Macro: when to expand, and what it pays back

Lesson

5 of 14

Watch

A written lesson, no video

Practise

A checkpoint en a plan for your next game

By the end of this lesson you will know how many workers a base wants, how many production buildings your income can feed, and when an expansion pays for itself. You will also know which macro mistakes cost you a little and which ones cost you the whole game.

What you will learn

• Why a worker is the best thing you can buy, and what one missed worker really costs.
• How many workers fill a base, and where extra workers stop paying.
• How many production buildings one base of income can keep busy.
• How to work out, in your head, when an expansion pays for itself.
• Which mistakes to panic about and which to ignore.

An economy compounds

Your income buys workers, and workers raise your income. That loop is the whole economy, and it means your mineral curve bends upward for as long as you keep feeding it.

The numbers to hold on to, using the working figures from the Economy Lab:

• A worker on an uncrowded patch mines about 42 minerals a minute.
• A worker costs 50 minerals. It pays for itself in about seventy seconds of mining, and then earns for the rest of the game.
• Nothing else you can buy pays back that fast. A zealot never pays back at all unless it kills something.

So the first rule of macro is the boring one. Your halls, nexuses and hatcheries should never stop making workers until your bases are full.

How many workers fill a base

A mineral patch takes two workers at full rate. A third worker on the same patch mostly waits its turn, so it adds about a quarter of a worker's income. A fourth adds nothing.

On a base with eight mineral patches, that works out like this:

Add three workers on each gas geyser, at about 38 gas a minute each. That makes a finished base around 27 workers: 24 on minerals and 3 on gas.

Two numbers to take into your games. At 16 on minerals your base is doing most of what it can. Past 24, every new worker on that base is 50 minerals for nothing. That is the moment you need another base, not another worker.

How much production one base feeds

Income is only useful if you can spend it. So match your production to your income.

A production building that never stops spends a fixed amount a minute:

A saturated base earns about 760 minerals a minute. Take off 150 for the hall still making workers and you have about 600 a minute for army. That is about five barracks on marines, or four gateways on zealots, all running without a break.

Use it as a check in your next game. If you have one base and two gateways, your money will pile up no matter how fast you click. If you have one base and eight gateways, half of them will sit idle and you paid 150 for each one that does.

Supply comes first

Supply is the jam that costs the most, because it stops a fully funded line cold. Every production building you have goes idle the moment you are blocked.

• A pylon, a supply depot and an overlord each give 8 supply for 100 minerals.
• One barracks making marines uses 1 supply every 24 seconds. Five of them use 8 supply in under 40 seconds.

So with five barracks running, you need a new depot about every 40 seconds, and it has to be started before you need it, not when the warning appears. Build supply on a rhythm. When your production doubles, your supply rhythm doubles with it.

The mistake that never stops costing

Here is the most important result in this lesson.

Say you forget to build a worker for 30 seconds, once, four minutes in. Right away you are down one worker for 30 seconds, which is about 20 minerals. The real damage is that you are down one worker for the rest of the game, and that worker would have mined about 42 a minute for the next fifteen minutes.

Over six hundred minerals for half a minute of not paying attention, because the loss compounds forward instead of sitting still.

Now compare pulling your workers off the minerals to dodge an attack. If you keep building workers while they hide, you lose only what they would have mined while they hid. That never grows.

The ranking that falls out of it: never stop making workers, never lose workers, and stop worrying about the minerals you miss while you dodge an attack. Most players have it backwards. They let workers die to save a few seconds of mining.

The Zerg exception

Zerg gets some forgiveness. A hatchery stores up to three larvae, so if you are late on a drone, you can make two at once and catch up. The limit is that third larva: once a hatchery is full, it stops making more, and every second after that is lost for good. A short lapse is free for Zerg. A long one costs as much as it does for anyone else.

When an expansion pays for itself

An expansion is not a mood. It is a purchase: 400 minerals for a command center or a nexus, 300 for a hatchery, spent on something that shoots at nothing. The question is how long until you get it back, and what you would have bought instead.

Worked example 1: you move workers you already have

Your main is full. Eight of your workers are the third worker on their patch, each adding about a quarter of a worker, around 10 minerals a minute. Move them to a fresh base and each earns about 42 again.

400 divided by 256 is about a minute and a half. Start the clock when the base finishes, not when you place it.

Worked example 2: you build new workers for it

Now the expansion costs 400 for the base plus 400 for eight new workers, and all eight earn a full 42.

About two and a half minutes. After that, the new base is pure profit, and the profit grows, because the extra income buys more workers.

What you did not buy

Payback time is half the decision. The other half is what you gave up, and it is always specific.

So the real rule is not "expand early". It is this: you can afford the payback time if nothing arrives during it. Two and a half minutes of being 400 minerals of army down costs nothing against a player who is teching. It costs the game against a player whose units are already walking at you. That is why expanding and scouting are one skill. Your scout tells you whether the next two and a half minutes are safe. Lesson 7 teaches you how to read it.

Common mistakes

• Stopping workers at 16 on one base and never expanding. Your income stalls at about 670 a minute while your opponent's keeps climbing.
• Building workers past 24 on one base. Each one costs 50 and adds nothing. Expand, or move them.
• Expanding with no army at all. The payback is real, but only if you are alive to collect it.
• Letting the bank climb while buildings are busy. You need more production, not more clicks. See the table above.
• Losing workers to save mining time. A dead worker costs you every minute for the rest of the game. A hidden one costs you seconds.
• Waiting for the supply warning. By then your production has already stopped.

Practise

Then the replay drill. Open one of your own recent replays and do three things:

1. Pause every minute for the first eight minutes and write down your worker count. Mark any minute where it did not go up by at least three per hall.
2. Write down the first time your bank passes 500 minerals. Count your production buildings at that moment and check them against the table above.
3. Find every supply block. For each, write how long it lasted and what was idle during it.

That list is your macro practice for the week. Fix the first item on it, nothing else, and play five games.

Checkpoint