EPTNever Miss An Inject? What the Data Say - Page 7
| Forum Index > SC2 General |
|
breakit
United States29 Posts
| ||
|
Hypemeup
Sweden2783 Posts
On May 22 2013 07:05 _SpiRaL_ wrote: As I said already this is a very small % of total APM and the difference this makes is very small. I dunno why this ALWAYS come up when discussing APM.Zerg unit spam makes a tiny difference only, its an APM intensive race. It actually makes a huge difference, I dont know what you are going on about. | ||
|
contv
35 Posts
On May 22 2013 05:27 Armada Vega wrote: well, just to inform you. Zergs apm is always artificially inflated, since holding down Z to make lings is counted as pushing the key down for each ling. If you make 25-36 lings at once, the game counts this as 25-36 actions a second, roughly. This will inflate the zergs current apm to 500 or higher for that one production cycle. This on its own is no big deal. But this effects your average apm based on the number of times you mass produce anything from larvae from the beginning of a game to the end of a game. Often times, a zergs average apm is 30-50apm off. After your game ends, if your average apm was 230, and the terrans is 200-190apm, you have exactly the same apm. The zergs apm is artificially inflated per production cycle that you make while holding down a key. Combine all mass production cycles in a game; drone, ling, roach, muta, overlords, ultralisk, banelings, broodlords, and this greatly effects average apm. Disagree. Making 1 pair of zerglings counts as 1 action and costs 1 supply. Making 1 marine counts as 1 action and costs 1 supply. For the purposes of average APM, it doesn't matter that zerg was able to hold down Z and briefly reach 500+apm, as opposed to rapidly queuing up marines - given that supplies for Zerg and the other races grow relatively similarly (watch Polt's games, where he often maintains higher supply than Zerg) and 1 action counts for 1 supply, there is no inflation. You could argue that the supply inefficiency of zerg units actually limits their APM - eg making roaches, where 1 action expends 2 supply. | ||
|
Markwerf
Netherlands3728 Posts
Comparing the largest difference to the standard deviation at a point is lousy way to say the difference is small, because you don't account for the fact that the positive difference is pretty much across the entire range. Test for example equality of intercepts of both regression lines as it seems very reasonable to assume from this data the slope is fairly constant but the intercepts differ. Basically a difference should be compared with some measure for the standard deviation of the difference while here it compares a difference with just a standard deviation of one group which is usually much larger, it's somewhat like applying a independant two sample t-test when you should be doing a paired t-test. Besides that the article kind of implies that inject efficiency being a thing should show up as while this is not really true at all. It's easy for a real effect to be hidden by confouding in an observational study, there are tons of examples where this is thought to be true. For example there is no shown correlation between wearing a seatbelt and lower risk of dying in an accident while it's pretty much an accepted fact seatbelts are very beneficial. Something akin to that is very well possible here and seems far more likely than the theory suggested in the article which sounds rubbish. Higher skilled players make hatches faster which are harder to inject all and meet more aggressive opponents who make it harder to keep injecting. The end result is pretty sound though, it's a shitty statistic as indicator for skill. | ||
|
petered
United States1817 Posts
On May 22 2013 07:32 LaLuSh wrote: Dsjoerg, Blizzard said they'd add more data to replay files. You know what exactly they're gonna add (or have added, I'm not in the loop. Have they already implemented this patch?). If replays contained supply data it would be interesting to plot inject rate against the average supply during a game (assuming this is possible to calculate). I would bet that that out of two 25 minute games, the one with the lower average game supply would generally have the higher inject rates. I believe that you are correct, but that differentiates the necessity of injects rather than the standard of injects for various leagues, which was the intent (at least part of it). | ||
|
dsjoerg
United States384 Posts
On May 22 2013 07:32 LaLuSh wrote: Dsjoerg, Blizzard said they'd add more data to replay files. You know what exactly they're gonna add (or have added, I'm not in the loop. Have they already implemented this patch?). If replays contained supply data it would be interesting to plot inject rate against the average supply during a game (assuming this is possible to calculate). I would bet that that out of two 25 minute games, the one with the lower average game supply would generally have the higher inject rates. The patch was implemented, it came out 2 weeks ago. Supply data is in there! Thanks for the suggestion <3 makes sense | ||
|
CrushDog5
Canada207 Posts
| ||
|
Glenn313
United States475 Posts
| ||
|
ZenithM
France15952 Posts
Injecting is very hard mechanically and it's clear that Zerg is not balanced with the assumption of perfect injects from the get-go. I've known for a long time that even high level players are somewhat bad at injecting (just look at replays of people you consider mechanical gods, they always have way too much energy on their queens past 12 minutes, because they have a lot of other shit to do ^^) What is interesting however is that there isn't much difference between high and low level players. That I didn't expect :D | ||
|
Pursuit_
United States1330 Posts
On May 22 2013 08:14 contv wrote: Disagree. Making 1 pair of zerglings counts as 1 action and costs 1 supply. Making 1 marine counts as 1 action and costs 1 supply. For the purposes of average APM, it doesn't matter that zerg was able to hold down Z and briefly reach 500+apm, as opposed to rapidly queuing up marines - given that supplies for Zerg and the other races grow relatively similarly (watch Polt's games, where he often maintains higher supply than Zerg) and 1 action counts for 1 supply, there is no inflation. You could argue that the supply inefficiency of zerg units actually limits their APM - eg making roaches, where 1 action expends 2 supply. It's not about how many action, it's about the amount of time required to do it. Terran's have 100 other things they'd rather be doing, but they need to only make as many marines as they have barracks (i.e. hit a button 10 times to make 10 marines because you have 5 reactored rax) whereas zerg can just hold the button down to make as many zerglings as possible, meaning an easy 500+ apm spike before returning to standard apm doing other things immediately. If actions themselves were the limiting factor, what you said makes sense, but actions aren't, time is. I get about 50% apm inflation playing as Z as opposed to T (avg terran apm is ~140, zerg is ~220) and when I play spammy I get upwards of 100% apm inflation as Z. Might differ for others, but I genuinely think just play playing Z you get a pretty significant APM boost. | ||
|
TrollPolice
44 Posts
| ||
|
sambo400
United States378 Posts
On May 22 2013 08:27 ZenithM wrote: It's really not a surprise for me. Injecting is very hard mechanically and it's clear that Zerg is not balanced with the assumption of perfect injects from the get-go. I've known for a long time that even high level players are somewhat bad at injecting (just look at replays of people you consider mechanical gods, they always have way too much energy on their queens past 12 minutes, because they have a lot of other shit to do ^^) What is interesting however is that there isn't much difference between high and low level players. That I didn't expect :D It also doesn't matter as much after that point in the game. By then you have 4-5 hatches and much more cost effective units. Early game when droning and making shitty units like roaches and lings off of 2-3 hatches its a huge deal that you get every last one that you can. But once you are up to units like infestors, swarm hosts, mutas, ultras, vipers, brood lords, etcetera, you can do way more with less larva. Plus spores and spines don't really cost larva late in the game since you were going to make the initial drone no matter what. And if you are Fitzy then half of your army doesn't cost larva to begin with. | ||
|
Evangelist
1246 Posts
| ||
|
Mindcrime
United States6899 Posts
On May 22 2013 08:40 sambo400 wrote: It also doesn't matter as much after that point in the game. By then you have 4-5 hatches and much more cost effective units. Early game when droning and making shitty units like roaches and lings off of 2-3 hatches its a huge deal that you get every last one that you can. But once you are up to units like infestors, swarm hosts, mutas, ultras, vipers, brood lords, etcetera, you can do way more with less larva. Plus spores and spines don't really cost larva late in the game since you were going to make the initial drone no matter what. And if you are Fitzy then half of your army doesn't cost larva to begin with. Swarm hosts, indeed. I've opened swarm host before taking a third in just about every zvp I've played since hots came out and ggtracker thinks my macro in zvp is just plain shitty compared to my zvz and zvt. | ||
|
Evangelist
1246 Posts
On May 22 2013 05:52 Ender985 wrote: Interesting. I wonder, can you do the same graph for Mule calldowns and protoss injects to see if it also follows the same trend? I suspect it does. It won't. With protoss it will, with huge spikes in energy usage for each major engagement. With terrans, lower leaguers will call down mules in huge spurts. Grandmasters and higher will eventually make the 19 orbital transition and their mule usage will spike. What would be a more interesting thing is to see how the number of constructed production buildings varies per league. | ||
|
Gene(S)is
Sweden419 Posts
But Queens are also used to transfuse and spread creep, is that considered when making these studies? Overall though I have to say cheers for an intresting topic. | ||
|
Steel
Japan2283 Posts
This is still very interesting and shows how few have mastered sc2 mechanically. This is still only data from the ladder which is pretty low level. To address claims that Zerg is the easiest macro, hold your tongue. I watch replays a lot and it's baffling at midmasters how many struggle with things like supply very early on in the game. I hate to target people in particular but I'm thinking about protoss specifically who make several macro mistakes. Zerg macro early on is extremely tight, as you have to prepare for every scenario. A classic example is the forge expand into 4 gate pressure, really popular in WoL and doing a resurgence nowadays with the mothership core. With standard timings, you scout it as the gates finish. The build works so that your inject at your third pops at the same time as your roach warren finishes, as the first wave of zealot are warped in. Get supply blocked once, or get behind in your inject, and you're looking at game ending damage. A lot of 2 base timings pretty much require you to max out. Sure there is some finesse in collosi all in but I think the macro required to defend it is much harder. It's tough to complain about tech switches being easy too, when you have idle max energy sentries sitting at your natural that you could use to make the fastest unit in the game to scout. Or have invisible detectors. Or have instant detection from your 3CCs. I really don't think T and P are in a position to complain about scouting. | ||
|
Wuster
1974 Posts
On May 22 2013 03:57 Cyro wrote: You're missing something important in that higher level players will more often use energy for creep tumors etc on purpose with early queens while lower ones are much much more prone to just floating energy, and also one of the biggest things: This is definitely true; I remember when Idra was teaching JP the 14 hatch build; he flat out said the first queen doesn't inject because you don't have the minerals to morph the larvae (or maybe it was the second queen, point was it was an FE buld where only one hatch was injected due to econ constraints, not skill). On May 22 2013 06:17 Eventine wrote: I swear, every time someone provides some data point, instead of sparking interesting debate, half the responses are like your data is wrong, my intuition is perfect and therefore i reject your findings. Agreed, the point of the article however shows that larvae management / macro is much more than just how often you inject with your queens. Half the anecdotal 'apologists' for why better players have worse inject uptimes even follow this conclusion; it's about planning out when you actually need the larvae, rather than just spamming them all out. So yes, controlling your injects is important, blindly spamming them is less so (but can't hurt right?!). | ||
|
habermas
United Kingdom304 Posts
| ||
|
da_head
Canada3350 Posts
On May 22 2013 04:06 SpikeStarcraft wrote: Im calling bullshit on this. I dont think what you showed there means anything. The inject percentage is correlated to the numbers of hatcheries. And in higher level play you have more hatcheries way faster. so after 10 minutes when the master zerg is on 4 base he does way more absolute injects than the silver level player thats still on one base. So thats not an useful comparison. "Wow, the silver level player hits injects on one base almost as good as master level players on 4 base. I guess injects dont really matter that much." Kappa Calling bullshit? And where's you're subsequent data to disprove them? | ||
| ||
