Hit them, don’t get hit: the key to good engagements

We all know how important macro is, but the most visceral and exciting parts of StarCraft (as both a player and spectator) come from micro in engagements. Commentators love to talk about army positioning. Stephano always seemed to make his units worth more than others with his army movement.

Over time, the community has invested a lot of thought here and layered heuristics and strategies around micro. These include ideas like “don’t attack into choke points,” “split your Marines against Banelings,” and “trigger Widow Mines with individual units.” A lot of this advice is generally correct, but there are exceptions and ways to over-apply these principles.

At its core, the idea of winning engagements is to hit them, don’t get hit. StarCraft is a game of numbers, and they don’t lie and in many ways lack the subtlety of real life. Damage is damage and should be applied liberally to enemies and avoided on one’s own army. Here’s how I see this applied in practice.

Always be shooting (except with Widow Mines)

Damage is damage, and you should hit them. Unlike Brood War, there is no high ground miss chance. Despite how much we talk about positioning, it is rarely worth it to miss taking shots for positioning. Thanks to the cooldown, you can kite and stagger movement to get positioning while taking the shots you need. It may not seem like much, but many engagements are won and lost based on the first volley (think about 2 Siege Tanks in siege mode firing at each other), but apply that at every point in an engagement.

Widow Mines happen to be a big exception here. Due to the cooldown, Mines can be more effectively used with good control. I myself lack the control to do anything fancy with them, so I just let them shoot early and let it be.

Arcs and flanks aren’t always better if everything is shooting.

Commentators love to point out how players will spread them armies to form arcs or separate their armies to flank. This sort of positioning is valuable because it exposes more of your army to fight, again on the principle that it lets you hit them. If you boil away the details, however, what really matters is how many units are firing on each side. If every unit on both sides is firing, the positioning is largely irrelevant: everyone is shooting away and doing damage.

The benefit comes when some of your opponent’s units are stuck behind and not a part of the engagement. That type of engagement is similar to ninjas attacking the martial arts hero one at a time: you’re hitting them with stuff while some of their units aren’t hitting you. For example, if you have your Marines in an arc around their ball of an equal number of Marines, you can imagine a few Marines in the middle waiting for their buddy in front of them to die so they can step up to fight. This is an effective engagement. If it’s all Marauders in the middle of their ball, however, everyone is involved at the same time, and the arc is less relevant and is just a good way to prevent escape.

Focus fire, but don’t over-focus

It’s better to kill 1 unit than to leave 2 at half health because you’re getting hit less. I think that should be obvious, but is often not really under our control given the large and chaotic nature of engagements. On the opposite end, however, you don’t want to overkill. If 20 Stalkers all focus 1 Marine at a time, they’re not hitting with roughly 75% of their shots. This typically, however, doesn’t happen. I believe even Siege Tanks have logic to prevent overkill.

It’s hard to control, but there is 1 notable case I should point out. Both Roaches and Hydralisks actually have melee attacks if they’re standing directly next to the unit they are attacking. It does the exact same damage, except the attack is immediate and not a projectile. The benefit of this is that it prevents overkill since your units won’t target and fire at units that current have deadly projectiles incoming. As such, it is worth it to stutter step Roaches up to melee range against, say, Marines and Marauders when possible.

Spread to minimize splash damage, clump to reduce surface area

When facing splash damage such as Banelings, Fungal Growth, Siege Tanks, EMP, Psionic Storm, or Colossi, you want your units spread out as much as possible to avoid getting hit. Conversely, against melee units and more generally, you want to clump to concentrate your army’s hitting power and reduce exposure to getting hit. As such, I think many great compositions mix these to exploit your opponent in different ways.


So those were a few of the examples I could think of based around this thinking. I know it doesn’t seem very sophisticated, but hit them, don’t get hit can pare away some fluff in micro and get to the core value of different engagements. Positioning in combat is key, but it isn’t intrinsically valuable: positioning is only important because of how it affects damage distribution.

Believe in the heuristics and advice on positioning: they’re usually right, but take the time to figure out when they’re wrong, too.

Great articles about engagements

First version of the ground-up tutorial

Check it out at http://starcraft.kevinleung.com/?page_id=430. Like my other guides, I’m happy to iterate on it, so comment if you see any big holes or have any tips on it. Otherwise, I might be iterating somewhat slowly on it while I refocus on my strategy guides for each race.

Also, check out the replays from DreamHack Summer 2013 and the 2013 MLG Spring Championship on the Spawning Tool. You can find build orders far more efficiently there than from my write-ups. Again, let me know if you would like to see a write-up for anything from those tournaments, and I can dig up the games and do some analysis there.

Roaches and Hydralisks have a melee attack w/ no delay or overkill


No delay means that Stalkers can’t blink away. No overkill means that all of your Roaches won’t waste extra shots when you focus fire, say, an Immortal. Additionally, because it’s melee, it isn’t affected by Guardian Shield or Point Defense Drone and is packaged with a different animation. It does, however, still count range upgrades.

How did I not know about this? And I thought that the Roach surround was just to stop kiting. I now have a good reason to micro my Roach-Hydralisk balls even more than just getting them all in range.

Prioritizing mechanics on the mental checklist

In Day[9] Daily #345, Day[9] suggests a process for improving mechanics: prioritize it at the top of the mental checklist at the expense of other things, see what it earns you, and slowly assimilate it into regular play. For Zerg, here are a few things I can think of that might work:

  • Creep spread
  • Overlord spread
  • Larva inject (as discussed in the Daily)
  • Expanding every 5 minutes
  • Never supply blocked
  • Army always split into 2 control groups
  • Constantly scouting
  • Keeping minerals below 400 at all times
  • Paying attention to alerts

Advice on Arranged 3v3 Ladder

Just played placement matches with Julie and my friend Jordan, and we went 3-2 to get placed into gold league. I’m pretty happy with that and thought they both played pretty well.

Quick observation: in 3v3, there are 2 main strategies: rushes, and air. In 2 games, we got rushed. In 3 games, we were attacked by air (1 banshee, 1 phoenix, 1 muta). So if you don’t get rushed, get defense in your mineral line soon.

The Macro Hatch is not to be underestimated

I once read somewhere (probably reddit), “I’ve never built a macro hatch and later regretted it.” I’m now a believer.

I’m traveling at the moment, and the best thing to do on planes and in the airport is to practice SC2 against the computer. I usually play against “very hard” and don’t have any problems, but today, in either my first or second game, I played a ZvT and lost. So I played the matchup again and was close to losing. Basically, the computer comes with a marine/marauder push around 6 minutes, then comes again around 9 minutes. I usually barely fend off the first attack with my 14 gas 14 pool 20 hatch opening, but I was taking a lot of damage. The 2nd attack was the real kicker as I would just be settling into my economy when it came.

I tried defending with banelings for the 2nd push, which worked well enough, but I felt like it was very early to be committing so much into it. After losing one 4-5 games in, I figured I needed a different approach. That time, I dropped a macro hatch right around 7 minutes. I can’t understate how much of a difference it made.

In the 2 games I played with the macro hatch, I just had more stuff. More drones, more zerglings, more money. I could tech faster while keeping a high zergling count for defense. Instead of trying to scrape together a defense for the 3rd push, I was desperately trying to keep my money low and oversaturated my main and natural expansion while getting ready for a counter.

I’ll need to play around with the timings on the macro hatch, since I have now only played about 4 games with it. Regardless of where it ends up in my builds, it’s a good reminder about how important of a resource larva are to Zerg.

Fun fact about my APM from Sc2gears

I installed Sc2gears a few days ago and took a look at the motley assortment of replays I have actually saved. I’ll be saving all of them now going forward, but for right now, there’s not much to look at. I’ll let you know if I learn anything important from my play.

Here’s one thing I did notice, though: APMs over time. APM (actions per minute) isn’t a perfect metric for how good someone is, but it’s okay. It’s not as important as some learners would suggest, and it’s not as useless as some pros would suggest. Fun facts about me, though.

I have a couple games from over a year ago. Back then, my average APM was 74, with an EAPM (effective APM: remove redundant clicks) was 61. Not great.

I played 2 games in the airport as I was flying back to the west coast for classes. That was at the end of my big spurt over break. My APM was 125 with an EAPM of 103.

I played 2 right after installing Sc2gears. My APM is at 114 with an EAPM of 91.

For what it’s worth, both games I played today were against the computer, 1 of which I let drag on for awhile at the end while I went for broodlords when I could have just won. Even so, both of the games at the airport were also against the computer.

Clearly even 3 weeks of inconsistent playing can be a drag.

Counting chickens before they hatch: proactive control groups

Zerg is kind of funny in that almost everything has a penultimate form: all units start out as eggs, and all buildings start out as drones. The nice part about that is that this means that things can be on control groups before they’re done. I see this being helpful in 2 ways.

First, you can put buildings on a group before the drone makes its way to the final position, which the other races can’t do. I’ve been thinking that one thing I can do is to put all of my tech buildings (spawning pool, evo chamber, spire, everything) on 0. That way, I never have to look at a building to start upgrades. This becomes very easy if I get in the habit of adding the drone to 0 when I send it to build.

Second, you can put eggs in a group before they’ve hatched, which other races can’t do (except maybe off of warp gates). One of the big mistakes I often make is bad rally points, either sending units to the enemy base long after an attack has ended, or not rallying new units to a push. While practicing my 10 pool (very necessary for multiplayer games), I started adding lings to my attack group once they were turned into eggs, which worked really well. This is also a relatively easy habit to get into: once you build all of the eggs you want, control-click them at the bottom to select only the new units (wouldn’t want to add larva to the group!), and then shift-add them to the right control group.

I think I have seen ViBE do both of these things before, but he also has insanely high APM, so it might be tricky to execute regularly.