Pretty comprehensive again rev nice work, you’d probably also find about some habits you didn’t realise you had.
Say P1 has a skill of 250 and P2 has a skill of 500. Depending on who killed who, the stat would increase or decrease accordingly. Larger gap in skill would mean a larger increase/decrease.
P1 kills P2. P1’s skill increases a lot while P2’s skill decreases a lot.
P2 Kills P1. P2’s skill increases a little while P1’s skill would decrease a little as well.
@KingCraftUK , Here’s what I meant about an “Overview”. Same basic layout, but it would include “all” player time, for every mode/map etc. Similar to what we get on the details page, but with more … details.
Part of my job is to design, calculate, code and display statistical data for my clients. Once you know all the variables and data streams it’s pretty straight forward.
Player stats are for the subject player first, then everyone else. Map based stats highlights/calculates all the good, bad and ugly in one page vs 376 battle reports.
Ah, I see. Personally I don’t mind having more stats like this, but my main point here would be that a stat like “skill” could be changed to instead be a reflection of the combination of all these individual stats. It could for example take into account your efficiency as a medic as well as your ability as a shooter as well as your efficiency within a squad (or as part of the larger team).
I think as it stands the skill stat is perhaps a little confusing and misguided, but it has the potential if done well to be the most true reflection of a players ability. And that’s a statistic that I am all for.
I totally agree, skill stat is totally broken. It needs to be gone. Its like a 1972 Cadillac that’s loaded and as soon as you drive it off the lot its lost 10,000 in value.
I see what your saying.
But the example I gave would be exactly like taking everyone of your Golmud conquest battle reports and combining the data to give you a complete overview of your performance (stats) for Golmud Railway.
That’s all that it is. Infact from a coding perspective that’s how it would be generated for the most part.
Who cares how good you are. You are still gonna play. The 8 year old that begged his mama to buy the game is gonna play.
Skill as a stat is completely broken. I mean if you thoroughly review a player’s reports you could get a better understanding of “how it pertains to them”. A quick glimpse … nice number … ok he/she is a good player just doesn’t cut it. But that’s how 90%+ use it.
That 900 Skill player could be a B flag metro support player. 2 ammo boxes deployed, spamming nades and suppression fire on a high ticket server. The example stats would highlight it in neon and strobes.
There is some truth to this , not gonna lie : /
True, but that’s a problem with its current implementation is all I’m saying. That doesn’t mean that there isn’t potential for it. In the example you’ve given there the aim would be that it would be reflected in the stat. If we have the ability to have all these amazingly detailed stats, then it wouldn’t be too big of a stretch to create a reasonable combination of them.
I agree. The data that Dice pipes downstream should be enough to code a proper “Proficiency/Efficiency” stat.
Skill algo as is:
Skill is normalized (as in between 0 and 1), but for display purposes it’s multiplied by 1000 and rounded to nearest integer. In other words 100% perfect skill would be 1000. OTOH, someone scoring no points at all, would have score of 0.
For a single game skill 60% of skill comes from SPM (capped at 1000), 30% comes from KPM (capped at 3.0), and 10% comes from KDR (capped at 5.0).
BTW. note that these calculations use SPM which excludes any bonuses and boosts (ie. as shown on scoreboard), while BattleLog displays SPM including bonuses and boosts.
To calculate this you first have to normalize each component:
normalized_SPM = min(SPM/1000.0, 1)
normalized_KPM = min(KPM/3.0, 1)
normalized_KDR = min(KDR/5.0, 1)
Now apply the percentages :
normalized_skill = .6 * normalized_SPM
+ .3 * normalized_KPM
+ .1 * normalized_KDR
skill = round(normalized_skill * 1000)
So above is skill for just one game. But what’s is displayed in BattleLog is overall skill, which is calculated to the formula is 90% skill prior to current game + 10% current skill:
new_skill = round(.9 * old_skill + .1 current_game_skill)
In the battle report you can also see how the skill changed, which is calculated as follows:
skill_diff = new_skill - old_skill
It’s pretty much based on Kills (KDR & KPM). Vehicle only players gain a huge advantage from the get go.
Impressive 
Why not replace the numbers with a handful of simple colored graphs (green-orange-red). With a quick glance you could see if your gaming is improving or not.
You people forgot to include a graph of the internets of the players. The internet is tied directly to the skill stat. And some OORs with crummt internet dominate the game. Factor in interference for some players, faulty equipment and high traffic on internet makes for packet loss. So player A shoots player B in the back. Player B turns around and shoots player A. Player A dies. This happens a lot because of packet loss.
Now in Explicits case she shoots players A,B,C,D,E,F and G and they all die. So don’t believe her when she says she is not skillful.
“Skill” is simply to general and loosely defined of a word.
What skill? Gunplay? Team play? What about vehicles or objective play?
It actually tries to factor in some of those but that makes the stat even more convoluted somehow, because people who know the equation can boost it, and people who don’t are left wondering why they lost 3 skill on what they thought was one of their best rounds that week
I definitely think that if there was no skill stat, then people wouldn’t go shoot I’m not gonna play any more because of this dum skill stat.
“I can’t aim, so accuracy cant be a factor in skill”.
accuracy should be a factor.
not saying we need a skill stat, but if you dont judge a player by his accuracy when determining what you think of him, you’re missing a huge component of real skill.
yes, this is how part of BF4’s skill stat and all of BF3/BC2’s skill stat worked.
Its a basic ELO system and is designed to measure head to head confrontations. something which doesnt often exist in a pure form in battlefield.
not only is it overboard, its almost certainly far more than DICE is willing to invest server wise to accomplish. DICE limited us to 100 friends because they didnt want to invest in more server. They limited our battlereports and they stop archiving after a few months. you really think they’re going to give every player a 20 map stat breakdown? hell no.
We dont need this. just give us the basic stats. Jesus, we dont need a thesis on a player’s map to map performance.
we dont need a map breakdown that spm, accuracy, KD and weapon stats dont already tell us. objective related stats help too. but this is way more information than anyone needs and it would take a lot of effort and investment to pull off for a full playerbase.
you’r not going to be able to graph this without turning origin into spyware, and frankly, it doesnt matter. if you blame OOR players for scrubbing, you just suck, period. playing OOR or with packet loss isnt an advantage. this myth needs to die.
You sound like someone else I know. If you have crummy internet or you are so far away, sometimes your internet drains the server and you own. Fact of host based servers.