|
I've watched a lot of GSL. I've also watched a lot of pro Terrans. But I repeat myself.
There's a ton of posts out there looking at the number of Terrans in GSL. But I couldn't help but feel that the "Terran saturation" in GSL had just as much to do with match duration as the number of players.
So I went through Liquipedia and slapped together a spreadsheet looking at the average match duration in every matchup. The data I used was 6 months of Code S + A. I found some very interesting results.
https://docs.google.com/spreadsheet/ccc?key=0AnLmn5yALkv_dEowUEhjOVRsck0yc2VOalFwT3c4S0E
TvT: 260 games, 21.3 avg game minutes
TvZ: 269 games, 20.9 avg game minutes
TvP: 238 games, 18.3 avg game minutes
TvX: 767 games, 20.2 avg game minutes
Terran is playing in GSL 44.7% of the time
ZvT: 269 games, 20.9 avg game minutes
ZvZ: 49 games, 13.0 avg game minutes
ZvP: 170 games, 19.9 avg game minutes
ZvX: 488 games, 19.8 avg game minutes
Zerg is playing in GSL 28.4% of the time
PvT: 238 games, 18.3 avg game minutes
PvZ: 170 games, 19.9 avg game minutes
PvP: 54 games, 11.2 avg game minutes
PvX: 462 games, 18.1 avg game minutes
Protoss is playing in GSL 26.9% of the time
Here's some observations:
1. There seems to be an "ideal" match duration of about 20 game minutes (which equals 15 real-life minutes).
2. The matchups that do not fit the mold are TvP, ZvZ, PvP, and PvX. I found this personally interesting because those are also the matchups that I think are the most boring to watch.
3. There have been so few ZvZ and PvP games in the GSL, that being very strong in those matchups is practically worthless. Keep in mind that this data *did* include most of the big non-T mirrors in GSL history. Inca's PvP run up to Code S finals is included, as is Nestea vs Losira Code S finals, and the Tassadar vs Puzzle Code A finals.
4. On the other hand, strength in the Terran mirror matchup is incredibly important. It's okay for a top-level Zerg to have a terrible ZvZ (*cough*JulyZerg*cough), however its a career-killer for Terrans.
5. Common knowledge suggests "the longer the game, the greater the probability that the better player will win". According to this data, that's great for Terran and Zerg, but it's bad for Protoss. It could very well be that one part of the reason why there are so few Protoss in the GSL is because the "best" Protoss players are more vulnerable to getting knocked out by lesser skilled opponents than the other races.
 
|
5 is interesting because pvp is pulling down the average. this might lead to the conclusion that because pvp is shorter and thus the better player has less of a chance to win than in other match ups, it's more probable that the lesser protoss will advance, and then lose a non-mirror that the loser of the pvp would have won
|
Are you the same robobob that used to play wow with aquabuggy?
|
Blizzard once had a post somewhere about "the ideal length of a ladder game", they said they wanted it to be 20-30 in game minutes, I believe.
|
50 minutes. Settle for no less.
Seriously, though, I don't know how you're interpreting the average duration as the "ideal" duration
|
Ideal duration is irrelevant if the game played is terrible.
Nice statistcs though, interesting to see that they all come up around the same length in the respective matchup's (Bar ZvZ and PvP)
|
This gets me thinking how strong an effect on success rates the mirrors have. Using that old adage you mentioned of longer games favoring better players, it seems to me that the best Terrans will advance because on average TvT is the longest mirror match up (and the longest match up in general) whereas the best Zerg and Protoss players are a lot less likely to advance by comparison because their mirrors are very short.
This may result in the higher rounds of the tournament containing Zerg and Protoss players who aren't necessarily the best representatives of their races, as well as Terran players who are, making the Terran players better on average.
Never thought about Terran being overpowered (or at least getting better results) because of TvT 
|
Seriously though, if you ignore ZvZ and PvP, everything looks within one standard deviation or two of "average", meaning that those two matchups are the easiest to win in the early game (duh) and need to be looked at as to how to actually have defensive options other than "do a defensive version of the offense your opponent is doing."
Of course, this is all common knowledge to anyone who's watched or played SC2 ever.
Also being good at XvT will do you well in GSL because most of GSL *is* T (though that number is, slowly, decreasing)
I like knowing the average durations so that's some good work, it just seems your conclusions are, for the lack of a better word, extremely obvious. Is there a point you're trying to make specificially?
Good job with the data though. ^_^
|
On January 29 2012 10:19 RoboBob wrote:I've watched a lot of GSL. I've also watched a lot of pro Terrans. But I repeat myself. There's a ton of posts out there looking at the number of Terrans in GSL. But I couldn't help but feel that the "Terran saturation" in GSL had just as much to do with match duration as the number of players. So I went through Liquipedia and slapped together a spreadsheet looking at the average match duration in every matchup. The data I used was 6 months of Code S + A. I found some very interesting results. https://docs.google.com/spreadsheet/ccc?key=0AnLmn5yALkv_dEowUEhjOVRsck0yc2VOalFwT3c4S0ETvT: 260 games, 21.3 avg game minutes TvZ: 269 games, 20.9 avg game minutes TvP: 238 games, 18.3 avg game minutes TvX: 767 games, 20.2 avg game minutesTerran is playing in GSL 44.7% of the time ZvT: 269 games, 20.9 avg game minutes ZvZ: 49 games, 13.0 avg game minutes ZvP: 170 games, 19.9 avg game minutes ZvX: 488 games, 19.8 avg game minutesZerg is playing in GSL 28.4% of the time PvT: 238 games, 18.3 avg game minutes PvZ: 170 games, 19.9 avg game minutes PvP: 54 games, 11.2 avg game minutes PvX: 462 games, 18.1 avg game minutesProtoss is playing in GSL 26.9% of the time Here's some observations: 1. There seems to be an "ideal" match duration of about 20 game minutes (which equals 15 real-life minutes). 2. The matchups that do not fit the mold are TvP, ZvZ, PvP, and PvX. I found this personally interesting because those are also the matchups that I think are the most boring to watch. 3. There have been so few ZvZ and PvP games in the GSL, that being very strong in those matchups is practically worthless. Keep in mind that this data *did* include most of the big non-T mirrors in GSL history. Inca's PvP run up to Code S finals is included, as is Nestea vs Losira Code S finals, and the Tassadar vs Puzzle Code A finals. 4. On the other hand, strength in the Terran mirror matchup is incredibly important. It's okay for a top-level Zerg to have a terrible ZvZ (*cough*JulyZerg*cough), however its a career-killer for Terrans. 5. Common knowledge suggests "the longer the game, the greater the probability that the better player will win". According to this data, that's great for Terran and Zerg, but it's bad for Protoss. It could very well be that one part of the reason why there are so few Protoss in the GSL is because the "best" Protoss players are more vulnerable to getting knocked out by lesser skilled opponents than the other races.
You say 'terran saturation' has to do with average game length, but in your own stats, theres almost twice as many terran games as games in the other matchups. If you multiply the number of minutes by matchup by the number of T matchups vs. Z and P, you'll see even more 'terran minutes' of gameplay. Even if terran matchups were say 25% lower than the average, there would still be more total terran minutes overall.
1. What does 'ideal' mean? Does it mean most entertaining? does it mean 'average'? If it means something like 'optimum', then optimum for who? for what? what purpose?
3+4. This is kind of interesting, but the relatively lower number of ZvZ and PvP games is almost totally related to the lower percentage of total players that Z and P comprise relative to terran. In a situation where there are more terran overall, of course your skill vs. terran will be relatively more important.
5. I'm not sure if this is true. Why do you think so?
|
Ugh ideal is a really bad choice of word here.
|
On January 29 2012 10:28 Odal wrote:
Are you the same robobob that used to play wow with aquabuggy?
Nope, afraid not =/
|
On January 29 2012 11:05 UmiNotsuki wrote:This gets me thinking how strong an effect on success rates the mirrors have. Using that old adage you mentioned of longer games favoring better players, it seems to me that the best Terrans will advance because on average TvT is the longest mirror match up (and the longest match up in general) whereas the best Zerg and Protoss players are a lot less likely to advance by comparison because their mirrors are very short. This may result in the higher rounds of the tournament containing Zerg and Protoss players who aren't necessarily the best representatives of their races, as well as Terran players who are, making the Terran players better on average. Never thought about Terran being overpowered (or at least getting better results) because of TvT
That is a really good point, I didn't think of it that way. But it makes sense, especially because TvT is almost 2x as long as the mirrors.
I guess its like saying that MVP is less likely to be KOed by a noob in a 20 min TvT than it is for MC to be KOed by a noob in an 11 min PvP.
|
On January 29 2012 11:33 caradoc wrote:Show nested quote +On January 29 2012 10:19 RoboBob wrote:I've watched a lot of GSL. I've also watched a lot of pro Terrans. But I repeat myself. There's a ton of posts out there looking at the number of Terrans in GSL. But I couldn't help but feel that the "Terran saturation" in GSL had just as much to do with match duration as the number of players. So I went through Liquipedia and slapped together a spreadsheet looking at the average match duration in every matchup. The data I used was 6 months of Code S + A. I found some very interesting results. https://docs.google.com/spreadsheet/ccc?key=0AnLmn5yALkv_dEowUEhjOVRsck0yc2VOalFwT3c4S0ETvT: 260 games, 21.3 avg game minutes TvZ: 269 games, 20.9 avg game minutes TvP: 238 games, 18.3 avg game minutes TvX: 767 games, 20.2 avg game minutesTerran is playing in GSL 44.7% of the time ZvT: 269 games, 20.9 avg game minutes ZvZ: 49 games, 13.0 avg game minutes ZvP: 170 games, 19.9 avg game minutes ZvX: 488 games, 19.8 avg game minutesZerg is playing in GSL 28.4% of the time PvT: 238 games, 18.3 avg game minutes PvZ: 170 games, 19.9 avg game minutes PvP: 54 games, 11.2 avg game minutes PvX: 462 games, 18.1 avg game minutesProtoss is playing in GSL 26.9% of the time Here's some observations: 1. There seems to be an "ideal" match duration of about 20 game minutes (which equals 15 real-life minutes). 2. The matchups that do not fit the mold are TvP, ZvZ, PvP, and PvX. I found this personally interesting because those are also the matchups that I think are the most boring to watch. 3. There have been so few ZvZ and PvP games in the GSL, that being very strong in those matchups is practically worthless. Keep in mind that this data *did* include most of the big non-T mirrors in GSL history. Inca's PvP run up to Code S finals is included, as is Nestea vs Losira Code S finals, and the Tassadar vs Puzzle Code A finals. 4. On the other hand, strength in the Terran mirror matchup is incredibly important. It's okay for a top-level Zerg to have a terrible ZvZ (*cough*JulyZerg*cough), however its a career-killer for Terrans. 5. Common knowledge suggests "the longer the game, the greater the probability that the better player will win". According to this data, that's great for Terran and Zerg, but it's bad for Protoss. It could very well be that one part of the reason why there are so few Protoss in the GSL is because the "best" Protoss players are more vulnerable to getting knocked out by lesser skilled opponents than the other races. You say 'terran saturation' has to do with average game length, but in your own stats, theres almost twice as many terran games as games in the other matchups. If you multiply the number of minutes by matchup by the number of T matchups vs. Z and P, you'll see even more 'terran minutes' of gameplay. Even if terran matchups were say 25% lower than the average, there would still be more total terran minutes overall. 1. What does 'ideal' mean? Does it mean most entertaining? does it mean 'average'? If it means something like 'optimum', then optimum for who? for what? what purpose? 3+4. This is kind of interesting, but the relatively lower number of ZvZ and PvP games is almost totally related to the lower percentage of total players that Z and P comprise relative to terran. In a situation where there are more terran overall, of course your skill vs. terran will be relatively more important. 5. I'm not sure if this is true. Why do you think so?
Well, I see two ways to look at "Too much Terran". Is it just because Terrans play the most games? Or is it also that Terran games they do play take a longer time to finish?
There's been a ton of threads about the first question, and the results are fairly obvious (just look at tournament results). But I couldn't find anything on the second. That's why I made this thread. And it was kinda cool to learn that Terran games lasted just as Zerg games. And it was also cool to learn that Protoss plays the shortest games, which I didn't expect at all.
Hm I wonder if the shorter game time for Protoss is partially because of warpin negating defender's advantage. I don't know why it would result in a "short" PvP and PvT but not PvZ though.
Oh and for the record, when I was talking about "ideal", I meant ideal for pure spectators to watch. It's the goldilocks principle: you don't want a game that's too short like a 7 min PvP, but you also don't want a game that's too long like a 30 min TvT.
|
o.o I'm surprised. It's nothing like BW's average game length of like 12 minutes or something.
|
interesting stats, already kinda knew terran play longer games though overall
|
can you show histograms for all data? using averages can sometimes hide that there is more than one peak. you would exspect one around macro games and another earlier peak for rushes timings etc.
|
United Kingdom16710 Posts
I would dearly love to see Goody's average game length.
For me, there is no ideal duration as long as the game itself is entertaining. However, this usually means most of the games ending in the first 10 minutes are underwhelming.
|
|
|
|
|
|
EPT
