SC2 skill "action latency" & "actions per minute"

(Credit to for the charts.)

It’s clear as day that 99% of the skill in SC2 is A) how fast you react and B) how many actions you can spam per minute (which is just another way of specifying how fast you react).

Is it time to reclassify SC2 as an “RE” game, aka Reaction & Endurance game? Strategy has little to no impact on the outcome compared to these two factors, so it’s hard to call it a “strategy” game anymore. Byun for example can go 1 factory mass marines in TvT, and it’s a horrible strategy, but it works because he radically outclasses his opponents in APM. The game is almost entirely your ability to spam APM and to do this for long periods of time.


Thanks for the share, this quite interesting data that you’re providing there. However, there are two notions that you might be neglecting :

  • First, and above all, it’s not because a graph shows some likely correlated variables that there aren’t others, nor that those are independent from each other. The mYGBLz graphs shows APM and reaction time as progressing with league, but that doesn’t means those are the only parameters that do evolve with those. The scouting, the adaptation, the game sense, the composition choice, transitions knowledge and experience in taking fights would logically also evolve from Bronze to GM.
    To prove it to you, take your graphs and just remove the APM from all of them : only reaction time is now shown. It seems correlated with league, as it lowers the better the players are. But is that correlation and the absence of graphical representation of other factors a proof there aren’t any other ? Nope.


  • Second, you’re assimilating APM and APM spam. And, it is likely that one of the biggest differences between high level players and beginners isn’t the maximum APM reached, but the amount of those APM that are indeed translated into useful actions. In other words, not only EPM, but also the part of those EPM that will concur to achieve what a good decision making produced. It is likely those good EPM are still correlated to your APM capacity, but the reverse isn’t necessarily true.
    There’s a gold player peaking at 300 APM in my smurfs thread, despite not having the wit to produce another thing than slowlots without wall to stop speedlings runby. Needless to say, EPM and good decision making weren’t proportional to APM in his case.



So we hold down 2 keys to generate massive amounts of apm and get gsl trophy? Sounds dope im in!


Strategy is what you plan before a battle. SC2 is a Real Time Strategy game. You have a strategy, and you execute it in real time; so yes mechanics are critical, that doesn’t mean there isn’t strategy. If what you want is 100% strategy and tactics with no mechanical requirement, go play Chess or Go or any other Turn Based Strategy game. That’s what they’re there for.

1 Like

Absolutely none of these are relevant unless you can play the game fast enough to actually do them. You can’t pick the correct fights if you can’t keep up with how fast fights happen. Reaction latency is 99% of the skill in the game.

The supposed variables affecting performance are a multiple input single output system. If one variable controls performance, it reduces the control of another variable by definition. The total control must equal 1 in other words. So if you have a variable that correlates highly with performance (reaction latency), any other variable by definition is either not distinguishable from reaction latency or doesn’t impact performance.

By definition any variable impacting performance must be both distinguishable (no / little correlation with reaction latency) and have a positive correlation with performance. If such a variable has a positive correlation (with performance) then reaction latency can’t have a correlation near 1 (given the requirement that it must be distinguishable), BUT it does on a per league basis. So no other variable exists that contributes to performance on a league-scale.

In other words on a per individual basis there is scattering between the reaction latency & and league correlation, indicating that other factors exist that impact performance on a “localized scale”, but on a league scale reaction latency is 99% of what makes up skill. In other words, if you are facing against a person of a similar reaction latency, other factors become relevant otherwise reaction latency is the driving force behind your performance. Reaction latency is 99%+ of the skill in SC2.

Nope. Reaction latency isn’t biased by spam. It’s clear as day in those charts. Many master level players have higher APM than GMs, but GMs have lower reaction latency. So it measures the same thing but more accurately. Reaction latency is 99% of the skill in SC2.

1 Like

You can have Serral-tier game understanding but if you have bronze reaction latency you will be bronze. Reaction latency decides 99% of your performance because SC2 has so much emphasis on the “real time” aspect and so little emphasis on the “strategy” aspect. That’s why Byun can spam completely nonsensical strategies and make them work (ex: 1 factory mass marine in TvT) yet Bronze level players TRYING to use the best strategies are still stuck in bronze. SC2 could be called a strategy game if strategy had an impact on the outcome. Unfortunately that just isn’t the case.

By the way, it’s very easy to prove that you are wrong. Take any equation with any number of variables and set it equal to 1. Then sample a random range of those variables and calculate a correlation table between them and the dependent variable. It is impossible for multiple independent variables to exist which correlate with the dependent variable near 1. If a variable exists with that high of a correlation, then it’s the only variable affecting the dependent variable.

1 Like

Even better, swap your mouse and keyboard hands and watch the mmr TANK…


Yep, it’s blatantly obvious that game knowledge, scouting, and literally every other aspect of the game, is dependent on your ability to keep up with the pace of the game. If you can’t play the game as fast as your opponent, no other factor is relevant. This is measured in reaction latency and it is definitely the driving force behind the skill distribution of the SC2 ladder system. It is what separates a GM from a Bronze.

When you have a GM vs GM then other factors become relevant but what makes a GM a GM is reaction latency, period. If you took a GM and made him play with his non dominant hand, for example, his reaction latency would skyrocket and he’d be Bronze league even though he is GM in all other aspects of skill.

1 Like

No one denies that quick reaction is important (it might not even be skill but animal reflexes that degrade with age).
Blessed be the simple-minded because they will inherit the Forum-Coup.
This clown is always unbalanced, like an adolescent his attention is caught after the impulse of the day and this juvenile moron balloons it out of proportions…

False. The other player has Bronze “reaction-latency” AND Bronze game sense and knowledge.
It need too much brain to understand who will win?
Having Serral game-understanding and Bronze reflexes puts somebody well at Dia1 level.

Alright. Switch your keyboard and your mouse to the opposite hands. Your rank should stay the same, right? Nothing changes except the speed that you can play the game at. Go ahead. Prove it.

You will soon realize that your ability to do anything in the game is dependent on how fast you can do it. To have Serral level scouting you have to have Serral level reaction speeds. If you want Serral level strategies, you need Serral level reaction speeds. If you want Serral level micro you need, YOU GUESSED IT, Serral level reaction speeds.

Winter low APM challenge disagree with you. O and also anger coaching jun around 1:38:00(slimy) who have around 100 apm(middle of replay)/200apm(end of replay) in masters.


Probably, and that would be a clever way to proceed. But you have yet to prove that there are only two variables involved in the current equation (which is ingame performance).

Let’s take an extreme example, of an entity with nearly instant reaction time, and about 500-600 APM average. According to your hypothesis, that entity should deliver ESL caliber level of play, and we ordinary players shouldn’t even get close to it. The only problem, is that this entity is stupid, doesn’t learn, and therefore has a poor decision making. I have named the game’s current AI insane bots, which I do managed to beat regularly while I’ve got >5x less APM and >5x more reaction time, and while the bot has been given vision and cheated resources.

Now take that entity, and make it learn from its mistakes. With Protoss, that entity has reach a level where even progamers do have a hard tome beating it. You’ll have understood I refer to Alphastar.

Alphastar’s performance (which was capped to ≈300 APM average) is incomparably superior to the regular AI bots, and that despite the AI bots having twice its APM and near instant reaction time. Learning capabilities, and this improved decision making is the difference between those two. Thus proving that APM and reaction time aren’t the only factors involved in RTS. :mag_right:

1 Like

What you are saying is, if you fly an airplane into the ground then airplanes aren’t safe. That’s absurd. Just because you can spam APM and accomplish less with more doesn’t mean that’s typically what happens. There will be a certain “efficiency” to how the APM is spent and some players will be more efficient and some less - that’s clear as day in the charts I listed above. However, that variability averages out on a per league basis but reaction latency & APM do not. Reaction speed is the driving force behind the MMR distribution of the SC2 ladder.

Uh, that’s exactly what I did. APM’s correlation goes to near 1 on the average per league. As I explained, it’s not possible for another variable to exist that both has an impact on performance and that is distinguishable from APM since any casual system of effects (whatever they may be) must add up to 1, meaning all the combined effects contribute to the entirety of all skill. If a factor impacts skill, it most both be separate from APM (no correlation with APM) AND a correlation with performance. Since APM correlates near 1 with skill, no other variables exist that impact performance.

And those bots can win with strategies that normally are never used because they are such bad strategies. For example, Alphastar beat Mana’s immortal push with EIGHT immortals using nothing but blinkstalkers. Strategy is not relevant unless both players have the same reaction latency, which is exactly why AlphaStar’s team had to impose human-like constraints on the AI (bringing it down to humanity’s level in terms of APM/speed forces it to win with other factors like strategy).

1 Like

“This army from Mana is very powerful. The immortals will ALWAYS take care of the stalkers.” - Rotti –> fast forward a few minutes –> AlphaStar micros dozens of blink stalkers with a 360 degree surround and wins –> “Wow. Wow. That is just unbelievable! Mana has made so many Immortals in this game that I can’t even count that high, and the fact that he has been defeated by literally pure stalker - the unit that it most counters - is so surprising.” - Artosis.

Are you going to claim that Stalkers are a counter to Immortals, and that AlphaStar used the proper strategy, or are you going to admit that strategy has no impact on the game when you have faster reactions than your opponent?

1 Like

It’s absurd to say that strategy is only 1% of the game while maintaining that playing Protoss affects performance. You’ve clearly underestimated the “select Protoss” strategy.

How can you fly an airplane into the ground if it does not fly?. :rofl:


That’s indeed absurd. But you’re the only one behind those words.

You said APM and reaction time were the only factors affecting performance, if this was true Alphastar and most players would be straight outperformed by the insane AI bots. Yet, it’s the contrary. Meaning that for your plane to reach as high as it’s meant to, the experience of the pilot matters as much as the plane’s speed.

You are contradicting yourself. If there is an APM efficiency variable, then reaction speed and pure APM aren’t the only variables involved in RTS.

Of course it is. APM per league and experience per league aren’t independent from each other. So APM per league strong correlation with performance could be in fact linked to the experience in strategical factors. And so APM per league and APM in general aren’t the same thing, as proven by the insane AI bot.

If those strategies lead you to victory, they aren’t bad. They could be bad to others, but not to people able to implement them. Casters at intermediate and low levels are a good example of that.

Anyway, experimenting and learning is what will tell you how efficient a strategy can get in your case. That’s why 280 APM Alphastar >> 600 APM insane AI.


1 Like

Ok. If serral gets a 500ms delay he will be bronze? Yeah sure.

Cap serrals apm to 50. He wont be Bronze.

1 Like

Why is batzy so eager to use strawman Arguments? Even worse. He compares it to something an AI did with INHUMAN micro Skills. Batzy batzy batzy.

At the same time he delivers the Argument that apm is not everything because mana has higher apm

Seems that most if not all people in this topic arguing with batz data don’t know what reaction latency is.

To put it simply, it is the time in seconds that elapses between the delivery of a stimulus and the person’s response.

To clarify, he is talking about reactions, which are distinct from reflexes, which are involuntary. What this means is that, in reaction latency, there IS a prefrontal cortex component to the circuit (among many other things). prefrontal cortex is responsible for advanced cognition, among other things.

In starcraft, reaction latency, would translate to the time it takes from a player receiving the stimulus, processing the inputs, deciding what to do, and then initiating a (mechanical) response as accurate and precise as possible.

So yea, if you give serral bronze level reaction latency, he’d be bottom of the barrel.

1 Like