|
Regarding gas timings..
I played 2-3 games vs protoss, doing this build.
As soon as both mineral lines were satured I got quad gas (for roaches) or double gas (for ling/bane).
If roaches, I went Lair, Roach warren.
If Zerglings I went Metabolic Boost, Banelings next.
In most of the games, if not all I had burrow, roach speed and burrowing claws WHEN they attacked.. as well as 30 odd roaches.
Same with the ling speed and banelings, and I always get burrow now.
If any one is interested, I have about 7-8 replays of me using this build, a few loses and a few wins.
Using it for both ling attacks and roach attacks.
I can upload them into a replay pack if you'd like.
Edit: 1500 Diamond.
|
On November 30 2010 15:50 Skrag wrote:
The spawn larvae cycle is actually *more* important than the current drone count. For example, at 6 minutes, your "winning" build at 43 drones has the benefit of finishing drones off 2 recent queen spawns, while the "inferior" builds that I've tried, and the 16h/15p replay, only have the benefit of 1. Shift the recording time 20 seconds or so, and suddenly things would look different.
The 13 pool/16 hatch build has thie same problem; its 'drone count' is very rapidly increasing at the end, but have all only just started.
However....
Many of these builds are production capped before 6 minutes. Hence the large mineral stockpile
toward the end. Given that, the total drone count seems a little irrelevent anyway.
Right now I'm leaning towards generating 'income curves'. If we can generate curves showing income as a function of time for each build, we'd have a much clearer picture of when each build is behind/ahead of another. It also would allow comparison with any models to be much easier.
Obviously this is a lot more work though.
|
I was so intrigued by Lomilar's build, that I had to try it out myself. I deviated a tiny bit, but not much... I think I followed it up to about 30 supply, so it was after I got my expansion. I've tried this build 3 times so far and won two games, and the game I lost was simply because I threw away a huge lead in one battle... felt like an idiot. Anyways, I am extremely impressed with this build so far. I would have never expected it to do so well economy wise.
Here is a replay of me executing the build. 1800+ Diamond
http://www.sc2replayed.com/replays/110784-1v1-terran-zerg-steppes-of-war
I really like this build so far. Like I said, I followed the build pretty well, and I scouted a very early 2 racks... I was able to drone hard and still pump lings in time to fend off the attack. I then had a far superior economy, tech switched to mutas, and eventually won the game. It was a far from perfect mid-late game, but I think that replay shows the strengths of this build early on. let me know what you guys think, what could have been done better, etc...
|
On November 28 2010 19:03 jdseemoreglass wrote:
However, now I face a slight dilemma that I am seeking consensus on. Douillos' build is ahead in regards to total minerals mined, but Lomilar's build is ahead in regards to drone count. In order to decide the leading build, we must come up with a method of comparing the relative worth of drones to minerals.
I have been thinking of valuing a drone as it's original cost + the number of minerals it could mine within the 17 seconds it requires to produce one. However, this may be a flawed method. Please give me input regarding this issue so we can reach a conclusion.
I think that once you have it narrowed down to 1 or 2 builds, you need to add in the extra touches, such as when to scout, when to add a gas, how much gas to harvest, when to build 1 set of zerglings (i mean really, you ALWAYS want at least 1 set), and possibly rallying appropriately to increase mining time.
When you start doing those things, you start to reduce the impact of oversaturation, thus giving the 'more drones' build a lead over the 'more minerals mined' build. It also more closely approximates a real game.
If it really is super-close and the choice is more drones vs more minerals, then I say DON'T decide which one is better. Just say that these are the two most economical builds, and leave it to the player to decide in the moment.
|
|
|
On November 30 2010 17:26 Lomilar wrote:
Okay. I have taken this build to the next level of silliness:
Double extractor trick @ 10 & 110 minerals
12 Overpool
16 OL (no really)
18 Queen
Send drone to hatch at 190 minerals
18 Hatch
21 Queen (should start right as previous queen finishes and you have minerals)
23 Overlord (should finish right as eggs pop)
29 Overlord
Maynard 8 drones at 30 (this should be when hatch pops)
Overlord at 37
I had studied a 12 overpool build like that:
http://www.teamliquid.net/forum/viewmessage.php?topic_id=131345
But:
- 9 OL 12 pool is better than 12 overpool
- With this build you get a lot of larva (early), and you need to have enough economy to convert them, so you need to build only drones at the beginning. So I feel it is a bit less flexible.
- Because of the different spawn larva cycles, it is hard to decide what the best build is.
|
A thought: to try and mitigate the effect of larvae inject on our 6-minute readings, perhaps we should value them with the following method.
A drones takes 17+40 seconds to produce.
a) A drone X time into production should be valued as (40+X)/57 worth of a drone.
b) Unpopped larvae due to injection, a time T into the larvae injection process, should be valued as
as T/57 worth of a drone for earch larva.
Edit: It's very important we do this actually, or any drones at 0/17 simply vanish from the output, and the larvae also vanish. This could give a false impression when reporting the results.
|
![[image loading]](http://www.gamereplays.org/community/uploads/repimgs/repimg-33-170226.jpg)
+ Show Spoiler +
9 overlord
14 pool (finishes @ 2:56)
16 hatch (finishes @ 4:20)
16 queen
18 overlord
18 extractor trick
22 queen
24 overlord
32 overlord
35 overlord
Drones: 42 + 15/17 = 42.88
Minerals Mined: 4000 + 664 = 4664
Queen Energy: 16 + 24 = 40
Remaining Larva: 0 (4 about to pop 1 second later from inject though)
(no sending to individual patches this time)
Notes: I made my second queen a few seconds too late and I oversaturated my main
|
On December 01 2010 00:16 Dragar wrote:
A thought: to try and mitigate the effect of larvae inject on our 6-minute readings, perhaps we should value them with the following method.
You can take them into account just by looking at the replays, its just a huge pain in the ass. You have to record hatchery larvae spawn times and queen larvae spawn times, and figure out which build is ahead when in order to compare them directly.
|
I'm trying to make it less of a pain in the ass. 
We do have queen energy written down, but it isn't being looked at currently.
|
There has been some debate regarding the criteria of larvae vs. minerals to determine the overall "economy" of a build. I thought perhaps it could be resolved by placing a constant conversion rate for one in relation to another. However, after considering my studies of economics, and the law of diminishing marginal utility, I recognize that the subjective value of a good relative to another is dependent upon the constraints the user is presently facing. To illustrate my point, consider the following thought-experiment constraints:
Suppose you have 1 larva and 50 minerals. Which would you prefer more of? Certainly you would prefer more minerals, as that would increase the utility or benefit derived from a single larva. Adding additional larvae would be worth nothing, since you do not have the resources to make use of them.
Now suppose you have 1 larva and 500 minerals. In this case, you would clearly prefer more larvae, since additional minerals would not be of much use given the larva constraint.
Therefore, the more valuable resource depends upon what is currently the greater constraint the player faces. So which is the greater constraint?
In an actual game, this answer would of course depend upon what the player was attempting to accomplish at that time: expanding, saturating, teching, attacking, defending... Up until now we have assumed our player's goal was simply to maximize long-run economy. My first instinct is to say that larvae are the current greater constraint, for we are beginning to stockpile resources faster than we can spend them on larvae. However, this problem could be assuaged by building additional hatcheries and queens, which could perhaps place the constraint on minerals.
The true constraint we have here is neither larvae nor minerals, it is mineral patches. If we had an infinite pool of mineral patches, clearly the 11 pool build would be the most efficient and economical. If we wanted a faster third or fourth hatchery, then we would prefer the 15 hatch build. I know from my own experience that it is very rare to want to take a third hatchery before the 6 minute mark in an actual game.
This is a difficult, if not impossible, question to answer. For now I think we have two very effective builds that serve to satisfy each constraint while still being generally efficient in the other. An earlier queen provides more long-term production capacity than a hatchery, and a hatchery provides greater resource gathering potential than a queen. Let us simply continue gathering data for now. That data, and not the theorycrafting, is after all the true value of this thread.
I am not content with the current pool of builds. We can't claim either of these builds are the leaders until we can test a greater variety of submissions. Go outside the box, if even to provide good evidence of what DOESN'T work, and why. I am still recieving requests to test more refined versions of previous builds. And we still haven't concluded the effects of delaying an exact build by supply timing.
For now I will be expanding the data sample timing to see if perhaps I underestimated the time required to overcome the initial investment of a second hatchery. If the results are significant, we can consider placing the data into a graph of some kind.
-jdseemoreglass
|
If you really want to take this serious, add scouting Drones (that eventually die), initial pair of lings, Lair-Tech (in time for cloak attacks?) and/or Metabolic Boost, and vulnerability to Overlord sniping. And of course test every build against every possible typical harassment or the biggest possible attack at the finish mark from T, P and Z. Also take into consideration the ability to harass a fast expanding enemy.
My bet is on 11 Overpool, ~18 Hatch. But 14 Ex, 14 Pool 21 Hatch might come close. A 1-Hatch-Muta revive after this test would be really fun 
Another build you might want to look at if you consider early game options:
+ Show Spoiler +10 Extractor Trick
11 Overlord
11 Spawning Pool
12 Extractor > put 2 on gas
15 Queen
17 put 1 on gas
18 Metabolic Boost
18 put 3 on minerals
18 Spawn Larvae
18 Hatchery > transfer 3 (10s)
18 Overlord
18 Extractor Trick
23 Spawn Larvae
23 Queen
25 put 1 on gas
25 Overlord
25 put 2 on gas
32 Spawn Larvae
34 Lair
|
On December 01 2010 02:51 slith wrote:
If you really want to take this serious, add scouting Drones (that eventually die), initial pair of lings, Lair-Tech (in time for cloak attacks?) and/or Metabolic Boost, and vulnerability to Overlord sniping. And of course test every build against every possible typical harassment or the biggest possible attack at the finish mark from T, P and Z. Also take into consideration the ability to harass a fast expanding enemy.
My bet is on 11 Overpool, 18 Hatch. But 14 Ex, 14 Pool 21 Hatch might come close.
Don't you realize this would add so many variables that it would make the gathered data convoluted and meaningless? In essence, you are asking to find the "best" zerg build, and that can't be done. There are subjective values to tech, scouting, etc. that cannot be measured in any meaningful way.
|
Sure, it does.
Just saying that testing economics only doesn't account for that nasty Terran opponent with his 1/1/1 or 3/0/0 who really wants to stop you from mining so hard.
And if you test economics, at least add gas to it at a reasonable time.. so we got all resources covered.
Not to say I don't like your dedication for this, really helps a lot in finding a good build. But finding the best build should be the goal.
|
I wonder how similar T and P best ecos will turn out
T gets mules, so an early rax might be superior for orbitals, but at what food
toss i can see as more linear w/ 15 nex since chrono doesnt need gateways
|
oo wow this is actually impressive and unexpected.
good to know that an overpool build is actually very economical as well.
thanks for carrying this on as well despite cynics who says hatch hatch hatch first is obviously the best build without testing anything.
Also, there is no way builds like 15 nexus will be viable anymore if zergs open pool first. This might revolutionize the game.
|
One build I want to try is some combination of double extractor trick hatchoverpool, or overpoolhatch, or pooloverhatch.
Hatchoverpool makes the most sense to me. That way you can then drone immediately after, and double-queen when the pool finishes.
|
I like how my informative thread got buried in garbage flaming posts.
In order to determine the best builds we must conduct another trial. Take your 3-4 best 6 minute builds and extend the trial time out by another 30 seconds. This should give you some very interesting results.
I suspect that the builds will get even closer to the same number of minerals mined. If this is the case then you need to shorten your experiment and retest at 5 minutes 30 seconds. Then compare the results again.
The scientific method is all about testings and retesting
|
On December 01 2010 02:38 jdseemoreglass wrote:
Suppose you have 1 larva and 50 minerals. Which would you prefer more of? Certainly you would prefer more minerals, as that would increase the utility or benefit derived from a single larva. Adding additional larvae would be worth nothing, since you do not have the resources to make use of them.
Now suppose you have 1 larva and 500 minerals. In this case, you would clearly prefer more larvae, since additional minerals would not be of much use given the larva constraint.
Therefore, the more valuable resource depends upon what is currently the greater constraint the player faces. So which is the greater constraint?
-jdseemoreglass
That is not true at all. 1 Larva is worth a lot more to the build than 50 minerals. You are not factoring in the opportunity cost of the Larva. The larva can be morphed into a drone. The drone can provide additional minerals over time. This can actually be calculated using Present Value of all future economic benefit. This calculation will be completely dependent on Saturation point as well. The point is that if the 50 minerals is not currently invested then it is not providing any benefit. Therefore you can calculate the value of 50 minerals by again taking the Present Value of the Future Value. The future value would then be considered by the time in which you would take to spend the 50 minerals.
The first present value equation however would need to be calculated like a perpetuities or an annuity that lasts forever.
If your looking for the most accurate results you will need to use Financial Concepts.
|
sigh..... you try to make a scientific thread and the first people to touch it are always these.
+ Show Spoiler + On November 28 2010 20:27 blitzkrieger wrote:
This is just a waste of time. Finding out the best economy does NOTHING except waste time especially since there are other limiters that will never allow you to use most of the builds that are better economy. For instance 14pool is generally standard to be safe because you are extremely likely to die if you are more greedy. You cant hardly go 15hatch now because of the counters.
For instance Fruitdealer goes (at least he did) 12 overlord with double extractor instead of 9 ovy because it is more flexible even though it is slightly less economical.
+ Show Spoiler + On November 28 2010 19:28 the p00n wrote:Show nested quote +On November 28 2010 19:27 ayadew wrote:On November 28 2010 19:13 mikell wrote:
.... obviously the best opening economy is 15/16/17/18 hatch. 14/15/16/17 pool. In a scientific environment dis is starcraft 2, bro
And of course, one of these.
+ Show Spoiler + On November 29 2010 04:24 Guerrilla705 wrote:Show nested quote +I regard to many of the comments, I want to clarify that this is simply to PROVE which general order timing will yield best economy in the long run ok that is a VERY vague statement. Like others have said, the best economic build is to never pool, you can't say that we are doing this build independant of other pressure, but we still need a spawning pool. If you are going to put in a spawning pool then what the other player is doing must be factored in, because Zerg is a reactionary race. We don't get to just start every game however we please, we must respond to what the opponent is doing. Also, you discredit yourself as a scientist if you are saying this single experiment proves anything. Science never PROVES anything, only attempts to disprove. If you want to make a scientific argument you need to recognize your uncontrolled variables, the problems with this experiment, and suggest ways to improve the experiment. It needs to be repeatable for others and they need to be able to do it better
And by the way
http://en.wikipedia.org/wiki/Scientific_laws
Science makes laws all the time starting with this exact same method.
|
|
|
|
|
|
EPT
