If can Measure it, can Improve it

It has been said that that which can be measured is that which can be improved. In order to improve your skills, having a common way to measure your performance would be very helpful. Simply being able to put a number to what you are trying to accomplish would go a long ways in this regard.

This is already put into practice in one sense, with having players earn tournament rating points under an ELO-type system since it was first proposed way back in the 90’s. However there are other things to measure and focus on concerning more incremental, individual improvement on more basic levels.

It would be good to be able to know and measure things like “How many frames am I actually spending in each input of my special moves like fireballs or dragon punches?” or even “How many milliseconds in those frames of each move’s inputs?” for even greater precision. Presumably, the less frames/time you spend in each required input of a given move, the faster your move will come out when you really need it at the last second, and you will also free yourself up to do more during the round.

Choosing the right thing to measure is important; you can’t just choose an arbitrary number to focus on as your measure of success. Even the ELO-based system is not perfect, but it can still give us a pretty good idea for certain things like seeding top players for a tournament. So even good measures won’t tell the whole story; it would probably be best to take a range of numbers, rates and ratios to get the whole picture, and each of those over a period of time.

When you express your progress over a period of time using a number to consistently measure your level of performance, it gives you something to focus on. It also keeps you honest; Did I really win all those matches this week because I’m getting better, or because my opponents had a bad day and let me get away with some things more often that they normally wouldn’t?


In starcraft there are programs that can take replays of games and measure actions per minute.
In those times people were shocked that top players had 200+ APM. People identified this and could practice it. Modern top players have 300-500 APM depending on the player.

If there was some way to measure such statistics in FGs then it’d be possible to improve those too.

because its good to be efficient and consistent

many other factors still come into play tho

Surely this would be possible with game roms, such as in GGPO? Measure inputs per minute like PP with starcraft? Someone code this.

yea it would only promote button mashing instead of skillfully timed executions

This is a tool that would help you learn movements… it wouldn’t give you extra points in the game…

yea, but fighters cant really be measured in that way like rts games can

rts games u really need to do as many things as u can as fast as u can
fighters its more bought proper timing and executing the rite move at the rite time

not how many total executions there r

now if this program can get specific…
like how many dp’s r executed and how many super’s r sucessfully executed
how many parries
how many combos…combo hit lengths and dmg
etc etc

this may be useful

but jus an overall total # of actions is quite indicative of nuthin at all

It would be cool to know how many moves executed vs. how many of those actually hit and such if only just because.

To be 100% honest, this topic seems kinda random and even after reading it, I’m not completely sure why you felt the need to bring it up BUT:

I couldn’t agree more with what you said there. The more I’ve played fighting games, and the more I’ve developed an understanding of them I’ve come to the conclusion that (well balanced) fighters are something like 80% timing/execution and 20% game knowledge. (Those are arbitrary numbers but I feel they’re pretty accurate.) Even the worst characters can win all day if you have perfect timing. I’ve been in plenty of situations where I know I input a move that should beat another move frame-wise and the only reason I get stuffed is because I didn’t time it right, or because I spent too much time (i.e. too many frames) inputting the move.

To shed light on your question about milliseconds though, a frame would be measured in frames-per-second, so that depends on the frame rate of the game. I don’t know for sure but I think most games are 60 frames per second, so if that’s true 1 frame is 0.02seconds, if it’s 30 frames, which is also possible (and seems more correct to me) then one frame is slightly more than 0.03 seconds. It seems pretty sick to think that anyone could develop reaction time THAT precise.

Then again, I’m not 100% sure I understand framerate, maybe the game is simply being processed at 60 frames per second (60hz) but that doesn’t neccesarily mean that a 4 frame move takes 0.08 seconds to come out since obviously an animation in a game could remain still for several frames at that frame rate and still look “fast” to the human eye. Can anyone elaborate on this?

Execution Aid Version 2

Well actually I DID code it:



btw 60 FPS is about 17milliseconds or 0.017 seconds (actually 0.0166666667 seconds).


Not even close. Otherwise, combo artists would be top at every game.
A top player can beat you with throws and one normal if they know the game better than you. At least in games where footsies/spacing are critical, execution is only a small part of the equation.

If you’re talking about Marvel or something then things change.

but combo artists tend to be pretty good players, just as top players can Execute with little problem. Their superior execution skills automatically make them a threat. Yes, that doesn’t mean they can consistently beat the best players, but there is a ton of overlap between doing well in a fighting game because you are Executing your moves well each time. If there is a top player that wins with only a small subset of moves, he’d better be sure to Execute those few moves perfectly, otherwise he is not a top player. Many pretty good players can do the exact same thing the top players can, just not as consistently. Execution counts for a lot in fighting games, more than in other games especially when you separate the top players from each other.

Penis enhancement pills are fake buddy.

On a serious note, you can always get better than your current skill lvl until you eventually join souls with Daigo. Daigo isnt 1 man, but many put together.

I agree with Kyokuji here. In games like ST, knowing character match ups and playing on those strategies will go a lot further than just good execution. This thread makes it sounds like you guys are talking about just 3s or something, but in my opinion, fighting games should be about mind games, not perfect execution. Otherwise you’d be able to play games mindlessly.

i think having more knowledge of a game like VF5 makes you better without he deep knowledge of the characters you wont know how best to fight them

so are you trying to figure out what would be beneficial stat tracking features for your program? i think snaek summed it up well, but seems nearly impossible given how hard it would be to make it discern a successful fireball from a failed/accidental one or even a mashed one in between rounds or during stun. in fact execution speed seems to be the only thing you could incrementally measure at all. i think you should focus on more refined ways of displaying the data so that someone could use it as an aide while watching a replay to have a better chance of telling what was luck and what was skill. like a simple place to start would be having it output a horizontal spreadsheet that reads like your score display but recognizes patterns and ranges of inputs as potential moves and then displays them.


1.0 1.5 2.0 2.5 3.0
(p1)qcf+p (p2)spd

I understand it’s not easy coding something like what’s in VF replays, but getting the program to display what is an actual successfully executed move in a streamlined spreadsheet or actual realtime overlay would be really nice in measuring execution consistency.

nice program btw. :slight_smile:

Execution Aid

Thanks man. Yes, I am thinking and trying to get us to think of the important stats to track.

It’s not impossible, I think it just takes a little imagination- like this, say we have a contest and we both do the following:

Use Execution Aid tool to record inputs for 10 fireballs in ST (in an emulator like GGPO’s), within say 15 seconds. The rules are that if any fireballs are messed up (don’t come out) then we have to try it over again from the start until we get a recording with all 10 FB’s in one go. (The data can be faked of course but we’re honest about it.)

We don’t even have to reverse engineer the exact timing requirements for a FB (in fact, if we do this a few times, I’m sure we’ll stumble upon the limits pretty quickly by interpolation).

We just take an avg of each of our input’s lengths (either in milliseconds for ultra precision, or just frames if you want); i.e. I take an average of all of my inputs, for my set of 10 fb’s, and compare it to the average of all of your inputs. The contest is who can get the lowest average.

I bet the players with the lowest averages are better, overall players (i.e. they usually perform better in tournaments).

One thing I noticed when I used Execution Aid- I roll the stick to the up-towards direction (i.e. past towards, when I hit punch). It rarely ever matters and I bet a lot of others do this to. I’ve been fb’ing for quite a while and have a better than average sense of what fb timing I can get away with- but have seen the best players squeeze out, just eek out a little bit more.

So, if you don’t believe this, that this stat, for one example, doesn’t really matter, and a top player’s stats won’t be all that different or necessarily lower, on avg, than an average level player. I think it does matter and, as Anotak said happened to Star Craft players, fighting game players are just as likely to be shocked to discover just what the top players are capable of.

Any takers? There’s little to risk here and a lot to gain from this.


Fuck, sweet.

This is some straight ass nerd shit.

As if I’m gonna be like “Hmmmm my 5 frames of DP vs. his 7 frames of HK, coupled with half-a-second input and 1.3 seconds lag, OH NOES odds against me!”, all this in a second worth of time? Ummmm, that’s not realistically gonna happen, sorry.

We are on a competitive fighting game forum, I think that’s to be expected.