|
With all the recent talk about GSL matches being very short, and maps being imbalanced, I decided to take a look all the games, and do some number crunching. So here's what I've got so far:
Match Durations
First, I wanted to check how long each game in GSL 4 lasted, and what the average time for the matches was. I plan to do this also for all previous GSL, so we can maybe see how the games have evolved (have games just recently started becoming shorter and shorter due to all of the cheese and all-ins, or was it like that from the start, or maybe the games are actually not that short at all...).
The process takes a bit of time, because I need to check each VOD on gomtv.net, and mark the in game clock time at the moment of GG. I then converted that in game time to real time using a factor of 1,38 (http://wiki.teamliquid.net/starcraft2/Game_Speed). I also took the opportunity to write down the number of views for each VOD.
M1 xx:xx is the number of the match and the in game time which was then converted into real time.
![[image loading]](http://i51.tinypic.com/25sse91.jpg)
The average time of all GSL 4 matches comes down to 12:13 minutes.
GSL Season 4 - In progress
![[image loading]](http://i54.tinypic.com/2eyg76q.jpg)
GSL Season 3
162 games were played in GSL Season 3.
Average game length was 11:23 minutes.
Average views on VODs is currently at around 69 000.
Map stats:
![[image loading]](http://i54.tinypic.com/15xl2tx.jpg)
Player stats (average match duration):
![[image loading]](http://i55.tinypic.com/iy2cn5.jpg)
GSL Season 2
Will be updated soon.
Race Win Percentages
I helped myself with Liquipedia here (such an awesome tool, doesn't get enough love I think). Hopefully all the numbers are correct, there was a lot to keep track of, and some mistakes might have slipped, but hopefully very few of them, as I did double check everything.
![[image loading]](http://i54.tinypic.com/2i1ffyc.jpg)
From looking at each GSL independently, it doesn't seem like one could try to come to some conclusions, as the win ratios are very hectic, and don't seem to follow a pattern - mostly because the number of games is too small to properly present match statistics.
But when looking at all the games played so far in 4 GSL tournaments, it's starting to look a bit better:
![[image loading]](http://i52.tinypic.com/1fg203.jpgg)
From the image above you can see the ratios per map (didn't want to calculate the percentage, as I feel the number of played games is still too low to go and represent it with a hard percentage. It's interesting to see stats PvT on Delta Quadrant - thank god that map is soon out of the map pool.
Hopefully someone will find all of this useful.
Cheers.
|
Good stuff man, thanks for posting!
|
Thanks for getting these stats up. I'm curious about match length depending on the map being played, which shouldn't be that hard to find since you already have length for each game up.
|
Thx for posting but i kind of hate shit like this cause some retard will come in here screaming about balance.
|
|
|
quite funny how at the time when the "OMG ZERG UNBEATABLE!" outcries started again gsl3 kicked around and shut it totally down.also how poor P performance in gsl2 was.
these stats are exactly what my "feeling" was. good to see im right ~~
|
On January 09 2011 23:28 Joroth wrote:
Thx for posting but i kind of hate shit like this cause some retard will come in here screaming about balance.
Perhaps when GSL starts to use big macro maps we will get some balance patches that are actually needed instead of band-aid for garbage maps like blistering sands.
|
Sweden4672 Posts
On January 09 2011 23:28 Joroth wrote:
Thx for posting but i kind of hate shit like this cause some retard will come in here screaming about balance.
So when you see a thread named "GSL Match Statistics", don't click it.
|
|
|
|
|
Dominican Republic913 Posts
nice stats keep them coming, can u please Update this every Round in GSL?
i mean, the Ro16 started, add the statics when Ro8 starts.
|
On January 09 2011 23:38 Xism wrote:
PvZ really has changed.
How? You have different veto systems for maps over the seasons and the results are heavily influenced by single players, you can't tell if anything has changed.
|
On January 09 2011 23:45 clusen wrote:How? You have different veto systems for maps over the seasons and the results are heavily influenced by single players, you can't tell if anything has changed.
Yeah, I guess you're right. Was looking too blindly at the statistics
|
Canada5565 Posts
Wow good stuff. Thank you for putting this together 
|
Statistics are nice but it's very low scale of played games. I mean in GSL4 PvZ is 80%. but only 5 games were played.
|
It kind pains me when I get one vod to watch
14minutes ..
"ok nice game..."
then I skip tartosis' comments etc etc
5-6 minutes of games left (7-8 in game)
I go fuuuuuuuuuuuu
|
On January 09 2011 23:53 Alpina wrote:
Statistics are nice but it's very low scale of played games. I mean in GSL4 PvZ is 80%. but only 5 games were played.
That's why the OP also provide GSL total
|
What if Fruitdealer and Nestea weren't around to keep those first 2 gsl's stats up?
You can look at the statistics from any perspective and make a balance statement that opposes the other.
Watch: Wow, 80% winrate for toss vs zerg in GSL4? 71% winrate for terran vs zerg? Looks like zerg is having a hard time, what would we do without 'X' zerg.
This doesn't prove anything.
Stop looking at the race, when it comes to GSL and start looking at the players.
|
On January 09 2011 23:32 BeMannerDuPenner wrote:
quite funny how at the time when the "OMG ZERG UNBEATABLE!" outcries started again gsl3 kicked around and shut it totally down.also how poor P performance in gsl2 was.
No. Don't go there if you can't even remember what happened in GSL2. Most of the Protoss players were trying builds lower ranked players wouldn't even try if being serious to win. The rest made continually poor choices that was out of character when compared to season 1 and 3 level of unforced errors.
|
I like having these numbers! Very well done.
Delta Quadrant seems a lot worse than I would have predicted.
Lost Temple and Jungle Basin seems to be on the verge of being bad maps due to PvT balance for Lost Temple and in generel for Jungle Basin.
Metalopolis and Xel'Naga Caverns seem very well balanced and has a lot of games to them.
Blistering Sands and Steppes of war seems to be better than their reputation.
If you remember to log the maps and matchups in the same spreadsheet as the game-times it would bring even more interesting stats, such as slowest matchups, slowest maps and it could tell something about if the strategies in different matchups are diverse when enough data has been accumulated!
|
These are meaningless statistics as all skill isn't equal. Some of the players are absolute garbage compared to others.
|
On January 10 2011 00:07 mutantmagnet wrote:Show nested quote +On January 09 2011 23:32 BeMannerDuPenner wrote:
quite funny how at the time when the "OMG ZERG UNBEATABLE!" outcries started again gsl3 kicked around and shut it totally down.also how poor P performance in gsl2 was.
No. Don't go there if you can't even remember what happened in GSL2. Most of the Protoss players were trying builds lower ranked players wouldn't even try if being serious to win. The rest made continually poor choices that was out of character when compared to season 1 and 3 level of unforced errors.
exactly what i meant. poor performance of the P players. i do remember the terrible 1 cannon FEs on xelnaga&scrap and the half assed bad executed 4gates etc ~
On January 10 2011 00:14 hmunkey wrote:
These are meaningless statistics as all skill isn't equal. Some of the players are absolute garbage compared to others.
yep but when you have seen all/most of the actual games its nice to see some numbers behind the trends you already noticed.
|
On January 09 2011 23:26 teamsolid wrote:
Thanks for getting these stats up. I'm curious about match length depending on the map being played, which shouldn't be that hard to find since you already have length for each game up.
Oh, that's a good idea, I basically started doing this just for match durations, but somehow digressed into match statistics.
Now that I have a spreadsheet done, I'll keep track of new GSL matches and update the thread with match data after every round.
And people please don't take match statistics too seriously, as we all know the number of games is too low to exclude all the other contributing factors like individual players, map veto systems and other stuff.
The focus will be on match durations, I'm really interested how the numbers break down for each map.
|
Nice stuff. Its sad to see there are so many short games... It will be nice when they remove Delta Quadrant, Blistering Sands, Steppes of War and Jungle Basin
|
On January 10 2011 00:14 hmunkey wrote:
These are meaningless statistics as all skill isn't equal. Some of the players are absolute garbage compared to others.
Just because the skill isn't equal doesn't mean that these are meaningless statistics.
If you consider that there is an approximately equal ratio of very good/bad Terran/Protoss/Zerg, and that these people are equally matched in the 3 MU, then the statistics are valid.
|
very nice job here tnx :D
|
On January 10 2011 00:14 hmunkey wrote:
These are meaningless statistics as all skill isn't equal. Some of the players are absolute garbage compared to others.
Uhh it is a nice indication of map balance and a little insight into the match-ups though.
This is how people did it in Brood War and it was pretty undisputed fact when one map had a huge racial advantage through proleague stats.
I know SC2 is new and all and not everyone is at that level where they are close in skill, but I'd rather look at these statistics then read posts on the strategy forum concerning racial imbalances from low diamond players any day. This bandwagon of "Protoss ez" is sort of annoying, and yeah, Protoss is "ez" if you're playing against terrible players. But I think whether a race is hard or not is sort of subjective, although statistically GSL seems to be that of the opposite opinion of a good portion of TL members.
Edit: Kind of like how Artosis and IdrA want to say how Zerg is the hardest, and as IdrA said many times how Terran and Protoss don't really take skill - but were terrible Terran/Protoss players in beta. How ironic, huh?
|
On January 10 2011 00:44 TeWy wrote:Show nested quote +On January 10 2011 00:14 hmunkey wrote:
These are meaningless statistics as all skill isn't equal. Some of the players are absolute garbage compared to others. Just because the skill isn't equal doesn't mean that these are meaningless statistics. If you consider that there is an approximately equal ratio of very good/bad Terran/Protoss/Zerg, and that these people are equally matched in the 3 MU, then the statistics are valid.
I think he is right. We needed way more games played to come up with valid statistics due to the fact that the skill difference among the players is huge.
|
Thanks for the info. There are an astounding number of under 10 minute matches, and having an average match length of 12 minutes is certainly very worrisome. No wonder viewership ratings are dropping! They just aren't very long or epic. I don't think enough action can be packed into 12 minutes for an RTS.
But, to tell the truth, it's not only just about the game length but also the quality of engagements and the control over the map. Games will only get interesting once we see a minimum of 3-4 bases each and at least 5 facs operational. Maybe I've been spoiled by SWL, but there we are.
|
woot nice to see the progress that protoss has made stats-wise throughout the 3 main seasons... really got dominated hard the first one and most of the second.... gogo toss!
|
great info, thx. but the statistics are severly skewed by the fact that all 3 gsls were won by players of a race that wasnt doing good overall at that time and didnt have much more presence in the later rounds of the tourney than the respective winner.
it also doesnt reflect major metagame shifts. nestea was able to overcome foxers 2rax aggressions in the gsl2 finals because foxer hadnt refined the strategy yet, but the seed was sown and the refined 2 rax strategies then went on to roflstomp zergs out of gsl3.
|
Interesting stats on Kulas Ravine. 100% zerg winrate (albeit 6 games) are these all Fruit Dealer games? I seem to remember Tastosis saying he was the only zerg who didn't veto Kulas.
|
Yeah, I think those are all FD's wins.
|
Calgary25968 Posts
Super interesting to me. Thanks!
|
|
|
Also lol @ 5-17 PvT on DQ ..... just wow.
|
It seems the shorter games overall get less views. HMMM
|
Somehow I doubt TvZ for season 1 is only 45%
|
Sweden33719 Posts
On January 10 2011 01:12 shadymmj wrote:
Thanks for the info. There are an astounding number of under 10 minute matches, and having an average match length of 12 minutes is certainly very worrisome. No wonder viewership ratings are dropping! They just aren't very long or epic. I don't think enough action can be packed into 12 minutes for an RTS.
But, to tell the truth, it's not only just about the game length but also the quality of engagements and the control over the map. Games will only get interesting once we see a minimum of 3-4 bases each and at least 5 facs operational. Maybe I've been spoiled by SWL, but there we are.
Eh 12 isnt so bad, avg bw was 15 I think.
Just needs less of th e 5 minute games and more evenly around 12-14
|
Here's the map breakdown for GSL 4, I plan to do this for other seasons as well, but it's talking a lot of time, I've just done all Ro64 matches for GSL3 and moving onto Ro32
Will keep the OP updated as I get new info
|
Which GSL season belongs to which balance patch?
|
That PvT statistic in the 3rd season is so deceiving. ><" MC basically contributed to like half those wins.
|
On January 09 2011 23:21 Odoakar wrote:As I always knew, Protoss is having the hardest time, imagine if there wasn't MC around to keep those stats up 
The same counts for Fruitdealer and IMNesTea in GSL1 and 2. I have the feeling the amount of matches played aren't enough and certain players are solely skewing the statistics.
Tho once the match count goes up it should give a clearer and more statistically accurate image and I praise/thank you for starting this up. Especially the average game time was an interesting tho not very surprising value.
|
Wow thanks for these numbers!
Question, what was the average length of SC1 matches? (perhaps not throughout all 12 years ofc, but recently, what is it around? like in the korean pro level).
Thanks in advance.
|
On January 10 2011 04:37 Nyxs wrote:Show nested quote +On January 09 2011 23:21 Odoakar wrote:As I always knew, Protoss is having the hardest time, imagine if there wasn't MC around to keep those stats up The same counts for Fruitdealer and IMNesTea in GSL1 and 2. I have the feeling the amount of matches played aren't enough and certain players are solely skewing the statistics. Tho once the match count goes up it should give a clearer and more statistically accurate image and I praise/thank you for starting this up. Especially the average game time was an interesting tho not very surprising value.
With the amount of significant balance patches through out the GSL seasons, we will have to wait until the patch frequency goes down before this happens. Comparing matchups with all the seasons' matches combined doesn't make much sense.
|
I dont think there is a big enough sample for these statistics to be useful. For example, according to the stats, PvZ on DQ is in favor of the zerg, TvZ on kulas zerg has 100% win ratio, and PvZ on scrap is in favor of the protoss.
Sc2 is so volatile atm, and builds are changing so quickly that there is too much randomness. One season 1 player will dominate the next 3 seasons he wont even qualify or get far.
Thanks for the stats though
|
Wow, without season 3, which had MC getting pvt the last 3 rounds, pvt would be in the low 30's for percent. Even the percent with GSL 3 included (44%) is pretty bad at a pro level. I'm actually really curious how the TvZ and PvZ numbers would look if all in's were removed from the equation. I've suspected for a while that, while zergs complain about the all ins, it is happening so often because even the pro's feel they have to in order to have a chance on many maps. My honest opinion is that the buff to toss thats coming won't make a big difference.
The real problem in my opinion is how few things most toss units counter (especially gateway units) and how many units counter them. Also, how many key toss units in the early game loose their greatness as the game progresses. Sentries, immortals, stalkers, void rays, get smashed by a lot more stuff than they smash they are all strong early and not as strong mid to late game. Psi storm and colossus are great, but so expensive they are a huge investment. Zergs and Terrans are becoming less and less afraid of them because they are learning when to scout for them and making corruptors and vikings for the colossus, using emp vs HT or just dodging the storms. I play toss and terran, in my experience it is 5x harder to feedback a ghost than it is to dump and emp into a gateway ball. 1 emp hitting 9 units is 900 dmg (100 shields per unit) and no energy for casting and people bitch about storm? Which takes a lot of gas and time to get out. Yet infestors start with fungal and Ghosts start with emp, neither of which are dodgeable.... I think toss macro is awesome, but overall toss has the worst mobility of any race and the weakest harass. Phoenix and DT harass gets shut down by the same thing, turrets or spores. Phoenix major problem is you have to wait for about 3 before you can do any harass with them and then the energy runs out pretty quick. Many people have figured out that when they see phoenix early that your ground army will be much weaker and they go for the kill. I don't think making obs 25 gas cheeper, phoenix build 10 seconds faster, and halucination 80 seconds instead of 110 are going to fix the issues that exist with protoss. Scouting isn't the big issue imo. The units themselves are. It's really hard to do damage vs zerg without a huge amount of risk of loosing the game and terrans units are just much more cost effective vs toss units. The combination of MMM with a sprinkle of either ghost, viking, or tanks is just extremely hard to deal with, yet it easily deals with almost anything toss can make. I think terrans real problem in the late game vs toss is the same as their troubles vs zerg in late game. They just can't replace their armies as fast as the other races. But unlike toss, every terran unit, except BC's, can be used for effective harass in some situation.
Zerg's units are pretty cost effective and they are usually ahead in economy after the 10 minute mark unless you've had some hugely successful push against them. The fact that so many naturals and 3rds are wide open is a good thing for zerg. Roaches, hydras, lings and banlings all have speed upgrades and creep to assist. There extreme mobility means they can deal with defending the key areas of maps much easier than the other races, yet still have the mobility to make huge attacks at any place anytime. The combination of this and their ability to reproduce 100 supply of and army in under a minute in the late game is why people are all ining them so much. The fear this scenario. I understand many will disagree. But watch closely in pro matches, how often zergs secure 3rds and loose the game. It's pretty rare. Usually, toss or terran never secure a 3rd vs zerg unless the zerg lets them and decides to take even more bases going up to 5 base.
Yes, I know it's a bunch of qq and I agree wholeheartedly that toss is the easiest to learn and play in general (zerg being the hardest) . But at high levels it looses its luster against an opponent who understands protoss's weaknesses and scout at the right times.
When these new larger macro oriented maps come into play. I have a feeling pvz is going to plummet for the toss winning percent and pvt will start swinging in Toss's favor. Zerg will start owning terrans and toss's in the months to come. Feel free to disagree and make a well stated, thought out rebuttal that doesn't include a slew of insults and vast generalizations.
|
we'd get a better idea of "balance" if op could post just the ro32 statistics
|
Very interesting, though it would be cool to see how much the outliers effect the average game time. Perhaps median would be a nice statistic to see!
|
Perhaps you could also include standard deviation? I know we could calculate it ourselves but if it was included in the OP next to all the other stats it'd give the average reader a general idea of the variability before they start making some bolder assumptions.
|
Could you also include patch numbers per GSL, so we know the point of shifts?
|
Australia8532 Posts
Excellent number crunching imagine a world without oGsMC - it would be a dull place to live.
The games seem shorter; as a lot of pros have said the Koreans are great at early aggression and they execute this well..
New maps will also affect any future results as any new maps im sure will be more macro oriented.
|
Could you also compute some t-statistics to give a sense of reliability and include at least 15 labelled graphs so that we don't have to look at numbers so much?
On behalf of the often socially retarded Starcraft community, I would like to say thank you, OP, for everything you've done, and please only take the above as suggestions if you ever feel the desire to put even more time into this. You've done so much cool stuff already!
|
Throught the way a tournament with brackets works and the small sample the stats are going to be very skewed. Plus a fruitdealer who i think lost like 1 game total in GSL 1 is going to mess stuff up. These stats however, if preserved are going to get better over time with more matches played.
|
responding to reborn, as a toss player i agree with much of what u said, however i think the patch is going to make the pvz matchup much better and even with macro maps no problem. i have a 60% plus win percent on shakuras and i almost always play a zerg on it, my trick is fast expansion stargate. now i'm gonna admit i rely on an error from zerg players to take a win with it, but now i believe it will be a viable, safe macro build against zerg and here's why. a fast void ray with some cannon at the choke will stop zergs only possible aggressive response to a protoss fast expand, which is a roach bust or a nydus worm, it does this by killing roaches unanswered (obv) and clearing overlords from your base immediately. the problem is what u do with your stargate afterward, i'm keen on a phoenix transition to force hydra and spores and queens, however the problem with this is it forces me to make two stargates if i want an efffective harass. no longer. i can now safely get two base and a void ray, phoenix harass, and have time to transition to turtle n gates or robo tech depending on how he wants to counter, heavy air or hydra, and boom a fair matchup is born, right now one stargate is not enough to beat blind muta massers and phoenix pump to slow to create an effective harass after the void ray which is absolutely essential for toss fast macro safety vs zerg. im only a 2.7 k protoss but i feel i have a very deep understanding of pvz because its pretty much the only matchup im winning so thats my 2 cents, and yes, obviously larger maps will help vs terran because they always blindly run up your ramp or take ur expo out as soon as stim is researched and theres not much toss can do except hope they are late.
edit and expansion: and as for gateway units i think early game dominance and late game weakness is as it should be, and this is my perspective as a night elf player in war3, think about it, the stalker is super fast, ranged, and does extra vs armored, typically the units that "counter" the stalker, it is built much like the huntress, however, as soon as a snare or stun ability comes into play, that expensive, powerful unit is a liability, this is balance, your opponent sees you getting something that is good vs their composition and has a route to counter, protoss has this except its options are typically at tier 3, making the tier 2 scout so important, however the only real viable t2 scout is the observer which is what this patch is for, it is lowering the scouting burden for all 3 tech paths, the problem with this, however, is at the highest levels the protoss tech choice is going to be revealed by their scouting method, pay attention cuz this is some good info. if you see hallucinations, they went twilight council. if you see phoenix, stargate, obs, robo. good protoss should realize this and they should not throw down any t3 tech until that scout has revealed the response that your opponent makes to which scout he sees, i'm not sure if this is good but it is at least an improvement, but i really think the obs should be built from the nexus as soon as tier 2 is reached from stargate, twilight, or robo, because cloak banshee and dark templar threat still force robo in any non-rush game against terran and toss.
|
Should do it like I saw a few weeks ago in the BW forum, the W/L rates WITH Flash/Jaedong and those WITHOUT.
I would say maybe S1 with/without FD, S2 with/without Nestea, S3 with/without MC, etc.
because the winners of the each season all won convincingly....
|
On January 10 2011 04:44 Yoshi Kirishima wrote:
Wow thanks for these numbers!
Question, what was the average length of SC1 matches? (perhaps not throughout all 12 years ofc, but recently, what is it around? like in the korean pro level).
Thanks in advance.
Not too sure but I'm pretty positive it'll be at least 20-25 minutes easy.
|
On January 10 2011 11:32 Silidons wrote:
Should do it like I saw a few weeks ago in the BW forum, the W/L rates WITH Flash/Jaedong and those WITHOUT.
I would say maybe S1 with/without FD, S2 with/without Nestea, S3 with/without MC, etc.
because the winners of the each season all won convincingly....
I actually completely disagree, and think it should go the other way. The game is balanced for the top tier of players, so perhaps we should only look at data from the top 32 or top 16. Or maybe do it like figure skating and remove both ends of the outliers.
Nice post, btub, this was very interesting to read. It amazes me that the current season PvT stats are decided so heavily by Delta Quadrant. Protoss should start vetoing that map!
|
What's the point in converting game time to real time? We all play in faster settings, or at least the majority; so wouldn't it make more sense to have a notation that we can all relate to much easier?
|
I don't think it's entirely accurate to take all the stats of all the GSLs, and compile it into one table. This will not show which race is having the most trouble. The fact that through GSLs, the win loss ratio between all the races have fluctuated dramatically shows that people are playing in different ways every time. Patches also have an effect on this. As the meta-game changes, so does the win-loss ratio of all the races.
Right now if you were to take GSL 4, which is the most recent and therefore the most accurate GSL to be taking statistics from, and put the numbers together, it shows that Zerg is having a really really rough time with the other races.
Of course, you have to take into account the amount of games that have been played, which isn't much at all. As you approach infinity games, the real percentage will present itself, so by the end of GSL we should have a nice percentage of win loss.
|
Wow this is really good stuff. Great number crunching.
|
i don't think we should remove a certain player, because those players are part of the statistic.
and i think DQ games are so damn short because its such an aweful map that no one wants to play on it haha
very nice statistics, when pared with following GSL games we can see how Metagame shifts affect each match up.
|
A lot of it is kinda pointless, but still interesting to see. PvZ pre-GSL um... season 2 I think most Protoss players didn't even know to ramp block so we had mostly FEs vs Protoss rather than the now required pool/gas first. The game is nothing like it was in GSL season 1 so all those stats are kinda dated.
|
Loved how TL juxtaposed this post with the new map post. Great work thanks.
MC damnit. Buff us T_T
|
Your table with the total GSL results doesn't match up with the seasonal results.
Almost all individual results are wrong. For instance, PvZ is 45-68, not 45-47.
Am I missing something?
|
United States17042 Posts
These are some great numbers, thanks for sharing them
|
I think these are great for determining some map balance.
|
Wow and lol.. every time I see that 3:06, I have to lol because of poor sanzenith 
|
On January 10 2011 13:35 Contagious wrote:Wow and lol.. every time I see that 3:06, I have to lol because of poor sanzenith
Guy looked in need of a serious hug at the end of that game.
|
The W/L stats really show how new of a game SC2 is. The fact that there's a sea change in almost every matchup, between every season, means that the metagame is way too volatile to be calling for balance justice just yet.
The sample size and quality of players makes this data difficult to use as absolute proof, but it's clear that things are not as black and white as saying "PvZ is obviously broken in the zerg's favor, Blizzard needs to patch it before things get out of hand."
Once we get some better maps and the metagame stabilizes, then we can talk about balance changes.
|
First of all thanks for the stats. Would have taken a good while to compile. Great job.
I don't really mind about the W/L stats. The main concern is the average duration of games... The only enjoyable game for me this season was Boxer vs Hyperdub on Shakuras Plateau. Having such short games does not allow for much variety or much of a spectacle.
The sample data IS small BUT the majority of the games have been short and crappy such that I am finding it really hard to stay interested...
|
Really interesting how close they %s are even with such low amount of games.
|
Shouldn't GSL 1's PvZ total be 18-13 instead of 18-33? Bringing the win% to 58.1%.
|
On January 10 2011 13:35 Contagious wrote:Wow and lol.. every time I see that 3:06, I have to lol because of poor sanzenith
I just felt sorry for sanzenith, scouted the wrong way, absolutely nothing he could've done, just bad luck.
Looking at the match times, it seems that despite being a VERY unfriendly map for every race due to the backdoor, most people favor playing safe on that map. Of course, 2 games isn't enough to tell anything at all and statistically insignificant, but I found that odd.
The win %'s per map in PvT are somewhat consistent, although sample size is still very close to insignificant. On the maps with back doors, BS and JB, Protoss seems favored, however, apart from that, it is hard to tell what is causing imbalance on some of these maps. Looking at the list, one definite possibility is the droppability(sp?) of each map in relation to rush distance.
|
OP your stats for ZvP is completely wrong for the first two seasons of gsl. PvZ winrate in gsl season 1 is 15-13 nin favour of toss, not 18-33, while your numbers for season two is also a lot lower than that on TL records. This corrected information will show that Protoss winrate is a lot higher than zergs in that matchup than your current data suggests
|
Netherlands6142 Posts
Nicely done! Wanna add this to Liquipedia? If you need help let me know!
|
Excellent job! Thanks for putting all this info in such a convenient way. Toss might have had a hard time, mainly because there weren't as many skilled players as there were from other races. Toss was poorly represented in GSL1 and 2, so I won't be too quick to assume toss is underpowered, since there weren't that many PvX games. The average match length on the other hand was suspected to be short, but 12 minutes for GSL 4 is really really bad (lol at the Blistering Sands average time :D).
|
Fantastic job!
BTW - Do you have any data on how it looks if we take away BitByBitPrimes cheese-matches and how it effects time on each map? 
|
On January 10 2011 14:21 5unrise wrote:
OP your stats for ZvP is completely wrong for the first two seasons of gsl. PvZ winrate in gsl season 1 is 15-13 nin favour of toss, not 18-33, while your numbers for season two is also a lot lower than that on TL records. This corrected information will show that Protoss winrate is a lot higher than zergs in that matchup than your current data suggests
I will go through the data again and make sure the numbers are correct this time. Just so we avoid any confusions, I based by numbers of data provided by Liquipedia for each GSL season. Did you take them from there also or from somewhere else?
|
On January 09 2011 23:48 Xism wrote:Show nested quote +On January 09 2011 23:45 clusen wrote:On January 09 2011 23:38 Xism wrote:
PvZ really has changed. How? You have different veto systems for maps over the seasons and the results are heavily influenced by single players, you can't tell if anything has changed. Yeah, I guess you're right. Was looking too blindly at the statistics
Shortest. Argument. Ever.
I wish all TL arguments were so subtle
|
Yeah, something was wrong with the numbers, although this time I got 17-11, while you say it's 15-13
Will recheck other seasons as well, and try to finish match duration statistics.
Edit: nope, it's 15-13. I think I have a good system now so errors won't repeat. Thanks for pointing out the inconsistencies, I really want this data to be accurate.
|
in addition to game length can you include who won the match?
for example
PvT - protoss win: 15 minutes
PVT - terran win: 10 minutes
|
really surprise at how close the statistic are.
|
The average time length says it all.
|
Interesting stuff. I always knew how bad Delta was, but I was surprised to see the average times of Xel Naga being so low in S4. Thought for sure that map would be as much of a macro map as Metal.
Overall good stuff, statistics are always nice.
|
Take out Kulas Ravine and Desert Oasis makes things look even better...so yeah it's all about the players.
|
Ok, I'm finally done with GSL 3 statistics! I'm really glad I managed to finish this off, as I feel the stats for GSL4 really don't speak that much considering the low number of games played. But with GSL 3, I had a lot more games to work with, so statistic should be a bit better here.
And this took so much more time than it might seem. And slow GOMTV player wasn't helping with that at all.
But let's see..
GSL 3
162 games were played in GSL Season 3.
Average game length was 11:23 minutes.
Average views on VODs is currently at around 69 000.
Map stats:
Player stats:
I've only taken into consideration those players who played more than 5 games, as there were too many of them, and I was getting really tired.
For those interested in raw data, here's my spreadsheet with all the info in it. You will be able to see number of VODs per each map, and even racial distribution (even though some fields are not filled in as I got tired of doing that, and didn't really need racial stats for these tables).
http://www.mediafire.com/?qyuqvwsoscdp8yr
If anyone spots something suspicious, please let me know, and I'll double check my calculations. Will update the OP now, and start working on GSL Season 2 data. And will also see if I can get this into Liquipedia
Edit: oh yeah, and I'm missing one Ro64 game, I think that was MKP getting a bye by his opponent not showing up, right?
|
I think one interesting fact is that on large maps, protoss mobility seems to be alright probably due to pylons. I think when they remove the small maps, that the advantage will really tip in the favor of protoss and zerg.
|
wow Odoakar, i am really impressed by your OP and info/statistics
GREAT JOB!
|
i love that bitbyebit has the shortest average game. very interesting though that scfou actually has the longest games
|
On January 11 2011 07:40 1Eris1 wrote:
i love that bitbyebit has the shortest average game. very interesting though that scfou actually has the longest games
bitbybit plays a lot of TvZ, while scfou plays a lot of TvT
|
I think instead of organizing average game length by games played, doing it by wins / losses would a more telling statistic.
So the player wins his games in an average of X. That may indicate his general playstyle preference or at which time that he prefers to decide the victor.
If He loses his games in an average of Y. That may indicates that he is susceptible to early aggression or mid game allins etc.
|
So, can someone help me about BW? I noticed quite a few match replays of BW games, even those games people recommend, to be about 10 minutes long. So is there a difference between SC2 and BW in how long a good game is? For example, the game where Jaedong had to kill one of his own drones was fairly short, and I wonder if that's the kind of game BW can give me and still be good, why SC2 games of similar length aren't that well received. Is that a difference in the game mechanics themselves?
|
Katowice25012 Posts
Very cool stuff! I debated grabbing these numbers for myself a few times so its great to see someone else did it already.
|
Here are some stats for GSL4, there's not that many games played so stats aren't that good but it's nice to see that at least average game time is a bit longer.
GSL 4
Average game length is 12:16 min. Average views per VOD is 110,929.
![[image loading]](http://i54.tinypic.com/28cixrk.jpg)
![[image loading]](http://i53.tinypic.com/xonq8h.jpg)
|
|
|
|
|
|
|
|
EPT
