Best Frag Or K/d

1246712

Comments

  • iNv|eKCommiNv|eKComm Posts: 394Player
    edited October 2015
    AliGator.. wrote: »
    eKComm wrote: »
    Everything is circumstantial. Literally, every stat could be subjective depending on how the light you shed on it. The only stat that matters is championships won. Checkmate :wink:

    Seriously though. We've been down this path before. Quantifying skill is very difficult. Its hard to define what skill is. At the end of the day none of this matters. The more matches you win in competition (not pub) define how successful you are as a player. Thats as close to 'real' skill we can get. Or at least its the least diluted statistic and most consistent.

    I agree with most of what you are saying but comp isn't everything. Are you saying people who don't play competitive have no skill and are unsuccessful players? I believe you have to look at everything to know how good a player really is. That includes competition, stats (not only KDR, everything), and reputation/history. Just as an example, lets say "no name gaming" has a member that never starts but his/her team is the best team in the game and literally wins everything, is he/she one of the best players in the game? No.

    Just because i don't explicitly say everything doesn't mean i'm implying anything. I'm simply saying the least diluted and most atomic indicator of skill is your competitive matches won. Pub is alot more diluted.

    I feel that KDR is the best recorded statistical indicator of skill. Surely, outliers exist. But outliers dont define sets, they're just exceptions. You can be the biggest KDR [TOS Violation] but if your bad a better player will still kill you. Furthermore, KDR is much harder to fake than other stats. Some will say that they can easily pick and choose the right servers and get some crazy KDR but over an extended period of time the regression to the average makes a victims of all players.

    Factually, better players tend to have higher KDR's. Understand that thats a tendency which means its generally true but occasionally false. There surely exists situations where good players have bad KDR's. When compared to other statistics though KDR has a much closer relationship to personal skill.
    2016 Flank Gaming Network Season 1 Champion
    2016 FraggedNation Season 4 Champion
    2015 FraggedNation Season 2 Main Champion
    2015 ESL Test Cup Second place
    2014 FraggedNation UMM Tournament Champion
    2014 FraggedNation Old School map Tournament Champion
    2014 TWL Season 1 Second Place
    2013 TWL 7v7 Beta Tournament Champion
    2013 TWL 5v5 Beta Tournament Second Place
  • iNv|roKis-iNv|roKis- Posts: 420Player
    ^
    This guy is so passionate
    Bhop is a string of jumps with no interruptions.
  • .dcG-Colts^.dcG-Colts^ Posts: 1,973Player
    SSKnecabo wrote: »
    W/L ratio is easy playing as a party. Depends on the players ofc but if you have 6 decent players every official server is yours.

    yah W/L ratio don't mean much at all I could join a match win 1 round get a match win for it. I could join a match loss 1 round and get a match loss for it.
    Pie charts + Graphs= Very Bad.



  • -=DA=-Tw!$t3d-=DA=-Tw!$t3d Posts: 65Player
    AliGator.. wrote: »
    HAHA So around 160 kills per hour? Yeh he is definitely legit.


    My point exactly, sir. Guy was 17-0 at the end of round two. Thats shady even for a h@cker like you Ali :D <3 (j/k)
  • .6ix.AliGator.6ix.AliGator Posts: 165Player
    AliGator.. wrote: »
    HAHA So around 160 kills per hour? Yeh he is definitely legit.


    My point exactly, sir. Guy was 17-0 at the end of round two. Thats shady even for a h@cker like you Ali :D <3 (j/k)

    exactly! <3
    ss_4_AliGator...png


  • .dcG-Colts^.dcG-Colts^ Posts: 1,973Player
    edited October 2015
    Can't there atleast be a system in place that flags guys average a lot more kills per min or hour than everyone else. This guy averages 2.6 roughly a min, 160 hour so if he got added to some sort of flagged list and the devs or admins or who ever deals with this kind of stuff can watch his demos and guys like this would get banned way sooner.

    That goes for any miracle stat too. Like I am a rusher in pubs I'm not usually sitting back camping and I average 0.8 kills a minute and 47 kills an hour. It would be a pretty easy way to catch unlegit players fast if you ask me.
    Pie charts + Graphs= Very Bad.



  • .6ix.AliGator.6ix.AliGator Posts: 165Player
    Can't there atleast be a system in place that flags guys average a lot more kills per min or hour than everyone else. This guy averages 2.6 roughly a min, 160 hour so if he got added to some sort of flagged list and the devs or admins or who ever deals with this kind of stuff can watch his demos and guys like this would get banned way sooner.

    That goes for any miracle stat too. Like I am a rusher in pubs I'm not usually sitting back camping and I average 0.8 kills a minute and 47 kills an hour. It would be a pretty easy way to catch unlegit players fast if you ask me.

    That would be a cool addition. Basically the things to look for are high KD (I would say anything over 4.5 or 5), high Kills per hour (anything over 70-75), and a high head shot % (anything over 70%?).
    ss_4_AliGator...png


  • SacchoSaccho Posts: 1,577Player
    AliGator.. wrote: »
    Can't there atleast be a system in place that flags guys average a lot more kills per min or hour than everyone else. This guy averages 2.6 roughly a min, 160 hour so if he got added to some sort of flagged list and the devs or admins or who ever deals with this kind of stuff can watch his demos and guys like this would get banned way sooner.

    That goes for any miracle stat too. Like I am a rusher in pubs I'm not usually sitting back camping and I average 0.8 kills a minute and 47 kills an hour. It would be a pretty easy way to catch unlegit players fast if you ask me.

    That would be a cool addition. Basically the things to look for are high KD (I would say anything over 4.5 or 5), high Kills per hour (anything over 70-75), and a high head shot % (anything over 70%?).

    I scanned the database for such things during open beta and didn't find much. You have to be careful not to accidentally catch false positives too; a bad player might show up with a high headshot kill % because he never lands more than one hit in a burst before dying so the only firefights he wins are from lucky heads. There are tons of ways the stats can lead you wrong. It can be done very well, but it can take more effort to really do right than you might think. I'd fit Gaussians over every stat and weight player evaluations on their position inside each distribution.

    The database error that replicated rounds a few hundred times also screws with how well the player stats can be used as a metric. That error made it look like I have a 94% chance to hit an arm with the AK. Imagine if I got headshots during those rounds instead? It'd look really suspect.

    The rise of stats analysis sites for other games (trying to do exactly what you guys are talking about in AAA titles) seems to have led to players being more cautious, more subtle, with what they do. The obvious players, the ones that stand out with crazy numbers, are really easy to catch as it is. People just need to be reporting them.
  • .dcG-Colts^.dcG-Colts^ Posts: 1,973Player
    edited October 2015
    Yah but we are not saying such stats leads to an instant ban we are saying they get flagged and a dev or admin or w/e reviews their demos. If the player is obviously unlegit he can be dealt with that way. But I think we need a better system to flagged such users I don't think reporting them is enough, because not everyone is going to report a unlegit player they see. So flagged unreal stats will help that process.
    Pie charts + Graphs= Very Bad.



  • -=DA=-Tw!$t3d-=DA=-Tw!$t3d Posts: 65Player
    I can totally program that into DA Live, and have it construct a list of "Well above average players" or whatever. All the pieces are already there, just need to glue them together. If you guys wanna add me on steam "datwisted718" we can start to put it together.
  • .6ix.AliGator.6ix.AliGator Posts: 165Player
    I can totally program that into DA Live, and have it construct a list of "Well above average players" or whatever. All the pieces are already there, just need to glue them together. If you guys wanna add me on steam "datwisted718" we can start to put it together.

    If I participate/help, will I be allowed in your server? You can use my stats as a guideline!
    ss_4_AliGator...png


  • -=DA=-Tw!$t3d-=DA=-Tw!$t3d Posts: 65Player
    edited October 2015
    Ok so i think i've figured out a formula to put a rating to each player based on their stats.

    Here is goes
    ((((A * KDR) + H) * K) / 1000) = PR
    

    A = Accuracy
    KDR = Kill:Death Ratio
    H = Head Shot Percentage
    K = Kills Per Hour
    PR = Player Rating

    Sample Output
    -=DA=-TW!$T3D
    A = 12.06
    KDR = 1.36
    H = 34.85
    K = 24.467659072966
    PR = 1.254006675744
    

    -colts.iog-
    A = 13.32
    KDR = 4.45
    H = 41.69
    K = 47.129770992366
    PR = 4.7584101984733
    

    AliGator...
    A = 14.25
    KDR = 3.71
    H = 42.54
    K = 61.888697424945
    PR = 5.9046458995704
    

    A Player with stats that cannot be honest...
    A = 15.73
    KDR = 5.89
    H = 28.95
    K = 143.43261841378
    PR = 17.44136336933
    

    It seems to be a good representation of player skill. Even with colts High KDR, he's statistically ranked lower than ali because his Kills per hour are lower. And the last guy, his Head shot percentage and accuracy alone wouldn't raise a flag, but when you figure his KDR and kills per hour, we raise a flag there.

  • =IK=El.Jefe=IK=El.Jefe Posts: 72Player
    Accuracy doesn't belong. Some people spam fire a lot more than others and different weapons are used such as sniper rifle which have a "native high accuracy rate". Maybe you could include pistol accuracy, since this is much less commonly spam fired and used by all classes. You guys said that matches w/l are an unreliable stat, but rounds w/l is ok. So maybe include that. Just throwing ideas out.
  • .shhfiftyfive-.shhfiftyfive- Posts: 495Player
    edited October 2015
    that formula will be useless. accuracy and headshot percentage in a formula for player rating will just lead to really good players having widely different ratings based on role/playstyle...

    plenty of good players don't bother with accuracy. they spam walls or prefire lanes as a form of map control and teamwork. this also lowers headshot percentage. suppression means to discourage the enemy from feeling safe while shooting your teammate. so you unload the entire mag on them, not at all going for a headshot. plenty of times when you are suppressing a target they will die to body shots because they are low health.

    also, your kills per hour and K:d will drop considerably if you play on servers where the majority of the enemy team actually has a clue. that may be rare, which just reinforces the point. plenty of casual players can get a really high rating because all they play against is clueless kids who just downloaded a free game and are super easy to kill without even being fired upon. that doesn't make them better than a player who regularly plays against a higher quality opponent, and thus has lower pub stats as a result.

    basically there's no need/point to try to use aapg pub stats to determine who is best (rating).


  • =IK=Doba==IK=Doba= Posts: 2,789Player
    But it's not for determining who's best rather who's fishy.. if your rating is drastically higher than some of the best players in the game rightly so a red flag should be raised for player review.. I like this
    _____________________________
    #Support Comp Mode

    https://www.youtube.com/channel/UCN4YhM6jUB2MxVj8i3b9rhw
  • .shhfiftyfive-.shhfiftyfive- Posts: 495Player
    i was responding to twisted's futile attempt at a formula to create player rating, not the idea of using insane stats to be flagged internally as an anti-cheat measure.


  • -=DA=-Tw!$t3d-=DA=-Tw!$t3d Posts: 65Player
    Shifty. Instead of being a condsending [TOS Violation]....let's work together to achieve the end result. Its ment to separate fishy stats from legit ones. You need to relalize that. Someone else can worry about rating all the stat jocks. Again. This is only to do two things.....1. Flag possible dishonest players and isolate those accounts effected by the stats bug. And from what I can tell it works.
  • .shhfiftyfive-.shhfiftyfive- Posts: 495Player
    im not being condescending. i'm simply saying, you can take each stat individually and find the "fishy", instead of combining a bunch of stats into a formula and some rating, which will not help you...


  • =IK=Doba==IK=Doba= Posts: 2,789Player
    The player at question has weird stats..normal accuracy, very low headshot %, mixed with stupid not normal high per hr kill, which makes a non realistic KD...

    You'd think someone with high hr kill rate would have high HS %
    _____________________________
    #Support Comp Mode

    https://www.youtube.com/channel/UCN4YhM6jUB2MxVj8i3b9rhw
  • TrIn@dOr^SuRTrIn@dOr^SuR Posts: 215Player
    That would be a good addition -=DA=-Tw!$t3d to your sig maker!!
This discussion has been closed.