[G] Positioning, Formations, and Tactics - Page 5

On April 06 2011 13:33 TimeSpiral wrote:


I'm not sure how to quantify the advantage, to address your question of "how much" specifically. But you also need to keep in mind that there is no real "back" to a unit in Starcraft 2 because all units are aware and have vision in a 360 radius. There is only a "back" to a formation that is oriented in a specific direction.

Option (b), the pincer, is far superior in this scenario. You are ...

(1) Reducing his ability to maneuver by closing from two angles.

(2) Dispersing the Siege Tank fire in multiple directions instead of focusing the splash on one big ball of units

(3) You achieve the "full surround" about twice as fast as you would with a full frontal charge (even if it is from the "back").

(4) You create the psychological impact on your opponent of being surrounded. This can cause him to scramble and fumble at the controls.

Hope this helps.

On April 07 2011 21:36 iChau wrote:


So it's worth it even if you have slow lots. I see, thanks. :D

I actually lost twice to this terran because of positioning. He abused the high ground from the low ground with medivacs and siege tanks, and it was annoying, especially at the taldarim altar's natural. I think I should have moved out fast so I can pincer him before he scans my army, and this also makes it so he doesn't know where I am.

You can visualize this by looking at Fig B.2 and seeing how all the firepower is concentrated into a single point. Additionally, the arc maximizes the number of troops able to engage because they are in range of the enemy's front line.

EDPS=k_1 * DPS_base * Life_remaining/Life_initial
    ≈k_2 * DPS_base * Life_remaining/Life_initial*{1 if a target is in range, 0 otherwise}
    ≈k_3 * DPS_base * Life_remaining/Life_initial*Number_of_targets in range

If you can Pincer your opponent's army (aka the Double Envelope) you will likely create an offensive advantage in the encounter.

On April 08 2011 03:30 Jagd wrote:
First about the 4.5 'exposure' of spread versus single balled units. As was pointed out yes the 4.5 comes from a particular split, and you can use derivatives to get the accurate amount for arbitrary splits. However someone complained that you didn't take into account creep, speed, etc., and I don't think this is a problem since you're talking about 'exposure' and not about 'effectiveness.'

Second, about concave vs. ball battles you state:

I think really what's important is the combination of these points. I.e. the fact that at the initial moment when you engage if you have a concave to your opponents ball then the first 2-3 shots your units fire will be focused on his front units while he is moving in to get all his units attacking, so
a) he is missing shots from his rear units if they are out of range (but not from his front units if he can move/stop correctly)
b) damage is spread unevenly over his army.

Point b) is the interesting point because it's saying that in general it's worse to loose 500 life on 5 units and 0 life on another 5(group A) than 500 life on 10 (group B). Suppose we're talking about ZvZ so the average life of a unit is 145, and suppose the average DPS of a unit is 10, also suppose all units are in range of each other. Suppose the effective DPS (EDPS) of a unit is proportional to the unit's base DPS times the percentage of health they are at.

EDPS=k_1 * DPS_base * Life_remaining/Life_initial
    ≈k_2 * DPS_base * Life_remaining/Life_initial*{1 if a target is in range, 0 otherwise}
    ≈k_3 * DPS_base * Life_remaining/Life_initial*Number_of_targets in range

The intuition for EDPS is that the effective amount of damage a unit does depends both on how much damage per second they can do, and how many seconds they are alive for to do that damage, and whether they can attack something. Whether the above formula for EDPS is correct though would require some statistical analysis of simulation results (i.e. maybe it's DPS_base^2 *c_1 +Life_rm/Life_0 * DPS_base^0.5, If someone is interested in testing this with me, and is good with the map editor, let me know.) Then for group A if 500 damage is spread evenly among the 5, then each has 31% HP left, so the EDPS of the group is 5*0.31*10+5*1.0*10=65.5, while for group B each has 65.5% HP left so the EDPS of five of those units is 10*0.65.5. So in fact whether damage is spread or not depends on whether EDPS is a linear function. In the definition I gave it is linear, so in this case it doesn't matter. You could argue that if you have a concave it is easier to focus fire with small groups of your concave so that for group B the 500 damage will be spread evenly, while for group A it will not be spread evenly (i.e. 3 roaches dead in group A, 7 full health), but again this will only make a difference if EDPS is not linear, in fact if you focus fire and overkill then you actually hurt yourself no matter whether EDPS is linear or not. Thus for now we should only say that concave is more effective because it is likely that if you have a concave more of your units have an enemy in range.

Third, about pincer/flank you state:

Bearing the previous argument in mind, if two armies are of equal size and equal range then having a pincer will not gave you an advantage no matter whether EDPS is linear or not. The case where flanking is most common is for Zerg, because flanking allows melee units to take shorter paths in order to get into position, which increases their EDPS as defined above, and that is the only and most important element of flanking. Of course flanking can prevent kiting and retreat as well, but those are tactical considerations and not properties inherent to deciding who will win in formation A vs formation B.

Finally, thank you for this thread, it's provoked me to think about this in my own gameplay:
-infesters/ultras and cracklings (HTs/Collosus and chargelots) make for a powerful combination because the formation that is most effective against one is countered by the other.
-in zvz crackling/infester will be superior to roach/infester once both sides have enough fungal growths available and the crackling player can set up good flanks.
-roach/hydra is superior to pure roach if used defensively, because defensively hydras can be positioned such that they will attack before roaches in their position would attack; offensively since hydras are slower they may not get into position quickly enough so their added range won't be a bonus, and roaches have slightly higher EDPS than hydras.

Bearing the previous argument in mind, if two armies are of equal size and equal range then having a pincer will not gave you an advantage no matter whether EDPS is linear or not.

On April 07 2011 22:40 TimeSpiral wrote:


Well, from a gameplay perspective, I think you are considerably higher level than I. But, to address your example, there are a few tactics you can use here.

That particular tank push is very common and powerful on T'DA. So, park a probe at the end of his natural's ramp. The moment he moves out, set up your Pincer attack. If it's the 2rax-reactor tank push it's getting to your base before charge or thermal lance unless you are doing some crazy rush. It's rarely ideal to attack into siege lines (obviously) but defensively you have to.

The tactical goal for this Pincer is to attack at the moment, or just slightly before, his tanks are activating siege mode and the SCVs are working on the bunkers. It's important in a Pincer that the two divisions of units are roughly the same size. So, take the number of gates you have into account when you're setting up the maneuver (overload the far team and warp in for the close team).

Both tanks and marines are really, really bad versus Zealots w/ Guardian shield. Your Terran opponent knows this, which is why he's leveraging the terrain to his advantage and constructing bunkers because the siege tanks remove much of the mobility of the marines. If you have robo-tech, even better. Spot the "kill zone" with an observer so you can see the exact siege time to 1a2a into his force.

I can almost guarantee you will crush the army losing almost nothing (assuming it's the push I'm talking about). As the game progress and the tank numbers get bigger this changes a little bit.

Sorry for such a long response :/

On July 18 2011 09:11 Kiaro wrote:
If this is true, then why do all protosses move their collosus/stalker army around in a big ball instead of spreading it out?