Using OODA Loops to talk through playing fighting games
This essay is a little different from the usual advice stuff! I’m taking a concept from military organizational strategy (and jet fighter piloting) called “the OODA Loop” and applying it to fighting games. It’s not the first time these have come up in fighting games (or even the first time that I’ve written about it), but in this essay I’m adapting it into a framework that I can hope we can use to be more specific about the kinds of work our brains and bodies do to actually play these games.
In fighting games, we normally just learn by doing, because we’re not very good at talking about how we actually play fighting games. When someone asks us how to play these games better, we’ll give them a list of things they should do to get better, but we’re not actually telling them what those things help us do better. Which is fine, but it is often unsatisfying to people who don’t want to feel like they’re dealing with mystical Mr. Miyagi Zen stuff. (I love the Zen stuff myself, but I am sympathetic to those who are just here to play the video game.)
In particular, the feeling like you aren’t sure if you should be doing X when you could be doing Y can create some real anxiety in people, especially if they’re used to having teachers, coaches, bosses, or UI elements tell them what to do next. And so I see plenty of new fighting game players with more tools and aids to learn fighting games than we’ve ever had before, but they don’t have any idea how to use them to learn, no one to teach them, and no idea how to mark their progress besides what the game itself gives them. Most of these people will burn out and churn out, and the ones that don’t may end up sticking around to learn something, but won’t necessarily have a great idea of how to help other people get to where they are.
Also, people who are new to fighting games typically don’t have very good words to describe why someone is better than them, leaving many to conclude that they just don’t have the right reaction speed/manual dexterity/whatever to play fighting games. This is incorrect! As you’ll see in this essay, getting better at fighting games often involves streamlining your thought process to speed up your reactions. However, new players often have a hard time believing that they’ll actually get faster if they play more. My hope is that in getting better tools for describing fighting game play, we can demystify the process of skill acquisition in fighting games and help people shift their practice to more intentionally focus their skills toward the kind of results they’d like to achieve.
Before we get to the framework, though, we’re going to get a bit more specific about what ‘getting good’ means in fighting games, with a focus on tournament play.
Defining the “tournament player style”
One of the things that makes learning fighting games hard as a new player is that the games give you so much stuff that you could learn that it’s not easy to tell what you should be learning first or why you should learn it. So I think it’s important to start from what I would generally describe as the generic “tournament player style”, which is largely about creating opportunities to consistently outplay your opponent with as little variance as possible. It is not the only way to get good at a fighting game, but it’s a good place to start.
Fighting games are an environment where two players subject each other to a series of skill tests. Any given fighting game can contain many different tests, and some games tend to emphasize a narrow range while others can get pretty out there, but they generally all tend to test our ability to know the right thing, feel our opponent’s intention, and execute the appropriate response in time; you may have heard of these called “Head, Heart, and Hands” or something similar. Head tests are about understanding the stuff inside the game (matchup details, move interactions, risk/reward calculus, etc.), Heart tests are about playing the emotional battle against your opponent (reads, misdirection, downloading), and Hands tests are about your ability to actually do the things that your head and heart are telling you to do (combos, reactions, totally sweet option selects).
Improving at a fighting game in the abstract could involve doing all kinds of different things that let you subject your opponent to nastier tests while passing your own. However, the fighting game community has largely agreed upon double-elim, best-of-three (sometimes five) tournament bracket performance as the primary format for meaningful competition, and that format has a few specific implications for how we think of getting better at fighting games. Perhaps the most significant factor here is that tournament sets are pretty short, and if you lose two of them you’re out of the tournament, so you need to aim for a high degree of consistent success in order to do well in tournaments.
For example, the tournament bracket format rewards players who invest deeply in their game knowledge, particularly matchup study, because you do not know who your opponents are ahead of time, and best-of-three sets don’t give you much time to figure the matchup out. However, at a high level, you generally have to assume that players ‘know everything’ that the other players know, so you generally cannot rely on knowledge checking alone to win. (There is some room for exception here with games at the very very deep end, like Marvel vs. Capcom and classic Guilty Gear games, where it’s entirely possible that no one in the world plays team XYZ or character A quite like you do, and so the only way your opponent could have that matchup practice is to play you.)
Once you’ve gotten to the point where game knowledge is not a notable differentiator between players, the successful tournament player typically shows a strong preference towards steering the game towards hands checks (high combo execution, strong reactionary play) over heart checks because it has lower risk of unwelcome variance than gambling/RPS interactions do. Again, these are short sets, and you have to play a lot of them to win a tournament bracket, so even if your tournament game is based around running a bunch of nasty gambling games, odds are that your luck will run out before the bracket ends. In general, the more you can make a fighting game feel like shooting free throws (a single-player test of execution) instead of rock-paper-scissors (a two-player test of reads), the easier it is to consistently win; you’d only want to play RPS against players you don’t think you could beat with free throws.
So, broadly speaking, the successful tournament player does as much as possible to keep the game within parameters that they can control. It’s hard to raw guess your way through a bracket, so you want to minimize the number of situations where you’re raw guessing. (This may sound obvious to you, but consider that if you’re playing in a ranked queue, or with a group of friends, or in a first-to-ten against Daigo Umehara, you would do different things to optimally prepare for those sessions than you would for Combo Breaker.)
As we start talking about structuring our gameplay in OODA Loops, keep in mind that the end goal of your fighting game play isn’t just “win the game”, it’s in finding ways to distill the chaos of two people simultaneously fighting and turn it into a neatly ordered series of skill tests that you can consistently pass more and better than your opponent can. Eventually you may decide that your own path to improvement will branch from the classic tournament player meta, and that’s awesome, but it helps to understand where the format of competitive play encourages you to build first.
Using the OODA Loop framework to describe fighting game play
Fighting games are kind of like turn-based card games in that each player is taking turns, with different phases to each turn. When you’re playing a fighting game, you go through the phases of your turn like this:
- Look at the screen
- Process the information on the screen into a situation
- Use your understanding of the game and the opponent to determine an appropriate action
- Execute the action
- Once the action is executed, you once again look at the screen to see how your action created a new situation, and the next loop starts.
This structure is called an OODA Loop, which stands for Observe, Orient, Decide, Act. The main difference from card games, of course, is that both players are going through their turns at the same time, and they’re often not in sync with each other since it’s all happening in real time. Many kinds of feints and traps are designed to manipulate the opponent’s OODA loop cadence to bait them into committing to something that you can beat on reaction instead of forcing a contest of reads.
Getting better at fighting games is, at a very low level, about cultivating a set of tools to plug into your OODA loops, and picking the right ones to use in a given moment to create a situation where you can consistently perform the winning action before your opponent can use theirs. Let’s walk through what is probably a familiar situation with this framework in mind.
A common starting point for new Street Fighter players is “Don’t jump, work on hitting your anti-airs.This is a pretty handy place for new SF players to start for two reasons: First, it teaches them to look for situations where there are clearly correct, reactable answers (Dragon Punch beats jump-in), and second, telling them not to jump forces them to rely on their grounded movement and attacks to deal with their opponent’s, and that’s where they start learning how to play SF-style neutral.
If we have New Player 1 jumping in at New Player 2, P1’s OODA loop probably looks something like this:
Observe: I am not within range to hit my opponent with any of my grounded attacks
Orient: I need to hit my opponent to win
Decide: I am going to jump at my opponent with a big attack
Act: Jump forward heavy kick
P1 doesn’t know much about the game, so their OODA loop doesn’t take much time to execute. This is one of the reasons why a new player trying to learn to play ‘properly’ often gets overwhelmed by the player who isn’t — if you’re trying to use new, unfamiliar tools, you’ll act more slowly than someone who is unencumbered by the weight of trying to do the right thing.
When we tell P2 “Don’t jump, work on hitting your anti-airs”, their OODA loop looks something like this:
Observe: Is there an opponent jumping at me y/n
Decide: I should Dragon Punch
If we compare this OODA loop to the previous one, we’ll see a couple notable differences. The first one is that the “Observe” phase is indeterminately long; P2 is just standing there checking to see if there is an opponent jumping at them. The second is that P2 is leveraging some knowledge of the game (DPs beating jump-ins) to look for a situation where they can get a consistent win. The third is that the action of inputting a dragon punch takes more time and is harder to perform than the jumping forward heavy kick.
Now, let’s give Player 1 some credit. After getting DPed a couple times, they figure out that DPs beat jump-ins, and so they figure out a new tactic: Walk up sweep. Now they’re just sweeping poor Player 2, who is diligently looking for jump-ins that aren’t coming in. So we’re going to tell P2 to throw a fireball if they see P1 approaching on the ground, because that fireball will beat P1’s sweep.
At this point, Player 2 is performing a basic shoto gameplan: If the opponent is jumping at you, DP; if not, throw a fireball. Player 1’s sweeps are getting countered by fireballs, and their jump-ins are getting countered by DPs. P2 is playing a reactive game, using their knowledge of two basic interactions to decide upon the correct course of action, and their proficiency in executing fireballs and DPs to make sure they don’t fail the reaction test.
So far, we’ve got the head check (knowledge of interactions) and the hands check (executing special moves). But we’re missing the predictive element of fighting games, so let’s give P1 a tweak to their Orient phase: Only jump in if you think P2 is going to throw a fireball.
We’re essentially asking P1 to make an educated guess here; by the time they visually confirm a fireball, they’re likely too late to jump. But if P1 can pick up on the fact that P2 is throwing a fireball whenever they see P1 on the ground, they’ll be right about every jump-in. Download complete.
This is where people start playing “the real game”; the neutral is framed by a couple hard interaction rules like “DPs beat jump-ins” and “fireballs beat pokes” and “jump-ins beat fireballs”, and then both players use the rest of their tools to nudge each other into those clear interactions where they can use successful predictions or reactions to win.
When we tell a new SF player to avoid jumping and work on their anti-airs, we’re introducing them to an important part of learning to play in competition. It’s not just about ‘not jumping’, it’s about learning to recognize and play for situations where you can cleanly win an interaction due to knowledge (DPs beat jump-ins) and execution (reacting to the jump-in with a DP input) instead of winning off of guesses, because that’s a more reliable path to victory overall.
However, reacting to something isn’t easy; it requires that you have the right option queued up in your mind to react to, enough time to actually react to it, the knowledge to know what you should react with, and the dexterity to execute the reaction. And if you’re caught fishing for reactable situations too often, you’ll become predictable. So players then have to learn how to make reads on the opponent’s behavior, which is handy because a prediction costs you zero time in your OODA loop (since you’re not waiting for anything like you would be in a reactive situation), but it’s risky because if you commit to a guess and you’re wrong, your opponent might be able to react to your guess with a punish.
As an aside: If you’ve ever wondered why people often play differently in ranked netplay than they do in an IRL tournament, you can think of it in terms of how the different play contexts change your OODA loops. Even though the game ‘rules’ are the same either way, there are a couple factors that change things up, including:
- Netplay delay means that situations are going to be less reactable than they are offline, meaning players have less time to execute reactive decisions
- Offline play gives you more potential information to observe; besides the game itself, you also can usually see and hear your opponent themselves, which can give you early cues to react to (like reacting to button presses) or more information on their emotional state via body language
- The cost of losing in ranked is some LP, then you mash to requeue and get into another game, while the cost of losing in a tournament might mean that you’re out of the bracket, so most players are going to be a little more risk-averse offline (which usually means they’ll be less predictive, more reactive)
So, hey, that worked! We took a pretty familiar situation and used OODA Loops to describe how players get better at playing out that situation by adding nuance and decision-making flows to their game. Now let’s try taking it for a spin on something a little bit more involved.
Using OODA loops to explain why dash FD brake is useful in GG neutral
It’s very common to see new players in classic GG games like XX and Xrd hard commit to a single course of action at full screen, especially if they’re playing characters like Sol, who can easily run to the middle of the screen and then swing with a big button, put out a projectile to set up a strike/throw mixup, or instant air dash over the opponent’s anticipated normal move or projectile and counter hit them for big damage.
Framed as an OODA loop, this action would look something like this:
Observe: My opponent and I are very far from each other
Orient: I bet they’re gonna run at me and press a button
Decide: I’m gonna run up and gunflame their button
Act: Run to about half screen, then gunflame
If you put two new players together, their neutral will often look like a series of RPS interactions, where they run at each other from full screen distance and collide in the middle with buttons/fireballs/IAD attacks that resolve in a chaotic scramble until someone bursts or air techs out back to full screen distance, where the cycle repeats.
Against a more experienced player, however, they’ll often find that their buttons are whiff punished, their IADs are cleanly 6Ped, and their projectiles are easily countered. If you don’t know what’s going on, it might just feel like your opponent is reading your mind. That’s because they are — partially, anyway. Once you’ve seen enough people commit to their decisions at full screen, you’ll be able to recognize that the speed with which they’re deciding and acting leaves very little room for observation and orientation; they look at the screen precisely once, decide to do their thing, and wait to see how it plays out before looking at the screen again. In most cases, it’s because the pace they’re comfortable playing at involves making a decision once every 6–8 seconds or so.
In order to beat this pattern, all you have to do is run, stop at 3/4th screen, and just wait for them to show what they’re going to do so you can punish it on reaction. Since your opponent committed early and you didn’t, you get more time to react to your opponent’s decision instead of blindly guessing what they were going to do and when and where they were going to do it.
Broken down into an OODA loop, you could structure it like this:
Observe: My opponent and I are very far from each other
Orient: They like to run to half screen and engage there
Decide: I’m gonna run up a little bit and see if they commit to something I can react to
Act: Run, FD brake, and Observe again
Essentially, what we’ve done here is observe a pattern in our opponent’s behavior, make an inference about how the opponent player’s mental framing of the action (they prefer to take longer turns to lighten the game’s cognitive load), and pace ours to take advantage by calling out the early overcommitment and taking a short turn that throws off the rhythm.
This isn’t an unbeatable tactic, of course. If the attacking player recognizes what’s going on, they can shorten their loops to avoid creating predictably reactable situations, essentially playing chicken with our reactive position to cut down the amount of time we have to react to stuff, or they could go in with even longer strings of actions that can disguise where and when they’re committing (run to half screen, jump back, pause, late airdash jH). By keeping track of each other’s decision speed and turn length, we can lead each other into situations where we overextend or fall short and can pounce on them instead of gambling on hard reads everywhere.
Also, if you want to read more about this concept, check out The OODA Loop and the Half Beat.
Why this OODA loop stuff matters
As I noted earlier, we’re all generally pretty bad at talking about how to play fighting games, and while that doesn’t stop us from getting better at fighting games, it does make it pretty hard to communicate exactly what someone needs to do in order to improve, especially when it comes to anything more nuanced than “Don’t drop your combos”. Even advice like “focus on anti-airing” doesn’t really tell the player how to do it.
But by breaking the actual process of active fighting game play down into a series of phased turns that involve different parts of your brain and body, we can get a little bit more specific in the kinds of work we’re doing to play the game better. This won’t save us from having to do the work in order to actually realize those improvements in our play, but it does make it easier to understand where someone is in their fighting game journey and show them how to work towards becoming stronger in smaller, more specific steps.
I often find myself thinking that getting into fighting games these days is in some ways harder than it was in the old arcade days, because the 2020s FGC is saturated with footage of high-level play that can seem so impossibly far from where a new player starts out that it can seem hopeless to think that their confused mashing could ever turn into deliberate, refined play. It’s a lot easier to stick with the journey when you cannot see how high and how far the mountain goes, and can only see the peaks and valleys in front of you on the path there, because all you can see is the immediate goal in front of you. All the information is out there now, but no one is there to help you parse it. And even if you do manage to absorb it yourself, you’re likely to be just as stuck as the rest of us when it comes time to explain how you did it besides just literally describing all the things you did to practice it.
Hopefully this framework can be one of many teaching tools that we can use to make things a little more clear. Take it for a spin yourself and see if it helps!
Thanks for reading! If you found this essay valuable and want to support my work, please do not hesitate to share it around on your social channels, follow me on Twitter, check out my Twitch stream (Mon-Thurs 830PM-1030PM PST), or join my Patreon.