0% in terms of getting players through to the next round.
There was only one group out of eight that had a rematch between in the final match (that awkward best-of-two). The average has been four out of eight over the past four seasons.
There have been many more comebacks in these bests-of-three compared to the Ro16 bests-of-three from other Code S seasons, at least as a percentage of 3-game-matches. Over all previous Code S seasons, the winner of the first game has won three-game-matches 60% of the time. In this group stage, they only won 44% of the time. Best guess is that, if this stat is significant, map-picking must be making comebacks a bit easier when two opponents are equally matched.
The is the fourth Code S in a row where the Ro16 is at least half Terran. By really quick estimation I'd say the chance of any one race randomly nabbing at least eight spots (assuming independence yada yada) is around 36%, so the chances of it happening four times in a row should be around 1.5%. If anyone wants to do a more deliberate calculation I'd love to see it.
Group H was just the fourth ever Code S group to have two Protoss players take first and second place. The three others were Choya/Tester in January, Tester/Anypro in March, and Huk/Puzzle in November. It has happened only once for Zerg, when Losira and Coca got first and second back in July. It has happened twenty-two times for Terran, however, largely due to the surfeit of Terran players in the GSL.
|
On January 03 2012 05:53 pullarius1 wrote:
Well... that was short :-/ You'll still be able to find my stuff on ESFI though.
Anyway, a quick stat: over ALL the individual SC2 tournaments Gom has run (So All Code S, Code A, Opens and special tournaments) only 5 Protoss players have ever made any finals: MC (Sx2), Inca (S), Puzzle(A), Oz(A), and Tassadar(A).
Over 22 tournaments there have been 44 finalists. Of those, 24 have been Terran, 14 have been Zerg, and 6 have been Protoss. Or, by percentage, a 55/32/14 split, meaning that Terrans have had four times more finalists than have Protoss players.
Unfortuante. I am starting to question whether Sundance know what's he doing, when layoff comes a month after a hiring spree.
|
On January 26 2012 04:15 Primadog wrote:
Unfortuante. I am starting to question whether Sundance know what's he doing, when layoff comes a month after a hiring spree.
Hopefully something else will come up. :-/
One more thing: I was very interested going into the new, Ro16 group selection ceremony as to whether players would pick opponents they had already beaten in the Ro32 for their groups, since obviously they should feel confident in their chances against just-beaten opponents. The way the groups turned out, though, only two players even had the opportunity to pick players they had already beaten. Ganzi picked last season's champion Jjakji, whom he had 2-0'd in the Ro32. Oz, on the other hand, declined to pick Curious, whom he had also 2-0'd.
|
I compiled a list of winnings at MLGs last year to compare them to how well the players did in terms of points v winnings. The discrepancies are pretty large in some places. The top ten in points went 2, 5, 8, 3, 7, 6, 21, 4, 27,11 in terms of winnings, while the top 10 in winnings were 16, 1, 4, 8, 2, 6, 5, 3, 12, 11 in points.+ Show Spoiler [Graph] +
The 21 and 27 in the first list come from Slush and Ret, who only earned $1000 and $500 this year respectively. The 16 on the second list, of course, comes from Leenock, who managed to top the winnings of every other player with just a single win. It is worth noting that Leenock's winnings doubled all but Naniwa's, while Naniwas'a winnings doubled all but DRG's, thanks to the incredibly top-heavy prize distribution. There will be an article out on ESFI soon that mentions some of this, and I'll link it here when it is published.
|
Did some quick calculating based on a snippet of SotG discussion from last night. When a Code S group is XYYY in terms of race distribution, and so the X has only one matchup to prepare for while the Y's have two, the X is 13/22 (59%) for making it out of the group. Interestingly, though, the X is far more likely to get 2nd place than third, something I think we can attribute to the Y's almost always being three terrans. Also, I don't think this sample is quite big enough to say anything definite.
When a group is XXYZ, so X has three matchups compared to he two of Y and Z, though, the X still makes it out 36/66 (55%) of the time. Again, because that XX is so often Terran, I think the most likely conclusion is that 2 v 3 matchups does not matter all that much. Also, none of this take into consideration that groups can often be somewhat predicted, and that it is still somewhat unlikely that a player will even play all three races in a 3-race group.
|
Holy shit. How does this not have a billion posts. Screw this I'm tweeting immediately. Absolutely brilliant piece.
|
On January 28 2012 06:35 Day[9] wrote:
Holy shit. How does this not have a billion posts. Screw this I'm tweeting immediately. Absolutely brilliant piece.
Well, how come I never saw this thread before? I spend HOURS in TL and never saw it until Sean tweeted about it.
TY for the Tweet, made me discover this masterpiece. One of the best bumps of, like, EVER.
Brilliant. Absolutely brilliant. Not to mention itis TOTALLY right! ._.
|
Mmm. I believe Fruitdealer lost his unofficial title in the round of 32 of open season 2 not round of 64 as it says :D
Just in case you care, and wish to fix it.
Was fun to read though!
|
|
|
United Kingdom14464 Posts
|
Wow, this is some really interesting stuff here. Thank you for writing it and thank Day[9] for tweeting about it!
|
Very interesting but sad Blizzard doesn't release their statistics because it doesn't say why there's so many terrans..is it just because
a)many more people play terrans to start with ?
b)they're just better players
c)the builds have been perfected more than with the other races ?
d)mvp fucks up the statistics chuck norris style
e)race is slightly better?
It'd be interesting to know what's the averaqe length of those games as well (with the detail for each matchup) if you ever study that, just to see if there's a pattern there ....I always get the feeling that T is stronger in the early game (banshees, mmm pushes, bunker rushes), Z mid game (roach pushes, early pools, mutalings) and P late game (deathball with right unit mix), so T would have an advantage there just by being able to end more games earlier but that's just a hunch from all those games I've watched, stats would be interesting....
Whatever the reasons for this may be, if you're a Zerg or Protoss pro and wanna qualify for the GSL, seeing those stats must make you wish you had played terran from the start.
|
On January 28 2012 06:48 UmbraaeternuS wrote:Show nested quote +On January 28 2012 06:35 Day[9] wrote:
Holy shit. How does this not have a billion posts. Screw this I'm tweeting immediately. Absolutely brilliant piece. Well, how come I never saw this thread before? I spend HOURS in TL and never saw it until Sean tweeted about it. TY for the Tweet, made me discover this masterpiece. One of the best bumps of, like, EVER. Brilliant. Absolutely brilliant. Not to mention itis TOTALLY right! ._.
LURK MOAR
|
On January 28 2012 06:35 Day[9] wrote:
Holy shit. How does this not have a billion posts. Screw this I'm tweeting immediately. Absolutely brilliant piece. Thanks! And thanks to the other kind words as well :-)
|
This thread is awesome. Thanks for putting it all together 
|
Wow, amazing man - i just saw this and am like.. wow! This is very intresting to read and while following the GSL quite alot. This is sooo cool to see the stats about the players, tournys.. man. I am for sure gonna read all off it when I get home.
Keep up the awesome work.
|
I missed all the january updates, thanks a lot as always, please keep up doing this !
|
Some stats I compiled about last season's GSTL for this article:
The matches are extremely streaky, which is probably to be expected. Only 31% of game wins came from single-game wins, whereas a full 48% of all game wins came as part of streaks that were 3 or longer.
Also, more games were ended by 4-kill streaks (37%) than by 3-kill streaks, 2-kill streaks, or even single-game wins.
Finally, if you win a game in the GSTL, you are 54% likely to win the next. If you win 2 in a row, you'll win a third 63% of the time, and if you win three you have a 62% chance of finishing out the 4-kill.
One thing that annoys me in the GSTL is when a team will send out two of the same race in a row. It seems absurd and only a good idea in a very limited set of circumstances, but it happens all the time. It turns out that the second player of the same race in a row only wins 31% of the time :-/ One confounding factor here was SlayerS, since they sent out Terrans 86% of the time, which forced the statistic towards their team average. Taking into account just Group B and the Playoffs, that second player of the same race won only 2 out of 18 times (%11)
|
wow giantic list of info, susprised this doesn't get more dicussion
|
On January 31 2012 09:32 pullarius1 wrote:Some stats I compiled about last season's GSTL for this article: The matches are extremely streaky, which is probably to be expected. Only 31% of game wins came from single-game wins, whereas a full 48% of all game wins came as part of streaks that were 3 or longer. Also, more games were ended by 4-kill streaks (37%) than by 3-kill streaks, 2-kill streaks, or even single-game wins. Finally, if you win a game in the GSTL, you are 54% likely to win the next. If you win 2 in a row, you'll win a third 63% of the time, and if you win three you have a 62% chance of finishing out the 4-kill. One thing that annoys me in the GSTL is when a team will send out two of the same race in a row. It seems absurd and only a good idea in a very limited set of circumstances, but it happens all the time. It turns out that the second player of the same race in a row only wins 31% of the time :-/ One confounding factor here was SlayerS, since they sent out Terrans 86% of the time, which forced the statistic towards their team average. Taking into account just Group B and the Playoffs, that second player of the same race won only 2 out of 18 times (%11)
I wonder if we can develop a statistical model to determine how far from a RNG this streakiness is. Based on my limited research, this type of streakiness do not exist in 1v1 bo3s.
+ Show Spoiler [all wrong, need redo] +Let's suppose every player sent out has an equal chance of winning their match. - Then 25% of game wins came from single-game wins, whereas 37.5% of all game wins came as part of streaks that were 3 or longer.
- 4-kill streaks will 12.5%, 3-kill streak 12.5%, 2-kill stream 25%, 1-kill 50%.
- If one win a game in the GSTL, you are 50% likely to win the next. If you win 2 in a row, you'll win a third 50% of the time, and if you win three you have a 50% chance of finishing out the 4-kill.
Let's suppose, instead, that there're a two tier system: I. 20% players win 70% of the time II. 80% players win 45% of the time (a rough model of GSTL S1 http://wiki.teamliquid.net/starcraft2/2011_Global_StarCraft_II_Team_League_Season_1/Statistics#MVP ) [list][*] Then 24% of game wins came from single-game wins, ... line one above seems wrong, gonna redo my math a little
|
|
|
|
|
|
EPT
