How to handle losing at StarCraft

Newcomer Learner Competitor Advanced

The short version

After every game, name the one thing that decided it. One sentence. Then stop thinking about it.
Judge the decision, not the result. A good decision can lose and a bad one can win, and you get better by noticing the difference.
Morale and stamina are in the resource taxonomy for a reason. They are spent, they run out, and they do not refill during a session.
When your play gets worse after a loss, playing more is not practice. It is practising being worse.
A series is not three matches. The better player wins a best of three more often than a single game, and the underdog should want fewer games and more variance.

The situation

You lose. You queue again in eleven seconds. You lose that one worse, and the third one you lose in a way that would have been unthinkable an hour earlier.

Nobody thinks they tilt. What everyone does is play one more, and the honest description of an evening that ends four games below where it started is not bad luck. It is a resource that ran out and a player who kept spending it.

Name one thing

The habit the class opened with, in its very first session, before any mathematics: spend a moment after every game naming the one thing that decided it.

One thing. Not a list. The value is in the compression, because a sentence you can say is a sentence you can carry into the next game, and eleven observations are not.

It also forces a real judgement. "I lost the fight at twelve minutes" is not the one thing; it is a description. "I took a fight in a choke while ahead" is the one thing, and it names something you can decide differently next time.

And then you stop. The thinking happens once, it produces a sentence, and the sentence goes in the review pile for later (T13 is the method). Continuing to think about a lost game past the point where it has produced a sentence is not diligence, it is a way of spending a resource you need for the next game.

The decision is not the result

The hardest and most valuable separation in any game with hidden information.

You scouted, you saw one base, you took the expansion, and their all in arrived thirty seconds later and you lost. Was expanding wrong?

Usually not. Given what you knew, at the moment you knew it, the expansion may have been the correct play, and it lost. T12 has the arithmetic: against a genuinely unknown choice, the unexploitable allocation loses some fraction of the time by construction, and the fraction is not a mistake. It is the price of the game being fair.

So ask the question in the right order.

What did I know at that moment?
Given that, was there a better play?
Only if the answer is yes: what would have told me?

Players who skip to question three invent scouting habits that solve last game. Players who stop at question one get comfortable and stop improving. The discipline is doing all three, in order, once.

Morale and stamina are real resources

Lesson 1's taxonomy put reaction speed, practice time, morale and stamina in the same box, and called them physical resources: things that exist in your room rather than in the game.

That classification is not a metaphor. It has the two properties that make something a resource. It is limited: your fortieth game of a session is not your fourth. And it is contested: an opponent who makes you play a long, uncomfortable, defensive game is spending it on purpose.

Which gives you a management problem rather than a character problem. You do not need to be tougher. You need to notice when the tank is empty and stop, in the same way you would notice a float and spend it.

The working

Show the maths

When to stop for the night. Suppose your win probability starts at $p_0$ and drops by $\delta$ for each consecutive loss, recovering on a win. After $k$ losses in a row you are playing at

p_k = p_0 - k\delta

(1)

The number is invented. The shape is not: almost everyone's play degrades with consecutive losses, and almost nobody stops.

Two things follow, and neither depends on the exact value of $\delta$ . The first is that games played below $p_0$ teach you the wrong thing: you are practising your degraded play, and practice is how habits form. The second is that if your rating moves with results, the expected rating change of continuing goes negative as soon as $p_k$ drops below the break even point, which for most rating systems is around a half.

Set $p_k = 0.5$ and you get a stopping rule you can actually use:

k^{*} = \frac{p_0 - 0.5}{\delta}

(2)

For most players, with honest numbers, this comes out between two and four consecutive losses.

Which is why the folk rule, stop after two or three in a row, is not superstition. It is roughly the right answer for plausible values of the two numbers, and you can find your own by checking your own history: your win rate in games played immediately after two losses, against your overall win rate.

A series is not three games. If you win a single game with probability $p$ , you win a best of three with probability

P_3 = p^2 + 2p^2(1-p) = p^2\,(3 - 2p)

(3)

Check it at a half: a quarter times two is a half. A fair player is still fair.

Put numbers on it. A fifty five percent player wins a best of three about fifty seven and a half percent of the time. A sixty percent player wins it about sixty five percent of the time. The longer the series, the more reliably the better player wins, which is the whole reason serious competition is played in series.

And the corollary, which is the useful half: if you are the worse player, you want variance. Fewer games, riskier builds, all in timings, unusual maps. If you are the better player you want the opposite: reduce variance, play the standard game, and let the arithmetic collect. Most players have this backwards and play their safest game when they are behind.

What a series changes

The second half of this is not about feelings at all. It is that a series makes each game worth more than one game.

Lesson 11's pre-game analysis treated this explicitly. In a best of three, beating an opponent's signature build in game one is worth more than the game, because it takes something away from them for game two. Losing to it gives them something. So the value of a game includes its effect on the next one, and that effect is sometimes larger than the win itself.

Three practical consequences.

Spend your surprise where it is worth most. A one off trick is worth more in the game that decides the series than in game one, unless the psychological effect is the point.

Expect the reputation. A player who is known for a build is more likely to bring it, not less, because they have something to defend. T13 has this as precedent.

Your own tilt is information they have. If you visibly fall apart after losing to the same thing twice, you have told them what to do in game three.

What the lecture said

This is the first thing the course teaches. In the lesson 1 lecture, from about 8:00, one of the teachers gives his theme before anything technical happens: at some point a player becomes a coach, take losing in stride, and spend a moment after every game naming the one thing that decided it.

The resource taxonomy that puts morale and stamina on the list arrives later in the same lecture, at 1:06:00.

And one more line from the same lecture, at 14:00, which belongs here more than anywhere else. The lecturer told the room there are dumb questions, and gave an example: asking whether the pylon goes at seven and a half or eight and a half without being able to say why you are building the pylon at all. A loss you cannot explain is a loss you will have again, and the point of naming the one thing is to make the explanation exist.

Still true today

Checkpoint

You expanded on a scout that showed one base, and an all in killed you. Was expanding wrong?
Why does the better player win a best of three more often than a single game?
Three losses in a row and it is late. What is the best move?

Sources for this page

Lesson 1 lecture, on YouTube, channel VioleTAK The facilitator who gave the progaming history opens with this, from about 00:08
Lesson 11 lecture, on YouTube, channel VioleTAK The pre-game analysis, where a series is treated as its own object