New AAPG PC Player Stats Web Site

12357

Comments

  • PredictionPrediction Posts: 117Player
    Keebler750 wrote: »
    jgvn11 wrote: »
    Keebler750 wrote: »
    Do you all agree that a high KDR comes from playing lower skilled players?

    Yes and no, I think it depends on how many kills, hours played, ect. Eventually you will play against the good and bad. Unless your entire game play is all about leaving and joining games looking for the specific one you want.

    Okay. So, my premise was that the better quality player you play against when YOU are a good player, the closer you will be to a 1.0 KDR. As you go pubbing more, your KDR creeps up from that level.

    The problem with even THAT is that a less skilled player can ALSO have a 1.0 KDR!

    So, I'm trying to get us to discuss a way we can look at this, or bring in ideas from some other game that's already done it.

    What if AAV implemented a match making service that followed Microsoft's trueskill? Seems like it covers a lot of ground on this topic already.

    https://www.microsoft.com/en-us/research/uploads/prod/2018/03/trueskill2.pdf

  • Keebler750Keebler750 Posts: 3,320Beta Tester
    Thanks for that. I'll take a look and try to digest it!
    ______

    This has been a test of the emergency flame-fest system. Please do not adjust your set.
  • -hyperlite--hyperlite- Posts: 94Player
    I'll touch on the KDR thing. I think that is one of the quickest ways to assume skill.

    For instance

    If you are in the top 1% - 5% of the game.

    How often are you going to play with people at the same level as you. You aren't. Most of the time you will be playing with less skilled players, which is only going to increase your KDR.
    Even if you have players as skilled as you in the server. Depending on the game play, you will still pick off all the usual players until you have to face off the same skilled player as yourself.

  • Keebler750Keebler750 Posts: 3,320Beta Tester
    edited December 4
    I'm interested in your assessment there, Hyperlite. Are you saying that picking off the low skilled players results in a high KDR that DOESN'T reflect skill, or does? Your choice of the word 'assume' can have a couple of connotations, so I'm trying to clarify.

    It seems to me the only time skill is actually tested is against tough opponents, ergo the KDR starts to stray AWAY from a skill assessment metric, since low skill players virtually walk right into your crosshairs (compared to a tougher opponent.)
    ______

    This has been a test of the emergency flame-fest system. Please do not adjust your set.
  • -pR|Arkeiro-pR|Arkeiro Posts: 714Player
    What if you're on the top of the food chain and everyone else is less skilled than you though? - Maybe that's why the hawkeyes, silvers and goldens stack up, thats why after thousand and thousands of kills the KDR is still high, thats why probably after hundreds of scrims and tournaments games the KDR isnt affected still. I don't know :awesome:

    Over 1.1k Golden Hawkeyes.
  • doogle!doogle! Posts: 573Player
    Keebler750 wrote: »
    Do you all agree that a high KDR comes from playing lower skilled players?

    Yes, it definitely can. I'd still say though, that those folks on that KDR top list are still really good players, regardless of who they're playing against. The community is so kick/ban heavy, you can't possibly go 25-5 in servers for too long w/o getting banned in private servers.

    But one thing i've found...especially in a FLO server..for every high skilled player, there are 8 other low skilled players to be killed. Definitely could still pull a 1.5+ in a FLO w/ good players. But BDX on the other hand, the odds of running into that high skilled player increase
  • -hyperlite--hyperlite- Posts: 94Player
    I used the word assume bacause there are people out there that strive for K/D only and play that way. A high kdr tends to reflect skill in most instances.
    Im just saying like Arkeiro said. If you are at the top of the food chain then your k/d is only going up. Doesnt matter how many other skilled players are in the server. You will pick off the easy ones until you have to engage with someone as skilled as you and then if you die, your k/d isnt going to change much. You've already killed all the bottom fraggers.
  • -hyperlite--hyperlite- Posts: 94Player
    Also, I noticed there is no average damage dealt per round. I think that would also be good to add as an indicator of skill.

    Some type of metric of KDR and damage would be where I would start for a toplist.
  • aaHollywoodaaHollywood Posts: 343Developer
    Prediction wrote: »
    What if AAV implemented a match making service that followed Microsoft's trueskill? Seems like it covers a lot of ground on this topic already.

    https://www.microsoft.com/en-us/research/uploads/prod/2018/03/trueskill2.pdf

    Yeah we are. There are a number of these services that natively integrate with UE4 that we would have to code by hand in UE3, and trueskill is one of them. Others, like Gamelift to manage server numbers and player backfill, will also be part of AAV. B)
  • doogle!doogle! Posts: 573Player
    Prediction wrote: »
    What if AAV implemented a match making service that followed Microsoft's trueskill? Seems like it covers a lot of ground on this topic already.

    https://www.microsoft.com/en-us/research/uploads/prod/2018/03/trueskill2.pdf

    Yeah we are. There are a number of these services that natively integrate with UE4 that we would have to code by hand in UE3, and trueskill is one of them. Others, like Gamelift to manage server numbers and player backfill, will also be part of AAV. B)

    Hopefully that's in a completely different game mode...dare I say, a competitive/ranked game mode. The queue times may be very long unless the playerbase is much larger than it is now.
  • =IK=Doba==IK=Doba= Posts: 2,780Player
    edited December 4
    Also, I noticed there is no average damage dealt per round. I think that would also be good to add as an indicator of skill.

    Some type of metric of KDR and damage would be where I would start for a toplist.

    Most certainly! As K/D ratio is a statistical representation used to measure how good you are ,the KDR or ADR (average Damage per Round as in csgo) also FPR (avg. Frags per round) is used to represent how useful you are for the team.

    If you're interested in establishing skill rankings, you're gonna have to have some mathematical equation involving more than just one variable like KD.. the more complicated the more accurate.
    _____________________________
    #Support Comp Mode

    https://www.youtube.com/channel/UCN4YhM6jUB2MxVj8i3b9rhw
  • Keebler750Keebler750 Posts: 3,320Beta Tester
    Good discussion! Thanks, Hyper, Doogle and Doba! :+1:
    ______

    This has been a test of the emergency flame-fest system. Please do not adjust your set.
  • Holy moly, this looks [TOS Violation] good. Nj


    Whatever you do, you cannot outrun the devil.

    Steam
  • Keebler750Keebler750 Posts: 3,320Beta Tester
    Okay, I've read to the point in that TrueSkill document where it starts to get math-heavy and I kinda choked on it. :(
    "Given a probabilistic generative model for match results, skill ratings are computed by Bayesian inference."
    :o
    ______

    This has been a test of the emergency flame-fest system. Please do not adjust your set.
  • PredictionPrediction Posts: 117Player
    Keebler750 wrote: »
    Okay, I've read to the point in that TrueSkill document where it starts to get math-heavy and I kinda choked on it. :(
    "Given a probabilistic generative model for match results, skill ratings are computed by Bayesian inference."
    :o

    Where is BCPULL when you need him :)
  • Keebler750Keebler750 Posts: 3,320Beta Tester
    Yeah. I'll message him..... :)
    ______

    This has been a test of the emergency flame-fest system. Please do not adjust your set.
  • doogle!doogle! Posts: 573Player
    See, we can all get along as long as you all agree with me.
  • Keebler750Keebler750 Posts: 3,320Beta Tester

    :D
    ______

    This has been a test of the emergency flame-fest system. Please do not adjust your set.
  • [soldier][soldier] Posts: 66Player
    edited December 5
    We have added a Top Players List to the AAStats home page. The list can be sorted by any of the stats columns (Kills, K/D Ratio, Exp Points, Matches Won, Win Pct, and Hours). When the list is sorted by K/D Ratio or Win Pct, only players with 100 or more completed matches are included:

    0hlo3pxzgw06.png

    Still planned for the Top Players List is a search field so you can find specific players on the list.

    We have also made some changes in the Player Stats page. The General Statistics section has been rearranged somewhat, and some new stats for Experience Points, Matches Lost, Match Win Percentage, Kills per Match, Points per Match, Points per Round, and Favorite Map have been added:

    fjash2vwiyyr.png

    Since we received some negative feed back about the readability of the blue background and table cell borders we are trying out some alternative looks on the Player Stats page.
  • -hyperlite--hyperlite- Posts: 94Player
    Thanks for the update but I'm hoping to give you some constructive feedback here.

    Are you able to get information on how much average damage someone does per round

    The Toplist - Not the direction I would have gone but it is an option. For me I would think the Toplist would have some type of K/D and Damage ratio metric.

    Top Kills - I would remove the unoffical leaderboard if you keep with the current format of Toplist because you can now sort

    Remove takedowns as I dont think it is of importance

    Remove Dev Kills - I dont see any value in parsing that info

    Highest round score - I look at those numbers and I'm confused as to if that is match score or if it is really per round score.

    Secures - I dont see the value in parsing that info

    Favorite map - I dont know if I would say "favorite map", I would probably change that to "most Played map"

    On a side note:
    The player with a 5.71 K/D is either cheating or found a way to manipulate the info being parsed. The guy is averaging 16 kills a minute
Sign In or Register to comment.