EPTI'm a 500ish zerg player.
A pretty common play I have to deal with at my level is the one-base all-in. Seriously, like 99% of games are one-base all-in plays. Trouble I was having was primarily centered around 2-gate zealot all-in, or 4-warpgate all-in, occasionally marine-tank allins would get me as well.
I tried a number of common openers. 13 pool, 15 hatch. 15 hatch, 15 pool. 14 gas, 13 pool. etc. All very similar, all basically based off the theory that you want to get your expansion up as fast as possible, and this is expected to give you a strong economy. You simply have to hold off the opponent and you win.
Now this was, in fact, the case, but the issue I had been having was holding off the opponent. When this inevitable one-base all-in push came, I wasn't ready for it. If I had spine crawlers, they had tanks, or just so much stuff that even a bunch of spine crawlers wouldn't manage it.
I started to consider the benefits of the fast expand. What does it do for me? Well, it gives me access to 8 mineral patches, 2 gas, and a larva every 15 seconds. It gives me 4 larva every 40 seconds if there's a queen managing it.
I started to watch my replays when I'm losing and take note of army sizes and income. What I found was consistently either I had such a tiny army that I'd get run over by the first push, or I had a suppressed economy such that the protoss player simply was easily out-pacing me despite the fact that I was on two bases. I thought to myself "Man, why is this protoss able to consistently out army me and out-econ me?" and I couldn't come up with a good answer for that. I mean, I had two bases, I had two queens, I was managing my spawn larvas correctly, but still, consistently, I either had a weak army, or a weak economy, and even if I tried to balance them both, (say I tried to match the protoss economy at minimum) I would end up significantly behind in army.
At first I just thought it wasn't fair. The fact that zerg needs to build military and buildings at the expense of drones just made things pretty difficult, especially early game, where you need to start building your tech buildings, and you have limited larva production. Also, you have no real means of early harassment for suppressing your opponent or limiting his worker count. The protoss or terran player builds a consistent number of workers, and never has to worry that if he builds a marine, that that's an SCV that he will be unable to build.
So the first thing I did is sit down and think about economy. I thought about why it was 1-basing players were beating my economy while I managed to fast expand. My issue there was simply drone count. I was underproducing drones to be able to survive.
Probes and SCVs both build one every 17 seconds. Larva from a hatchery spawns every 15 seconds.
This means a terran and a protoss can build 0.0588 workers per second.
Zerg can build 0.0667 workers per second.
However, when you become at risk of being supply capped, you need to build a supply feature for every 8 workers. For Terran and Protoss this doesn't impact your ultimate worker count, for terran it temporarily suppresses the number of SCVs you have mining, but once that depot is done, you reclaim it. For zerg, however, building an overlord consumes a larva. So when you consider that 1 of every 8 larva has to be an overlord to maintain drone production, you get 8 drones per 135 seconds.
So in reality, zerg can build 0.0592 workers per second off a hatchery.
This means, off a single hatchery, nexus, or command center, you get about the same number of workers produced for all three races with a tiny bonus in favor of the zerg.
--- now ---
In the case of protoss, they have chronoboost. Now, a chronoboosted nexus can create probes at one every 11.3 seconds, or 0.0882 probes per second, but you can only have the nexus chronoboosted half the time because of energy constraints. A nexus that has all it's chronoboosts spent on itself can produce 0.0735 probes per second.
In the case of terran, they have MULEs. A mule lasts slightly longer than it takes to create another mule (it takes about 80 seconds to get the energy to create a mule, the mule lasts 90 seconds) So it can be assumed that a mule is up 112.5% of the time after the OC is built. A mule is about as effective at mining as slightly fewer than 4 workers. That means, you can consider the mule's impact as having a consistent 4 SCV advantage over the opponent.
In the case of zerg, they have spawn larva. This is a very different ability because it allows increased unit production, but since larva is used for so many things, you can't simply count it as drone production. In short, spawn larva allows the zerg to make 4 units every 40 seconds, or 0.1 units per second. This allows now a drone every 6.7 seconds, or 0.148 drones per second. However, this needs to be metered because drones built are at the expense of military.
--- so ---
We sorted out that in theory, zerg can outproduce both terran and protoss, terran has the mining advantage on one base because mules don't count towards maximum saturation, and protoss has the early economic advantage because chronoboost starts having an impact at 10 supply, which is before the orbital command or the queen start getting into production.
Now lets take into account military production. If the game were centered around drone production, zerg would win the race, but a few zealots or hellions thrown in the mix and your massive drone count means nothing without something to defend them.
So, consider a gateway. A gateway will build a zealot (2 supply) every 33 seconds. It will build a stalker (2 supply) every 43 seconds. This means a gateway is producing 0.0303 units per second if working on a zealot, or 0.023 units per second if it's working on a stalker.
Now, consider a 2 gateway push. We'll make the odd assumption that the protoss is spending chronoboosts on probes and using two gateways for an early zealot attack. This means the protoss is creating 0.0735 probes per second, and 0.0606 zealots per second. In order for a zerg player to keep up with that rate of production the zerg player needs to manage to have 0.1341 units per second. Great, that's totally possible, because a queen and hatchery produce 0.1667 units per second. But you need to take into account a few other things, one is buildings. Early on, you're going to need to build an extractor or two, a spawning pool, maybe a roach warren, maybe a hatchery. We need to take those lost larva into account. Likewise, you need to build overlords for your army. If we consider for the early game you build on average one building per minute that loses you about 0.0167 units per second. So your larva rate with building accounted for is about 0.15 units per second.
So lets assume we keep up with the probe production, to make 0.0735 drones per second, that will cost us 0.0827 larva per second when we consider overlords as well. That leaves us with 0.0673 larva per second for military. This means we can make 0.0598 zergling pairs per second, or 0.0538 roaches per second. Building 2 zerglings per zealot isn't going to be enough, but building 0.89 roaches per zealot probably will stave off the attack. Likewise, you could start to build 0.0673 spine crawlers per second, because they don't consume supply.
The issue with zerglings is that you won't have enough larva to keep up without cutting drones. The issue with roaches here is that you will be spending gas, while your opponent will be pooling it. The issue with spine crawlers is that they can not be used for a counterattack, and can be maneuvered around.
Likewise, this is simply a 2-gate push. Assuming you have that near equal number of roaches, and the 2-gating protoss doesn't throw his zealots away and simply backs out, he'll soon have 4, and up to 7 gateways. For this example, I'll say he builds 1/2 zealots, and 1/2 stalkers from his gateways, he can make 0.0263 units per second from the gateways. Once he gets to warpgates, he can make 0.0357 units per second. 4 gateways plus probe production means you'd need 0.1787 units per second to match his production. 4 warpgates would mean you'd require 0.2163 units per second, and 7 warpgates would require 0.3234 units produced per second to simply match in unit generation.
So let's consider what we'd need to use speedlings to manage this attack. Speedlings good because they cost no gas, in the event of a successful defense, they can be used for counter-aggression, and in the event of a rout, they can chase down the fleeing opponent. In general, you need to have 4 times as many speedlings as the opponent has zealots to manage a small force. As the zealot force gets larger and more difficult to surround, the ratio goes up, but we'll consider the small force. To defeat 0.0606 zealots per second, you then need about 0.1212 zergling eggs per second. For 0.1212 zergling eggs per second, you need 0.1515 larva per second. To get this force, and maintain income parity, you need 0.234 larva per second.
0.234 larva per second is more than one queen and one hatchery. It's approximately what you would get from 2 hatcheries, one queen if you ignore structures. So let's consider that approach. Putting down an additional hatchery gives you larva ample enough to build speedlings to match two gates worth of zealots.
So I'm going to ignore the ability to fast expand for the moment. We're just going to plop that hatchery down at our main, because my goal at this point is not to overtake my opponent, it's simply to see if I can match what he's doing, because my experience previously showed that I always came up behind. For the time being, I'm ignoring resource spending, but I'll just do a quick sanity check to make sure I'm not completely overextending myself on mineral expense with this second hatchery. The hatch in my base costs me 300 minerals, plus I'm down a drone which cost me 50 minerals, I've also a queen, which is 150 minerals, but provides defense. I've also spent 200 minerals on my spawning pool. In total I'm looking at 700 minerals expended. In contrast, the protoss player has spent 300 minerals to make two gateways, and is thus 400 minerals ahead, but does not have the queen for defense.
At the point I make that hatchery, then, I will be down about 400 minerals, but I will be able to maintain production parity with the protoss player. In the game this is relatively unimportant, as you can not spend money fast enough on two gateways for it to. You can build zerglings off two hatches, 1 queen, and drone as fast as the protoss match the zealots at a 4:1 rate, and still have money.
So, 2 gate zealot vs 2 hatch 1 queen zerglings is about perfectly equal in small numbers. Interesting, and good to know. Also, this means the zerg defender has an advantage, as he has a queen, and additional time to morph units while zealots are en-route. Likewise, when the attack is in your base, you can ramp zergling production up in place of drones, and replace those drones after the incident is over. Likewise, unlike roaches, if the zealots see the lings and retreat, you will kill them all and suffer fewer losses than if they committed fully. Good stuff.
So let's consider the implications of a larger gateway count. Assume our protoss player decides to commit to the attack, our armies clash, his zealots all die. My zerglings all die, my queen gets the last hit on the last zealot. He decides to up his gateway count to 4 gates, and go stalker/zealot and warpgates complete. What is it I need to deal with that? 4 warpgates + probes means he's making 0.2163 units per second. Or simply 0.1428 military per second.
Now, let's consider a couple of options. One option is to go roaches, the other is just to power through with zerglings. Straight zerglings get a bit more complicated when the mix is zealot/stalker. If weapons aren't upgraded for protoss and we'll assume they aren't for now, the zealots will nearly kill the zerglings in one shot, and a stalker hit will finish them off. Since the stalkers can fire from behind the zealots, this makes them quite effective. However, the stalkers are at a bigger risk to the zerglings alone, and zerglings vs. stalkers normally resolves in the favor of zerglings, at least until blink. Again, however, the mass of zealots and stalkers often results in a larger ball, where the zealots will migrate to the outside of the ball. Any time ranged units are unreachable inside a ball, it tips the battle in favor of the ranged attackers. We'll ignore that for this moment just to look at what would be needed to match the supply investment of the protoss.
For the zerg to match the protoss zealot+stalker investment, he would need to make 0.2856 zergling eggs per second. Counting overlord requirements, that is 0.3213 larva per second. Considering what's needed to maintain economy parity, you need 0.404 larva per second generated. That's a lot, about what you would get from 3 hatcheries, 2 queens. 3 hatcheries and 2 queens costs 1350 minerals, while 4 gateways cost 600 minerals. This is going to have a pretty noticable impact on your economy and is unreasonable if you aren't already ahead.
For the zerg to match the protoss zealot+stalker supply with roaches, however is more reasonable. You would need 0.1428 units per second, which for roaches, translates into 0.1785 larva per second. Considering drone production as well, you need about 0.2612 larva per second. This is a bit more than 2 hatch, 1 queen, but a bit less than 2 hatch 2 queen. This means you could supplement some more larva-expensive zergling production and have a better balance against a zealot+stalker force off 2 hatch 2 queen, as a protoss with 4 warpgate, and match worker production.
I'm not going to go much past lair tech in these examples, hydralisk production is pretty important, and mutalisk harass can be quite potent. These things are based on gas economy. In general, however, gas expendatures on the zerg side in these examples are lower than that of the protoss in general. Unless the protoss goes heavy into zealots at a 4 gate production, in which case the zerg is spending gas on banelings or roaches.
Ultimately we'll consider a 7 warpgate scenario. 7 warpgates is about as much as I've seen fully operational on a 1-base play. 7 warpgates allows for .3234 units per second as a zealot/stalker mix. The zerg player can create 0.3333 units per second from 2 hatch, 2 queen. In order for the protoss player to maintain the 7 warpgates from one base, he's going to have a fully saturated main, so we can ignore probe production for now (although then we have to account for chronoboosts). Chronoboosting warpgates constantly would give him a 0.3503 unit per second production rate. This is edging out the zerg production, when you consider that the zerg must pay for overlords as well. A zerg player can produce about 0.2666 new units per second accounting for overlords, but when replacing units, can create 0.3333 units per second. In order to keep up with 7 warpgates, the zerg player would need 3 hatcheries, 2 queens, which cost 1350 minerals, while the protoss 7 gates cost 1050, not an unreasonable difference.
So what that is telling me is that for a Zerg, one basing is reasonable (production-wise) against an opponent who is one-basing. However, to maintain unit parity, you require more than one hatchery. By reasonable I don't mean it's a good strategy, I just mean that you can maintain a level of production consistent with your opponent, you can have a similar investment in your army, and an economy that is matched. This is interesting as I had always thought that as zerg, I was stuck being behind on one base. And while this is actually true to a limited extent (zerg spends more in buildings than other races to achieve similar tech levels) it's not by a large margin.
So, armed with that understanding I started to look at production rates. A hatchery produces 0.6667 larva per second and costs 300 minerals, and a drone, it takes 100 seconds to build. A queen produces 0.1 larva per second and costs 150 minerals, 2 supply, and provides defense, it takes 50 second to build. In the interest of production, the queen first is going to be optimal. Not only is it cheaper (cut fewer drones to build earlier), it's faster, and provides more production. After the queen is out, the hatchery is the next thing to go down.
The hatchery then can go down at the main or at the natural. Benefits of building it at the natural are that you get two more gas, you can drone longer before you're saturated, and you have twice the potential mineral production than if you stayed on one base, you also get creep at your natural sooner, which allows for spine crawler placement earlier. Benefits of building it at your hatchery means it's already on creep, it's protected by your ramp and defenses, and in general is significantly less vulnerable. My general strategy dictates that I use more minerals than gas, so the ability to secure two gas early is relatively unimportant. Thus until I'm saturating my minerals, I get little value out of having that hatchery at the natural.
So I try out initial build orders. 9 overlord allows me to avoid clipping larva while I wait for it to pop. 12 pool is a great timing for me, as it is a fast pool that also doesn't sit me at 3 larva, but lets me get my pool out fast. I will be at 17 drones by the time the queen finishes. After that it's about Hatchery timing. Initially I simply drone until I can build a hatchery, queen spewing larva every 40 seconds. As soon as larva spawn, I build zerglings with them, but every generated larva makes a drone. Doing this I know I am keeping up with a non-chronoboosting probe generation rate. This means that my larva from spawning will give me enough to deal with an early zealot attack. When the hatchery is complete, I build military from it as well, and start spawning larva on it instead. All larva from my main hatchery goes into drones.
This way, I know I'm keeping up in economy, I know I'm keeping up in military. My decisions now are tech, scouting, and deciding how to harass my opponent or deal with what he's sending me. Instead of expanding blind at 15 food like I was before, I begin expanding after I earn it. Defeat a push and retain a bunch of my army? Ok, I can afford to expand. Successfully harass with mutas, and kill some scvs and supply depots? Ok, I can afford to expand. Notice my opponent stick up 4 cannons at his base? Ok, I can afford to expand. Notice my opponent expand, ok, it's safe for me to expand too, unless I first want to take his expansion with lings or something.
This is allowing me to avoid putting up a risky expansion. With mass reaper play, early gateway pressure, sentries blocking my ramp, hellion harass etc. all a risk, that early expansion is a bit of a liability. I need to invest in spine crawlers, I need to underproduce drones, and I can't be aggressive myself. I'm immediately put on the defensive, because if I lose that expansion, which is really quite exposed, I will probably be unable to recover. That expansion is giving me the production I need just to keep up, it's not like it's giving me some bonus over my opponent, except as another site for gathering resources. Without it I lose more than the resource gathering, I lose the ability to spend my resources as well.
If I wait to expand until I'm ahead, or at least until I know I don't need to defend it immediately, I can drone more freely, I can make sure I've got creep all the way down to it, I don't have to invest as much in static defense, and I know I'm not going to get surprised.
I already know that if neither of us expands, I can stay within reach of his income, and his production so I'm safe, until I have an opportunity to cause him damage, he makes a mistake, or he expends a bunch of resources on things that do not threaten me like base defense, a new expansion, or low-threat units like corruptors in ZvZ when I have no air commitment and anti-air protecting my overlords.
So taking that strategy one step further, I start iterating on timings. When can I get my gas? How does that impact my hatchery time? What if I delay my zerglings further to drone faster? What if I get them earlier and delay my drones? When's the soonest I can push in and deal damage? What's the best time to get mutalisks? How many roaches do I need to be safe against a 4 warpgate push, and how soon do I need hydralisks?
And then making further iterations I start to consider other changes more integral to the strategy, what about that fast expand that I dismissed originally? How does a fast expand change what I'm doing? Can I safely still get a pool at 12, a queen reasonably early, and still get an expansion up soon enough that it won't be too threatened? Is that a reasonable thing to do on all maps, or should I avoid it on Lost Temple vs. Terran for threat of tank drops on highground? When can I get away with breaking the drone every 16.875 seconds rule? What about severe edge cases like 6 pool all-in with drones, or proxy 2-gate? Can this manage them? Do I have to make structural changes to do it, or simply scout them? When can I take a third? Will it ever be worthwhile to have a 3rd hatchery in base? Can I make due with the standard 2 base 2 hatch in some circumstances?
What I learn from this is, in a lot of cases, especially in situations where there's not a ton of 1-base all-in play, fast expanding is reasonably beneficial. If my opponent isn't committing his entire force and economy to the attack, it means I don't need to commit my entire larva economy to defending. If my opponent is planning to expand before the very late game, the fact that I've expanded before him gives me an edge.
However, the earlier the attacks are, and the more all-in they are, the more tenuous hold I have on my expansion. Losing or canceling that expansion sets me back immensely. Also, due to the larva economy, having the more exposed hatch at the natural means I have less opportunities to take advantage of the fact that I just defeated an all-in attack. I don't start gaining an economic edge until I have enough drones to saturate my main, unless I'm focusing on early gas.
The early expansion takes a gamble that either the opponent will not use early aggression, or will not do it well, or it expends resources and larva in base defense indicating that the opponent may have an opportunity to expand themselves. If the opponent takes the opportunity to expand, then the stakes are more or less equal, unless the zerg overinvests in base defense, in which case the opponent is ahead. If the opponent pushes early, then the expansion could fall, or be canceled, unless you have overspent on military at the expense of drones, and will have the opportunity to compensate for that later.
In general though, I've learned that a fast expand build works, it works because it allows for creep generation from your nat early, which means that you can put down spine crawlers which are quite efficient compared to other options. I've learned that you can compete with the production of any 1-base build with your fast expand, and in general you can keep up with an opponent producing off x bases by having x+1 bases. These are points I already knew.
But what I also learned is why all of that works. I've learned that you can do a lot with a 1 base build, and in fact, on certain maps, or against certain types of openers, a one base build may be much safer. You can one-base against an opponent who is one-basing, and maintain an even economy, and a reasonably even army, but to do so, you need a second hatchery. If you're in a situation where that second hatchery can not be placed at your nat safely, then placing it at your main is far better off than expanding right away. The production lost by being forced onto one hatch if you lose or cancel your nat is far more damaging than the resources and 350 mineral investment that it cost.
I've learned about gas economy, how if you consider 100 gas as 1 "unit" of gas, 4 roaches, 1 muta, two hydras, two stalkers, 2/3 of a void ray, 2 reapers or 4 marauders all take one unit. If you're on one base with your opponent, and you build 4 mutalisks and have spent all of your gas, and your opponent is one one base, and builds 6 stalkers and 3 void rays, and you've killed the 3 void rays with your queens, and the 6 stalkers with your zerglings, he's not going to be having much to defend against your mutas, but he's probably going to have a reasonable number of zealots available.
If you can take that expansion early, and start collecting that gas, you could have 10 mutalisks and be in the same position, but you'll have traded even more minerals, and thus maybe you wouldn't have been able to take out the stalkers with zerglings.
Thinking simply about this BO or that BO is interesting to follow, but it's not innovative. I like to know the "why". I wanted to know why nobody hatches in their main. The best answer I could come up with was because you can hatch at your natural. This makes sense, hatching at your natural is an advantage over hatching at your main.
The next question is, can you hatch at your main and have it be viable? This is a question not often asked. I think there's an amount of feeling that if you can't defend your nat, that with better play you could. I disagree with that. I think that while it's possible to defend your nat in almost every circumstance, it usually comes at a cost higher than defending your main. Thus, I believe expanding to your nat is only beneficial when the cost differential for expanding early is covered by the increased income from doing so. That means the fast expand puts you at a disadvantage, and you need to do something to even the scales. My issue there is that as Zerg it's difficult to secure an advantage from behind early, we have no reapers, or hellions, and our opponents have walls. That means that advantage doesn't get realized until Mutalisk time unless the opponent makes an opportunity for us. However, if the opponent doesn't abuse that disadvantageous timing through early aggression, the fast expand is quite a boon.
So, FE is popular because it works a lot. It has weaknesses, and those weaknesses are centered around early all-in maneuvers, or aggression to pay for expansion. It can counter all-in maneuvers just fine most of the time because of the potency of spine crawlers. An opponent who fails an all-in attack vs. a 2 base zerg is at a severe disadvantage. However, the aggression to pay for expansion gives it trouble. 5-rax reaper is popular now, you need to defend against it, and going lean is not going to cut it. You can not share defense between your main and your nat, which means you overinvest in spine crawlers, queens and roaches to make sure both areas are covered. The opponent then has the opportunity to expand, and now you're in a situation where you have a 2 base terran that has a huge production capacity vs. a 2 base zerg that has overspent in static defense, and taken some possible losses due to his defenses being spread thin.
The 1 base situation vs. the terran would be different. You could get away with fewer defenses and deal with the reapers, you would have a much easier time actually killing them, as they wouldn't just be able to run and harass something else. You could spread creep to your natural before you expand, and you could possibly delay any expansion attempts by your terran opponent with zerglings if they tried to expand before they were able to control it. Once you have defeated the reapers, or delayed the expansion, and have creep to your main, you could much more comfortably take your expo. While both builds survive, one is behind, and the other is on par, assuming equal play. It's a bit of a tricky situation because if the reaper player is not strong, the FE will come out on top anyways because you can afford to not overspend on defense, and end up with the economic advantage. But if you don't overspend on defense, and you FE, and the reaper player is strong, you could lose outright, or at least lose your expansion or something equally as devastating.
Anyways, just my thoughts. I have a ton of charts, spreadsheets, numbers and whatnot. And while those numbers mean very little in the heat of the game, I think they give me a bit of understanding as to why things work in such a way, why things don't work, and why certain perceptions are maintained. I think one of the perceptions that will soon change is that hatcheries should exist one per mineral field. Despite the fact that, in general, it's hard to spend all the larva from 2 properly maintained hatcheries with resources coming from 2 bases, in reality, you can, and likewise, there are points in the game where even the best players let their macro slip, and being able to have twice the larva spawning after that slipup could mean the difference between a win and a loss.
I don't know how popular the blog section is on here, so if anyone actually reads this whole thing (I'd be surprised) and has arguments against my assumptions, I'd be happy to hear and consider them. I'm not trying to tell people how to play at all, I'm simply trying to understand for my own benefit, how these cogs fit together. So telling me something like "You're stupid, just 14 gas 13 pool, that's the way zerg plays" doesn't help. I know how 14 gas 13 pool works, I know how x gas, y pool effects a bunch of things for tons of variations on x and y. I'm interested in why it works. What timings does it enable? What holes does it open? Is it something that nobody's considered? Is it something that someone's considered and thrown away because it has a big weakness?
I'm done now
