Now, I know you guys all hate the "skill" statistic and think it's worthless, but I think that's MOSTLY because you don't understand it.
Now that you'll understand it, you'll understand why it's a fairly worthless (but still quite telling) statistic.
Essentially, every game of Battlefield has a base theoretical "perfect" game. Long story short, your gameplay matched up against the theoretical "perfect game" is compared to how well you've done in recent games. If you did better, you get a bump in skill. If you did worse, you drop skill points.
So what is a "perfect" game of Battlefield? It's made up of three different (or I should say "different") stats, each weighted differently, and not the way you'd expect.
1000 Score per minute (weighted at 60%) - Note that this is BASE score per minute, and does not include any awards/boosts
3.0 Killer per minute (weighted at 30%) - Holy shit that's a lot of kills per minute
5.0 Kill/death ratio (weighed at 10%) - Hey look, a metric I can actually achieve
Alright, so there's a "perfect game" of Battlefield. I mean, that's pretty reasonable, eh? If I had a teammate who always had the above scores, I'd be pretty freakin' happy.
---------------------------------------------------
Okay, so we know what a "perfect game" looks like, but what does that mean for my skill stat?
One important factor is that the "perfect game" of Battlefield is comprised of the maximum values for those statistics. If you go 18-0 in a game with 1.1 kpm, your k/d will only reflect a "perfect" 5.0. If you somehow score 1k+ per minute in the base game, again, that is dropped down to 1000 max. If you somehow manage 3.0 kills per minute, well then fuck off because I hate you.
In order to calculate skill for a given game, you need to have a base skill score to work with. Once you have that, you plug it into this formula:
nSPM = (SPM/1000)
nKPM = (KPM/3)
nKDR = (KDR/5)
oSKL = (Starting overall skill)
nSKL = (Game skill)
So for the first part we have:
nSKL = ((nSPM*0.6)+(nKPM*0.3)+(nKDR*0.1))*1000
Now, you're NEW game skill is an amalgamation of this number with your old skill, weighted at 0.9 for your old skill and 0.1 for the game you just played:
New skill = (oSKL*0.9)+(nSKL*0.1)
Got it?
No? That's cool. We'll do an example then, using some of my recent games. The numbers are rounded off for simplicity.
-----------------------------------------------------------------------
Link: http://battlelog.battlefield.com/bf4...152/190879989/
nSPM = 423/1000 = 0.423
nKPM = 1.74/3.0 = 0.58
nKDR = 2.57/5.0 = 0.52
oSKL = 372
nSKL = ((0.423*0.6)+(0.58*0.3)+(0.52*0.1))*1000 = 479.8
New skill = (372*0.9)+(479.8*0.1) = 382.78
Well, what do you know? http://imgur.com/hUlCbgp.jpg
------------------------------------------------------------------------
So what have we learned? Well, for one thing, it's easy to see why skill starts to plateau at some point. For the above game, I only got an 11 point bump. Had I played that game earlier in the night when my skill was around 330 (started that night at 391, damnit :/), I would've earned 15 points. If I had already been at a skill of say 500, I would've lost 3 points for that game! Fuuuck that.
What else have we learned? Well, while the skill stat is an OKAY indicator of very recent performance, it has several glaring flaws that keep me from taking it too seriously.
While a skill level of 400+ likely means that the person is at least a good player who can do well one on one, it doesn't really say anything about their teamwork. Score per minute is very heavily influenced by k/d and especially kills per minute, so your amount of kills are being counted threefold here. In a game where you can have a 0.8 k/d and still be MVP, this is obviously flawed. Even a perfect score of 1000 spm with no kills and at least one death will only give you a max skill of 600, and it's fucking impossible to get 1k spm (remember, awards don't count) without killing a shitton of people.
I like the a) spm is weighted so heavily and b) k/d is weighted so lightly (I might even bump that up to .25 myself, because fuck you, k/d does actually matter in any gametype, regardless of what people say. If you're costing your team more tickets than you're taking from the other, you need to have an exceptional PTFO score to make up for that, and that's not easy with how far CQ flags are from each other). But I don't understand why kpm is weighted so heavily, especially since it's already so heavily featured in your spm. Maybe replace that stat with some sort of objective measure? Say, defense, attack, and captures scores per minute? Not sure what the theoretical perfect score would be for that, but surely one exists that might make this a better measure?
Say maybe 0.5 spm, 0.3 PTFO, and 0.2 kdr would be a good split. Not sure. Anyhow, now that we understand how skill works, we can discuss it rationally.
That said? It's still quite useless, but again, pretty hard to ignore when gauging the skill of an enemy player.
Reply With Quote
Originally Posted by BlameItOnGod13
