|
1115 replays total, from MLG Providence, Nov 2011
(in brackets - number of replays)
Race Avg Length
Z 0:11:35 (673)
P 0:11:50 (573)
T 0:12:48 (610)
Mu Avg Length
PvP 0:08:02 (90)
ZvZ 0:08:51 (173)
TvT 0:12:00 (111)
ZvP 0:12:05 (242)
ZvT 0:12:58 (258)
TvP 0:13:01 (241)
As you can see, the common wisdom that PvP is the shortest matchup, is true, and ZvZ is quite short too, on average. The other common wisdom that Terran games take longer is also true, but TvT is actually still shorter than TvZ and TvP. Overall, mirrors are shorter than non-mirrors. And you can conclude that a group with terran participation would take longer to play all matches than a group without any terrans, for example.
My goal was to get some predictive idea about how long some broadcast could be. And the conclusion is for example that a 2T group of 4 with RR Bo3s will take at least 15-20 minutes more on average, than a non-terran group. Also a very mirror-based group should be expected to take significantly shorter than a regular non-mirror group, in some distributions with up to ~45 minutes difference.
N.B.!: All times are in replay time, not real world time (so in real time they come as 72% shorter).
 
|
Israel2206 Posts
This is interesting, but I think the skill discrepancy in many of the matches in MLG's open bracket may have resulted in lower game lengths over all.
Also, are these replay game lengths, or real world time?
|
On January 19 2012 19:58 Noam wrote:
This is interesting, but I think the skill discrepancy in many of the matches in MLG's open bracket may have resulted in lower game lengths over all.
Also, are these replay game lengths, or real world time?
This is definitely true. The fact that there are so many noobs trying to cheese pros in the open bracket as well as the mentioned skill discrepancy (IMMVP vs Low NA Masters) makes games shorter.
|
This is interesting, not very surprised by the fact of pvp and zvz being so short. I am a bit suprised of TvT not being the longest MU tho... but maybe there is more 1-2base timing pushes than I think.
Thanks for the statistic tho, always funny to read this =)
|
On January 19 2012 19:58 Noam wrote:
This is interesting, but I think the skill discrepancy in many of the matches in MLG's open bracket may have resulted in lower game lengths over all. No, I just made brief calculations for every matchup from just the championship, and the games aren't longer, actually PvPs are much shorter for example. Of course, there are too few games for such pool of data to matter, so I'll just leave the whole data.
About TvTs I was surprised they aren't longer too, but it seems the long games don't happen as often as it appears; too many TvTs are around or below 10 minutes, and they counter-balance the long ones.
|
Surprises me that the Z matchups are apparently the fastest overall. Seems like any time I ladder, I'm guaranteed to play an 18 minute game minimum.
|
On January 19 2012 21:07 MrBitter wrote:
Surprises me that the Z matchups are apparently the fastest overall. Seems like any time I ladder, I'm guaranteed to play an 18 minute game minimum. That's a good point - this isn't ladder (or online tourney), where people play comfortably - in LANs as we know they tend to play worse than usual, because of multiple logistics factors and stress. Might affect the game length.
I was surprised that Zerg comes up with the shortest games too... not sure what to make of it, but the difference with Protoss isn't that big anyway, and ZvT is still pretty epic. (:
|
On January 19 2012 21:15 figq wrote:Show nested quote +On January 19 2012 21:07 MrBitter wrote:
Surprises me that the Z matchups are apparently the fastest overall. Seems like any time I ladder, I'm guaranteed to play an 18 minute game minimum. That's a good point - this isn't ladder (or online tourney), where people play comfortably - in LANs as we know they tend to play worse than usual, because of multiple logistics factors and stress. Might affect the game length. I was surprised that Zerg comes up with the shortest games too... not sure what to make of it, but the difference with Protoss isn't that big anyway, and ZvT is still pretty epic. (:
To be fair, you had twice as many ZvZ's in your data pool than any other mirror.
Yea, upon seeing that, this all makes perfect sense. :D
|
I'd like to see you do a new one from let's a Code S season, should probably be a little longer games in general, and a more reliable result imo
|
On January 19 2012 21:19 MrBitter wrote:Show nested quote +On January 19 2012 21:15 figq wrote:On January 19 2012 21:07 MrBitter wrote:
Surprises me that the Z matchups are apparently the fastest overall. Seems like any time I ladder, I'm guaranteed to play an 18 minute game minimum. That's a good point - this isn't ladder (or online tourney), where people play comfortably - in LANs as we know they tend to play worse than usual, because of multiple logistics factors and stress. Might affect the game length. I was surprised that Zerg comes up with the shortest games too... not sure what to make of it, but the difference with Protoss isn't that big anyway, and ZvT is still pretty epic. (: To be fair, you had twice as many ZvZ's in your data pool than any other mirror. Yea, upon seeing that, this all makes perfect sense. :D Hehe, indeed if we just average the matchup averages for each race, we end up with Protoss slightly faster  :
P 0:11:02
Z 0:11:18
T 0:12:39
|
What happens if you just take the winning games (i.e. PvZ where Protoss wins, ZvT where Zerg wins)?
|
nice post, it's good to see the numbers to support the modern notions of matchup length!
|
On January 19 2012 23:43 Dragar wrote:
What happens if you just take the winning games (i.e. PvZ where Protoss wins, ZvT where Zerg wins)? ZvP means Z wins, PvZ means P wins:
TvZ 0:11:33 (130)
PvT 0:11:41 (98)
ZvP 0:12:04 (117)
PvZ 0:12:06 (125)
TvP 0:13:55 (143)
ZvT 0:14:23 (128)
Lol, apparently when T wins ZvT it's the shortest non-mirror, when Z wins it's the longest of all. That way you could predict a ZvT outcome by its length to some extent, and even somewhat for PvT (it has similar big difference).
That also tells us that Zerg's goal in ZvT is longer game while Terran's goal is shorter game; but in TvP Terran's goal is longer game, and Protoss' goal is shorter game. (on average, of course)
|
these results are funny because for many people T has an avantage in the early game vs P. Seems like it is not the case here.
|
On January 20 2012 00:55 Nhed wrote:
these results are funny because for many people T has an avantage in the early game vs P. Seems like it is not the case here. This could still be true, because the averages don't tell the whole story of back and forth advantages in different game stages.
|
On January 20 2012 00:26 figq wrote:Show nested quote +On January 19 2012 23:43 Dragar wrote:
What happens if you just take the winning games (i.e. PvZ where Protoss wins, ZvT where Zerg wins)? ZvP means Z wins, PvZ means P wins: TvZ 0:11:33 (130)
PvT 0:11:41 (98)
ZvP 0:12:04 (117)
PvZ 0:12:06 (125)
TvP 0:13:55 (143)
ZvT 0:14:23 (128) Lol, apparently when T wins ZvT it's the shortest non-mirror, when Z wins it's the longest of all. That way you could predict a ZvT outcome by its length to some extent, and even somewhat for PvT (it has similar big difference). That also tells us that Zerg's goal in ZvT is longer game while Terran's goal is shorter game; but in TvP Terran's goal is longer game, and Protoss' goal is shorter game. (on average, of course)
There's no way that's accurate overall.. I think 90% of top terrans would tell you they'd much rather play a 30-40min game vs Z than vs P. x_x
|
On January 20 2012 01:29 Pokebunny wrote:Show nested quote +On January 20 2012 00:26 figq wrote:On January 19 2012 23:43 Dragar wrote:
What happens if you just take the winning games (i.e. PvZ where Protoss wins, ZvT where Zerg wins)? ZvP means Z wins, PvZ means P wins: TvZ 0:11:33 (130)
PvT 0:11:41 (98)
ZvP 0:12:04 (117)
PvZ 0:12:06 (125)
TvP 0:13:55 (143)
ZvT 0:14:23 (128) Lol, apparently when T wins ZvT it's the shortest non-mirror, when Z wins it's the longest of all. That way you could predict a ZvT outcome by its length to some extent, and even somewhat for PvT (it has similar big difference). That also tells us that Zerg's goal in ZvT is longer game while Terran's goal is shorter game; but in TvP Terran's goal is longer game, and Protoss' goal is shorter game. (on average, of course) There's no way that's accurate overall.. I think 90% of top terrans would tell you they'd much rather play a 30-40min game vs Z than vs P. x_x That's because of how the average works, and the matchups work. Each matchup is designed to switch the advantage multiple times throughout the game stages. The average may capture some kind of advantages around the mid-game lengths, rather than the direct comparison between early and late game.
P.S. By the way, of 1115 replays, there's no game above 30 minutes, the longest are 28 minutes. Funny.
|
Wow, thanks for this very much.
|
|
|
|
|
|
WCS
