Tuesday, March 29, 2011

Core Concepts of the Protoss Matchups

To celebrate my ascension to the almighty Diamond League (lol), I want to share some conceptual epiphanies I've had that drastically improved my play in various matchups. I feel like many players don't understand crucial and unintuitive specifics about PvX matchups, and this lack of understanding is all that holds them back from higher levels of play.

Protoss vs Terran
Scouting is the most important part of PvT, hands down. There are a few diverging paths the Terran can take, the largest being bio vs mech. Early game there are cloaked banshees and helion drops to worry about, and the way Protoss needs to respond to bio or mech in the late game are very different. Observer use will win or lose games against Terran, so it's important to get them. Countering bio and mech aren't difficult once you know the Terran has committed to them. So the most important building in PvT is the Robotics Facility. Observers, colossus, immortals. The other most important building in PvT is the gateway. I've noticed that it is just not tenable to have two-robo colossus or robo and stargate off of two bases. Gateway units are strong. Colossus are made of glass. They need to be in a supporting role, not a main role, within the army. Off two bases, there should only be a single tech building (robo, templar, stargate) with a collection of gateways and at least one upgrade building being used constantly. It's too fragile to add anything more until you're on three bases.

Protoss vs Zerg
Stalkers and sentries. Add immortals constantly against pure roaches, add Colossus against anything else. Build Dark Templar when you hit three bases. Be active with your army to force the Zerg to make units - don't commit unless the Zerg just isn't making units, at which point it's an easy win. Against an aggressive Zerg, defending a lot while building up and expanding will mean an easy win later in the game. Against a macro Zerg, they'll feel on the back foot all game long and eventually just die.

Protoss vs Protoss
This matchup is one I didn't understand for the longest time. I thought I could get away with a really technical build designed to hold the 4-warpgate rush and just win every game because I wouldn't die to 4-warpgates. The truth is that any counter-4-gate build is actually really weak against anything else.

PvP is based on the 4-gate attack. What this means is both players need to deal with the possibility of 4-gate. Neither player actually needs to do an offensive 4-gate, as often both will do a defensive build instead. But against a well-executed offensive 4-gate, there are only about three builds that will defeat it which aren't 4-gates themselves. The defensive 4-gate is probably the safest build with the best transitions after the 4-gate timing has passed.

After the timing for 4-gate has passed, or the 4-gate has been defended, the game can really only go in a few directions. The safest bet is to immediately make a robo and get an observer. The opponent can branch into a few tech patterns after the 4-gate. He can go for robo tech, which includes immortals or colossus. He can go for stargate tech, which usually means phoenixes. He can also go for twilight council tech, including blink, charge, and DT.

If you spot the opponent going robotics tech, you want to begin colossus production and make only zealots. On large enough maps, you can expand after your first colossus or two is out, because his re-enforce time for his colossus will be a huge disadvantage if he tries to attack. Otherwise, 1-base colossus pushes will probably decide the game. Do not get thermal lance until you've committed to expanding and the opponent is doing the same. At this point, get ready for dozens of colossus and war of the worlds. Teching to mothership late in the game is viable for a colossus toilet, which will end the game. Against immortals and stalkers, this is the best choice, but if the opponent is going for colossus before you commit to it yourself, you can throw down a stargate. He'll have very little anti-air and you'll get map control at the very least, allowing you to expand and harass. Blink stalkers or immortals is a good followup yourself.

Against a stargate build, you want to expand and get blink stalkers. That's actually it. Defend the harass while expanding, and you ought to come out on top. Use observers to get high ground vision and harass with blink, too.

If you scout a twilight council play, robotics play is going to be fine. If it's blink stalkers, go heavy on the immortals and do a timing push. If it's charge, a 1-base colossus timing push ought to end the game. You'll have mostly zealots yourself, negating the benefits of his charge upgrade, and you'll have colossus. If he's making a dark shrine, make observers and expand. Perhaps a cannon at your natural, depending on the map. You should gain a commanding lead and be able to win with virtually anything. Pure zealot+stalker with upgrades is a fine choice.

And that's all I know about PvP. Early game is defensive 4-gate, then the mid-game transition is a sort of rock-paper-scissors game. Use the observer to respond to the opponent. Don't expand until you know he isn't going for a robo-based timing push. 1-base 2-colossus timing attacks will beat a good number of followups. Far too often I see opponents who get past the 4-gate timing, and then begin expanding while getting colossus AND getting range. A 2-colossus timing will actually kill that. It should be your standard followup until you see otherwise.

Saturday, March 19, 2011

The Development of a Build Order - Part 4

Day 4.

Experimented against Muta-Ling a bit more. I believe I definitely need to research hallucination to spot the spire in time. If I wait until my first few stalkers to push, a Zerg could have a good number of mutalisks and zerglings already on the field and force me into an early defensive posture. Now I need to determine when to begin researching hallucination and whether to chronoboost it. Gas is at a premium before I take my expansion nexus, because it is all going into sentries. Directly after the nexus goes down and I have 7 or 8 sentries seems to be the reasonable time to begin the research. I'll need to examine whether it has to be chronoboosted in order for a phoenix to scout the spire in time. I doubt it.

If I scout an early spire, I believe I can skip the robotics facility and put down a fifth gateway and push, which should win the game. If I can't succeed with the timing attack, as has happened because of scouting the spire late, I have to get blink very quickly. I have been making the templar archives off of two bases. I don't know if I can survive purely on blink until I get my 3rd base. I know that iNcontrol also favors dark templar to regain some map control, which is a solid idea, but the gas consumption conflicts with the high templar tech. It may be a better response than the high templar. I'll need to try it out in the future.

I did have one game against an opponent who made pure roaches with burrow. I was able to crush it with good forcefields, stalkers, and a single robotics facility making immortals. Once on three bases I added high templar, anticipating a hydralisk switch, but none came, and I walked in and killed his roaches. A large stalker army with some immortals and templar mixed in can dominate roaches, and forcefields help immensely as always. I also decided to push much more aggressively all game long, relying on forcefields to escape if he had too many units. This kept me economically even or ahead.

Wednesday, March 16, 2011

The Development of a Build Order - Part 3

Day 3.

One notable game to report on today. It was a very nonstandard game, so the build and timings were not very helpful practice. My Zerg opponent opened with a fast roach warren off of one base, prompting me to put down a forge and cannon early. Upon seeing the cannon, the Zerg expanded, and I proceeded with my 3-gate expand normally. My observer was unable to scout much beyond two bases, a roach warren, a hydralisk den, and two evolution chambers. As I had seen a roach opening, I expected roaches, and began immortal production from a single robotics facility. I researched hallucination but never used it once, which was a big mistake. As I moved out to poke and scout the Zerg's army, he dropped a large quantity of hydralisks in my main. I cleaned up his army but many buildings were destroyed in the process. I figured we were about even, economically, because he had made so many hydralisks unprompted. I also discovered in engaging so many hydralisks without colossi that guardian shield makes an absurdly big difference. It is 100% necessary whenever engaging hydralisks.

I moved to take a 3rd base while remaking my buildings. I forgot to begin my templar archives for a very long time, which was certainly what lost me the game. I didn't have storm researched until well after my third base was up, and I continued to try to engage hydralisks with charge zealots, which does indeed work very well. The rest of the game wasn't very relevant to my strategy, except that Zealot+Templar seems very viable against pure hydralisks. I simply didn't have enough of them, fast enough, and I continued to suffer from drops in my main, hurting my macro a lot. From watching the replay, my interpretation of the game state at all times was fairly good: the Zerg played aggressively with large numbers of hydralisks, and I thought (correctly) that we were economically very even, or I was ahead. This was true, and some Dark Templar harass actually brought me substantially ahead later in the game. What hurt me was my lack of production buildings, because they kept getting destroyed, and not enough high templars. The game was surprisingly close, but I simply couldn't quite catch up to my opponent's hydralisk army.

The game

Interesting fact: a single psionic storm brings the health of a hydralisk to 1 hp. This makes me think that as soon as Templar are on the field, I should switch from Zealot support to stalkers. Beyond that, this game has made me think about the ordering of the tech route I want to take. Early roaches should prompt me to go stalker+immortal first. This combination then prompts the Zerg into hydralisks. At this point I should already have psionic storm, which should be enough to beat virtually anything the Zerg can throw at me at that point. If, on the other hand, the Zerg opens with Hydralisks fast, I have to go for Zealot+Templar to open. If I execute this correctly, I hope it will be so dominating that the Zerg will need to get air units or roaches. Mutalisks are weak enough to high templar that I should be able to safely transition to blink stalkers. The most dangerous counter after fast hydralisks would be a very fast tech to brood lords, which require a large number of blink stalkers to counter, or void rays.

While my path from stalkers to immortals to a third base to templar seems reasonably clear, my path from zealots to templar to a third base to immortals does not. I'll need to work on rushing to storm and charge off two bases and trying to stay alive. An obvious problem here, as well, is knowing that a Zerg is going for fast hydralisks. If the Zerg began making hydras or just the hydra den early, but then made roaches or even mutalisks, I would be in some serious hurt, I think.

Saturday, March 12, 2011

The Development of a Build Order - Part 2

Day 2.

My practice partners were pretty busy today but I got a couple PvZ's in, between ladder and custom games. The rest was practicing the build order execution and some specific timings against an easy AI, as well as some testing in Unit Tester Solo.

In a platinum ladder game on Shattered Temple, I scouted roaches during my initial push coinciding with the observer. I did not use hallucination to scout. I went for double-robo immortals off of two bases upon seeing roaches. My 3rd base felt delayed and I was unsure of when to take it. I was surprised how much money double robo immortal uses. My opponent made roaches and zerglings and went for hydralisks later. The immortal and stalker combination seemed very powerful with decent forcefields. However, after perusing the replay I discovered that my opponent simply didn't make more than 40 drones all game long. I won by superior macro, not necessarily by the strategy. Still, I was able to discover some aspects of my build in a live scenario.

Replay

In a custom game against a diamond Zerg friend, I again poked at the front once my observer was out with my army. I scouted spinecrawlers and zerglings, making me suspect mutalisks. I then used my observer to scout to confirm the spire. I rushed for a twilight council for blink, as well as armor upgrades, while massing stalkers. My 3rd base again felt late - I took it while close to maxed. My templar archives also felt late - I didn't get storm before the game ended. Against good macro, I don't know if I would have won. The zerg didn't drone enough and didn't harass with his army at all, and a single push was able to win the game with good force fields and an excellent chokepoint engagement area. A goal of my strategy is definitely to be adaptable to roaches, hydras, lings, mutas or banelings, depending on my scouting, and I feel like I did that in this game. Still, I want to get a 3rd base sooner and I want psionic storm earlier. I'll focus on these points in the next few games.

Replay

In Unit Tester Solo, I wanted to find out just how powerful equal-resource armies of immortals and zealots fare against roaches and hydralisks. Immortals in decent numbers are ridiculous against roaches, as I expected. Chargelots against unmicroed roaches are actually cost-effective, which surprised me. Immortals are not terrible against Hydralisks in combination with stalkers, but it's not something I would want to stick with in a long game against mass hydralisks. A surprising find was how good sentries were against hydralisks. Sentries are actually an important damage dealer against Hydralisks, in combination with other units, but they die quickly to hydras like everything else. 8 sentries can actually kill 5 hydralisks.

By far the coolest find, however, was the effectiveness of zealots against hydralisks. Unlike roaches, hydralisks can't micro against zealots, and charge is a solid upgrade in this fight. Zealots are incredibly cost-effective against hydralisks, as long as no roaches are present in the fight. I'd like to start experimenting with staying alive against a hydra rush by using sentries and rushing to charge on my zealots while teching to psionic storm. I'd like to use a comparison to Protoss vs Terran here: in PvT, Marauders are weak against zealots and strong against stalkers, and they are the tank of the bio army - they absorb the most damage. The marines are good against zealots, while having very little health. In PvZ, Roaches are the tank unit equivalent to the marauder, except they are strong against zealots and weak to stalkers. Hydralisks are strong against stalkers and (apparently) weak to zealots, and have very little health. Both marauders and roaches can micro against zealots. And unlike marines, hydralisks cost a significant investment. I'm very excited about the possibilities here. It is also difficult for the zerg to transition from hydralisks to mass mutalisks, and the Zerg will not have enough gas to field both units in large numbers. Overall, reacting to a hydra rush with fast chargelots seems like a cool idea that I need to test more. Naturally storm will be necessary when the zerg has too many hydralisks, and when roaches are on the field, because the zealots won't be able to reach the hydras before dying.

Friday, March 11, 2011

The Development of a Build Order - Part 1

I have a strategy I want to develop in Starcraft 2. The opening builds and strategies that I usually employ are created by professional players far better at the game than I. However I got an idea for a strategy, and this time I have no strict and specific build order to follow. I have to develop and test and hone timings on my own to discover if the build I want to create will actually be viable. I may need to abandon this build if testing proves it won't work. But without testing, it isn't Starcraft, it's Theorycraft. The next few posts on this blog will document the development and testing of this build, with replays and possibly commentary videos included.

"So what's the damn build already?!"
This is a Protoss vs Zerg strategy for the Protoss. The typical Protoss style which I use to great success is the Blink Stalker + Sentry + Colossus death ball, with a 3rd base Dark Templar/Archon transition. Extremely safe and well-rounded strategy. It's good against just about everything the Zerg can throw at it and transitions well throughout the game. But is this the BEST strategy against everything Zerg can throw at the Protoss? I'd like to find out.

In a nutshell, what I would like to develop is a strategy involving Immortals and High Templar. Immortals are one of my favorite units and I feel like they are criminally underutilized. With 50 damage against armored targets and a hardened shield, Immortals are serious threats to any armored unit. Typically we only see these in large numbers against Terran mech - Tanks and Thors - but both Marauders and Roaches are armored as well. With the standard Protoss PvZ strategy, the answer to Roaches is Stalkers and Colossi, neither of which is particularly strong against the Roach. Combined together, however, they can do OK. I want to do better than OK! I want to wreck roaches entirely. And with the Immortal in larger numbers than we ever typically see, I believe that is possible.

So the Immortal will be the key to defeating mass roach, but what about anything else the Zerg throws our way? Zerglings and Hydralisks are light units with high DPS which circumvent the Immortal's hardened shield. In the standard Protoss strategy, the Colossus deals with these two threats. If the Colossus count is very low because the Zerg keeps making Corruptors, Hydralisks and Zerglings can be devastating to the remaining Stalker army. It can be hard for a Protoss to deal with a Zerg who overmakes Corruptors and then masses Hydralisks. Alternatively, many Zergs will play with Zerglings and Mutalisks, or even Zerglings and Banelings. There just isn't a strong answer to large masses of Zerglings using a standard Protoss strategy. There are only ways to do OK against these Zerg strategies and eke out an economic victory. That's not good enough for me. I want to crush my enemies!

That's where the High Templar comes into play. Psionic Storm will melt Zerglings almost instantly. They soften Mutalisks substantially, and are a critical way to keep the Mutalisk count low enough for blink stalkers to deal with them. Storm deals with Hydralisks because they can't micro out of the storm easily, and don't have a lot of health. What storm is weak against is armored targets - the Roach, the Brood Lord, and the Ultralisk.

It is my belief that the Immortal deals with armored targets better and the High Templar deals with light Zerg units better than the more general-purpose Stalker + Colossus army. Now that I have this strategic concept, I need to fashion a gameplan around it. What are my goals with this build?

1. A safe fast expansion
-I want to win a game of any length using this build. The units I need to make require a lot of gas. I'll want three bases reasonably quickly both to get the economy I need for the units I want, but also to stay even with the Zerg's economy.

2. Immortals in time to fight mid-game Roaches
-If we scout Roaches, we need to be able to create out Immortals in time. This means a reasonably fast Robotics Facility.

3. Ability to scout the Zerg's tech in time to determine my Templar timing.
-What the Zerg is doing determines whether I start to go mass Immortal or High Templar off of two bases, and when to start getting Templars. I need to scout what Zerg does soon after taking my natural expansion.

I now have a very basic set of goals that my build needs to meet. As a general framework, we're looking at some kind of reasonably fast expansion to open. I'm going to elect for a 3-gateway Sentry expand, which is a standard and very safe Protoss expansion build. I feel that forge fast-expanding is too map-dependent and too risky at this point in the development of my build.

Upon taking my natural expansion, I need to scout the Zerg's tech, or optionally to force the Zerg's tech. There are four ways to do this:

1. Hallucinated Phoenix
-For the reasonably small cost, we can research hallucination and send fake Phoenixes across the map all mid-game long. This is a versatile and long-term scouting solution, offering more uses later in the game.

2. Observer
-We could build an observer to scout the Zerg with. The problem I foresee is that this is a slower scout method and we want the observer to be with our army. We also may want to utilize the Robotics Facility for Immortals early and often.

3. Poking the Zerg army
-Poking at the Zerg army will force him to make units and show what units he's building. This is a strategy we ideally want to do all game long to control the Zerg's economy. The issue with this method is that the Zerg can hide tech, like Mutalisks or Hydras, until it is too late to spot.

4. Forcing Roaches with a Zealot + Sentry attack
-I like to think of this as the Huk method. Upon expanding, we want to attack with pure Zealots and Sentries and a proxy pylon. Against a Zerg making Zerglings or small numbers of roaches, this attack can actually win the game, but our ultimate goal is to force the Zerg to make large numbers of roaches. The timing will hit the Zerg before Mutalisks come out. The issue with this method is how risky it is. Forcefields have to be perfectly placed and the Zerg can blindly kill the Protoss by building a lot of units early.

I'll need to test all of these methods, but my goal is a tactic which will determine what the Zerg is doing and give me time to respond via 2-base Immortals or High Templars. What timings I need to learn to defend specific midgame Zerg attacks I will have to discover through testing. The next question is when and how I will take my third base. With the standard Protoss style, the key timing is when the first Colossus is produced, because it immediately gives the Protoss some amount of safety. Without Colossus, it seems like a key timing will be the completion of Storm research, perhaps the completion of forge upgrades, or the beginning of double-robo Immortal production. This is something I need to discover through testing.

So now I have a basic framework for a build order, based on the goals of my desired strategy. 3-gate Sentry expand, some method of scouting, a robotics facility, and then a response to scouting - either double-robo Immortal or High Templar and Storm. After one of these responses, at some point, we need to take a third base. Once taking a third base, we can complete our Immortal + High Templar gameplan. Potentially we can move onto stargates or Dark Templar. This is when we might get Archons out, as well. This is the build order I intend to figure out through testing and playing with practice partners. Stay tuned for the next post, detailing my testing and experiences developing and refining this build order further.

Any diamond-level Zerg player that wants to assist this process by acting as a practice partner should message me on Battle.net! My ID is "GomJabbar", my character code is "678".