[G] (T) 3 racks : Cheesing your way to GM league - Page 38

On May 23 2011 14:26 blah_blah wrote:


Your opponent is so bad that his marines didn't get to your base until about 6:30, your post basically has nothing to do with this thread.

On May 23 2011 15:57 LicH. wrote:
And your post denouncing him for posting a relatively informative post with graphics somehow "has something to do with this thread"?

On May 23 2011 16:43 blah_blah wrote:


How is it informative when you can't counter it that way against any sort of competent player (and, conversely, if your opponent executes it that poorly, you can literally do just about anything and win)?

On May 23 2011 15:54 LicH. wrote:


LOL

Combat ex, optikzero, perfect, and countless other tosses 4 gated their way to GM. So basically, cool story bro, tell it to me again some time.

On May 23 2011 10:03 Nimue42 wrote:


Pretty bad mid-Gold player here, but build order analysis is one thing for which my training is useful.

I posted something that might be relevant to your interests on screddit yesterday (right in the middle of the mod drama), including basically the builds you suggest and some that seem to be stronger, but I'll try and explain it better here.

For a preview, here's a build for 7 marines at 4:00 (assuming perfect micro, so, really a few seconds later, but anyway):
8/11 depot (stop building SCVs)
8/19 rax
8/19 rax [one SCV can leave to scout then]
8/19 - 14/19: 6 marines
14/19: OC
14/19-17/19: 3 marines

At ~4:00, you'll have an OC, 7 marines, 8 SCVs, and 2 incoming marines. However, I think that the following build, with 9 marines at ~4:10 is even more interesting:

9/11 depot
10/19 rax (stop building SCVs) [one SCV can scout]
10/19 rax
10/19 rax
10/19-19/19 train 9 marines
19/19 Orbital Command

That gives you 9 marines and 10 SCVs at 4:10, but you'll have to wait until your troops start dying before training reinforcements.

Now, taking a step back, it seems to me there are two basic questions regarding the marine-scv rush:

* How does building an OC affect the marine count, if at all?
* At what times is it most interesting to push out?

I study operations research, and that's exactly the sort of things operations research is about So, I wrote a mathematical model for that build, and abused computing power to not only provide good builds, but also proofs of their optimality. Obviously, in a game as complex as SC, there's no such thing as a *best* build, so the objective was to produce as many marines as possible by the given time limit: each completed marine was worth 100 points, and each incoming marine was worth the amount of seconds already elapsed (e.g. 20 points if it's completed 5 seconds late). Since the process was fairly computation-heavy, the program stopped once it'd proved that if there existed a better-scoring build than the one it had found, it was at most 5 points better (i.e., that any better build only got the incoming marines a couple seconds earlier).

Also, in the interest of realism, all the builds assume one SCV was off scouting from 1:30 to 3:00.

With respect to the OC question, my program explored three choices:

* No constraint on the OC (it may or may not be built; in the end, it was never built)
* OC must at least be started by the time you push out
* OC must be completed by the time you push out

And, finally, the program computed these builds for each push timing from 3:00 (180 seconds) to 5:30 (330 seconds), in increments of 10 second. The results are summarised in the following graph.



My interpretation is that having the OC completed before the push only hurts a bit (slightly slower reinforcements), except if you want to leave at 4:10 or at 5:00, in which case it costs you one marine. Eyeballing the builds shows that, basically, 4:10 hurts because you don't necessarily need a second supply depot yet, but do need the money, and, at 5:00 you want a third depot, and the OC hasn't fully paid off yet (iirc, I don't really care about a push at 5:00).

However, even in these cases, you can start upgrading your CC to an OC a bit before rushing, and it doesn't really penalize your DPS much, but you still get a mule to finance your SCV repairs and a potential recovery.

And now, that leaves the question of when should a Marine + SCV rush leave its base. Intuitively, I'd go for the 4:10 mark, with the OC started just before the push, but, again, I'm not a very strong player, and someone with a better grasp of the metagame would know better.

You can find and sanity-check the build orders in GraphStarcraft format here. Again, they were designed by a program for an idealised version of SC2, so you can assume that you'll be a couple seconds late.

On May 23 2011 20:45 3D.Hydra wrote:
I hate imba terran race, requires no skill to a click that high-dps army ;(

On May 23 2011 17:35 -orb- wrote:


Good point and agreed, 4gate is dumb.

Maybe what I'm saying is cheese is still silly strong in sc2 compared to bw it seems? Maybe that's just my perception but these cheeses are so easy to execute :\

+ Show Spoiler +

On May 23 2011 21:38 zatic wrote:
After hitting five Protoss in a row last night and going 0-5 I am determined to only use this from now on. Bring on the flames!

Good work with the guide OP!

On May 23 2011 22:06 zatic wrote:

Actually thinking about it I may have hit this last night. At least the guy went 3 raxx scv allin. But I didn't have too hard a time stalling with hellions and scv until blue flame was finished. Maybe he just screwed up the build. Is this supposed to kill factory / hellion builds?

On May 21 2011 20:57 Mactator wrote:
I don't understand how this thread can be 31 pages long! You are discussing a very simple allin build that has absolutely no potential and is easy to counter if you know it's coming. A 6 pool can be just as strong but seriously a zerg who only 6 pools is a joke. Why are you rushing for masters if you aren't good enough? I don't get it.

On May 23 2011 20:45 3D.Hydra wrote:
I hate imba terran race, requires no skill to a click that high-dps army ;(