Me my ELO and I

dePARAdePARA Join Date: 2011-04-29 Member: 96321Members, Squad Five Blue
edited November 2013 in NS2 General Discussion
The forum is full of threads about stacking, imbalanced teams and why there are no mechanism to solve that.
One try is the ns2stats ELO based random from shine-adminmod.
So you need ns2stats and shine running on your server, thats it.

Why i am writing this?
Cause im tired of this stacking paranoia and things like "NOO, dont vote for this ELO crap"
In the end everything is stacking. People whining about stacking after pure random also wich is nonsense for sure.
Random cant be stacking, its random. But thats another story.

If an server run ns2stats, every player get an ELO score wich is shown in this picture.
dabciqoa.png
1500 is the base score every player become after the 1st game on an ns2stats server.

If you want to now how this score is calculated exactly ask the ns2stats crew.

What happen if you use the shuffle mode from shine-admin in combinatin with ns2stats?
- It splits 8 players with the highest ELO score (these are the carry players typically)
- It use the average KD/R of the last 3 rounds for the other players to mix the teams.

So the shine ELO vote is a mix between KDR and ELO.

Why there are still "stacked" teams after a succesful vote wich can end in an stompfest?
There different reasons for that.

- Sometimes the carry players leave after roundstart and rookies join in
- People reconnect to the server to join the other team (The votesystem block the ability to swicth the teams)
- One team has great teamplay (this is something no system in the world can "balance")

To summarize this:
We have systems against stack but its the human factor that counter these systems.
I think EGO-shooter is not only the description for an game genre in this case.
And as i said before: in the end EVERYTHING is stack.
«13

Comments

  • 2cough2cough Rocky Mountain High Join Date: 2013-03-14 Member: 183952Members, Reinforced - Shadow, WC 2013 - Supporter
    dePARA wrote: »
    - One team has great teamplay (this is something no system in the world can "balance")

    Booooooom, qft. Nail on the head there.
  • CCTEECCTEE Join Date: 2013-06-20 Member: 185634Members, Reinforced - Shadow
    True words.
    1 minor point: Its 'Elo score' not 'ELO score'. It's based on a chessplayers name and not short for anything.
  • cooliticcoolitic Right behind you Join Date: 2013-04-02 Member: 184609Members
  • Kouji_SanKouji_San Sr. Hινε Uρкεερεг - EUPT Deputy The Netherlands Join Date: 2003-05-13 Member: 16271Members, NS2 Playtester, Squad Five Blue
    edited November 2013
    coolitic wrote: »
    Whats ELO?
    A British rock group
  • GhoulofGSG9GhoulofGSG9 Join Date: 2013-03-31 Member: 184566Members, Super Administrators, Forum Admins, Forum Moderators, NS2 Developer, NS2 Playtester, Squad Five Blue, Squad Five Silver, Reinforced - Supporter, WC 2013 - Supporter, Pistachionauts
    edited November 2013
    Nice post depara :)
    coolitic wrote: »
    Whats ELO?

    I originally planned a wiki page about this,but i forgot to wrote it.

    It's a system developed by Arpad Elo to rank and rate Chess players.

    Basically a player starts with a startvalue of x (at ns2stats.com 1500) and losses or gets a y amount of points based on the difference between his and his opponents ELO rating every time he looses or wins a game.

    As NS2 is a teamgame we average the other teams ELO Ratings to determ your ELO rating change.

    ELO is used in the same way in games like Guild Wars, World of Warcraft, Leage of Legend, Star Craft 2,Counter-Strike and many more .

    Issues with the ns2stats.com ELO rating:

    Next to the other mathematical issues of the ELO System ns2stats.com has the big problem that it doesn't cover alot of games.
    Therefor your Elo can be pretty different from your actual skill level due to the fact that it could be the case that you have just luck and win all games on a ns2stats enabled server.

    Normally those things would be covered by the theory of probabilit , but as the random samplings of ns2stats are yet ways to small you have to see the elo as suggested
    skill rating of a player.

    Your Elo will come closer to your actuall skill level if you play a lot at ns2stats enabled servers against other by the ELO rating even players.


    Got interessted and want to know how exactly the ELO rating works, go here to get the basics: http://en.wikipedia.org/wiki/Elo_rating_system
  • MouseMouse The Lighter Side of Pessimism Join Date: 2002-03-02 Member: 263Members, NS1 Playtester, Forum Moderators, Squad Five Blue, Reinforced - Shadow, WC 2013 - Shadow
    *Moved to General Discussion*
  • UncleCrunchUncleCrunch Mayonnaise land Join Date: 2005-02-16 Member: 41365Members, Reinforced - Onos
    Also the main problem is : ELO is a single number.

    One number doesn't tell much. K/D is never gonna do it. You can have a whole team of pistoleros with zero teamwork that never gonna do anything. Right now it is close to useless.


    I would see :

    *Combat
    -K/D & Assist (Multiplied by classes. KD when fade etc.)
    -Enemy structure harassment. This would give a player teamwork meter for field unit. Killing Rts takes time.

    *Builder stats
    -Base elements
    -RTs
    -Gorge tunnel

    *Commander
    -Delay between alert and focus or dedicated actions


    As every action gives points it is possible and would be more accurate.

  • VetinariVetinari Join Date: 2013-07-23 Member: 186325Members, Squad Five Blue, Reinforced - Shadow, WC 2013 - Silver
    Have fun pressing that in one formula (which gives you a single number in the end, too, btw, because otherwise theres no other way to compare).
  • GeekavengerGeekavenger Join Date: 2012-08-31 Member: 157117Members, Reinforced - Shadow
    dePARA wrote: »
    Why there are still "stacked" teams after a succesful vote wich can end in an stompfest?

    Also because the game is designed to snowball if the teams balance is off by even a little (As something as simple as good team work can accomplish). The team that can push the battlefield in their favor early, gains a marked advantage in both tech and resources. So yes the other team can come back but it is much harder. For example, even if the leading team which has managed to tech up has a set back (loses a higher lifeform, failed shotgun push... etc) the lost res isn't as much of set back because they already have the res flow to recover, where as a comparable setback for the losing team is a MUCH bigger deal, might as well spam concede (or my preferred method for conceding, crazy all in full team pushes over and over again).

    This is how the game is designed and quite frankly a good thing, no body wants a 2 hour trench fight every round. Once and a while it is super satisfying, every time it would be tedious.
  • dePARAdePARA Join Date: 2011-04-29 Member: 96321Members, Squad Five Blue
    I dont think we need a super complex formular to "balance" teams a bit.

    A gorge speeding up the hivebuild, healing up the teammates are very important for sure.
    As skulk biting res like an idiot is very important also.
    But are these two the reason for stomping a team? Not really.

    NS2 is a shooter and the team with the better aimers/biters/.. win normally.
    If one marine can kill 4 skulks alone again and again this one marine is the winning factor.
    Alien can counter this with suicide bile attacks on the base for sure.
    But is this real skill? not really.

    So we need to split the carry players. (Thats what the Elo vote do)
    This mechanism alone, where every team has 2-3 carry players, is the basic balance we need.
    All the other factors would be only cosmetic.
    And the "perfect" formular would not end in perfect balanced teams cause the human factor i mentioned above.

    The KDR is needed to split the carry players, ok. But in my opininion the Kill death assist should be removed from scoreboard and implementing some scoreboard like in TF2.

    So you see only the scores on the scoreboard (wich would promote the support of an gorge or the resbiter also) but internal the KDR is used for balancing.
  • Ghosthree3Ghosthree3 Join Date: 2010-02-13 Member: 70557Members, Reinforced - Supporter
    dePARA wrote: »
    biting res like an idiot

    wDFhRll.png
  • xen32xen32 Join Date: 2012-10-18 Member: 162676Members, Reinforced - Supporter
    - It use the average KD/R of the last 3 rounds for the other players to mix the teams.
    K/D doesn't mean all that much in pub NS2.
    I spend all my day making enemy commander's life a nightmare. I kill harvesters like I'm poisoned and there's antidote in them, I suicide bile ten times a round like if there are seventy two gorgettes waiting for me in heaven, I'm bombing upgrades like if it is a revenge for Pearl Harbor. Most of the times I end up having 1/30 kill/death ratio, but help my team to win on macro level and leave enemy far behind. None of the currently existing systems (hive, ns2stats) appreciate that. Only people you play with regularly do.
  • GrimfangGrimfang Join Date: 2003-02-04 Member: 13086Members, Constellation, Reinforced - Shadow
    The soloplayer against structures is only good, if the rest of the team can keep the enemy occupied. I often take the role as well, and while it can distract the enemy team, and help on some level. If you are good at killing enemy players you probably contribute more in the long run, since you deny enemy players map control. But ofcourse everything helps, so contribute where you can :)
  • RapGodRapGod Not entirely sure... Join Date: 2013-11-12 Member: 189322Members
    edited November 2013
    This is... idk how to put it. People complain about pubstomping (which DOES exist), but want the ratings to be based on points? or commanding? do you really think your 50 point gorge is going to transfer over to marines and decimate aliens? or is just plain horrible? The same goes for commanding. No one gets pubstomped by gorges or commanders (on either side). If people are complaining in-game about a pub stomp it's all in their K:D.

    "Oh look he's 14-1 and no one has more than 5 kills on our team... he's pubstomping."

    The point system is entirely unimportant imo (other than making players "feel" important) and has little to NO connection to pub stomping. At least try to implement a K:D balance. See where that goes. I bet 1 penny (worth less than it is to make it) that that system will either stay or need minor tweaking.

    Just looked at your last post depara... i guess i just stole from u lol but I also agree the whole "assists" concept is just to make the player feel like they contributed.. even though they know they hit the thing. Just because you didn't finish the kill doesn't mean you didn't help. If you really need such a petty way to tell you "Hey, you helped!" then... i think a less complicated FPS would be better suited for you; that way you know it was YOU who killed the enemy. Last hate on assists: 95% of the time it screws with my head. I'll be shooting a skulk or fade or anything and all of a sudden a +27346527852 pops up and I think I killed the lifeform.. but he's still there (goes for either sides).
  • Ghosthree3Ghosthree3 Join Date: 2010-02-13 Member: 70557Members, Reinforced - Supporter
    People have to accept that you CANNOT mix up the teams by ANY formula for a balanced game (MORE balanced maybe). The only way to get games that are more balanced is proper match making, which I doubt will happen because we'd need at least 5x the player base we have now.
  • MestaritonttuMestaritonttu Join Date: 2004-07-29 Member: 30229Members, Reinforced - Shadow, WC 2013 - Gold
    Be honest, you don't care about the balance, you care about those 2 guys with the insane kdr being on the same team. I'd wager this is the case in 83,9% of stack complaints.

    Anyway, I guess the important question is why isn't ns2stats and shine on more servers?
  • Ghosthree3Ghosthree3 Join Date: 2010-02-13 Member: 70557Members, Reinforced - Supporter
    Actually I don't care about the two guys at all :P They can be good, means I have to be better. Regardless, matchmaking keeps them out of the pleb-tier games.
  • UncleCrunchUncleCrunch Mayonnaise land Join Date: 2005-02-16 Member: 41365Members, Reinforced - Onos
    Have fun pressing that in one formula (which gives you a single number in the end, too, btw, because otherwise theres no other way to compare).

    Not a single number. 3 bars. So you can profile a little more players. Distribution based on profile isn't that complex...
    Have la little gorge, have a little com, have a little pistolero.

  • cooliticcoolitic Right behind you Join Date: 2013-04-02 Member: 184609Members
    Well, I guess it sort of works then but doesn't seem to accurate (then again, no game has a very accurate way of showing skill level).
  • WalsaWalsa Join Date: 2011-06-28 Member: 106813Members, Reinforced - Shadow
    dePARA wrote: »
    I dont think we need a super complex formular to "balance" teams a bit.

    .....

    The KDR is needed to split the carry players, ok. But in my opininion the Kill death assist should be removed from scoreboard and implementing some scoreboard like in TF2.

    So you see only the scores on the scoreboard (wich would promote the support of an gorge or the resbiter also) but internal the KDR is used for balancing.

    I completly agree with this. You should only see your own K/A/D in your own scoreboard.
  • IronHorseIronHorse Developer, QA Manager, Technical Support & contributor Join Date: 2010-05-08 Member: 71669Members, Super Administrators, Forum Admins, Forum Moderators, NS2 Developer, NS2 Playtester, Squad Five Blue, Subnautica Playtester, Subnautica PT Lead, Pistachionauts
    RapGod wrote: »
    No one gets pubstomped by gorges or commanders (on either side).
    Pssh... then you haven't seen @currently gorge - the man is insane with a gorge and while he definitely does not pubstomp, he easily could.


  • OnosFactoryOnosFactory New Zealand Join Date: 2008-07-16 Member: 64637Members
    "Whats ELO?" - Some dudes surname, made a rating system for Chess leagues.


    PS:

    RapGod wrote: »
    "No one gets pubstomped by gorges or commanders (on either side)."

    Pssh... then you haven't seen @currently gorge - the man is insane with a gorge and while he definitely does not pubstomp, he easily could.

    ^^^Gorge is borderline OP as a support class.

  • xen32xen32 Join Date: 2012-10-18 Member: 162676Members, Reinforced - Supporter
    edited November 2013
    Grimfang wrote: »
    The soloplayer against structures is only good, if the rest of the team can keep the enemy occupied. I often take the role as well, and while it can distract the enemy team, and help on some level. If you are good at killing enemy players you probably contribute more in the long run, since you deny enemy players map control. But ofcourse everything helps, so contribute where you can :)

    I unno, we had a game of all good players 10v10, formed teams so one had best shooters in our community (those kind of people you'd think are cheaters if see for first time) vs team of people who focus more on macro level of the game.
    And we (2nd team) won matches on both sides.
    Dem shooters consume tons of meds and ammo instead of building, all go lerk and fade, never go gorge or buy GL, so they end up with no means to kill enemy structures, they will never sacrifice themselves for greater good, and killing certain structures is the only way to win the game. Sure killing enemies would help achieve that. But your team can win without a single kill and vice versa - you can have full team of 100:0 and game won't end.

    Having a few shooters on your team is a great advantage of course, but you can't overdose KD, as it will get you nowhere.
  • driestdriest Germany Join Date: 2013-02-21 Member: 183251Members, NS2 Playtester, Reinforced - Shadow, WC 2013 - Supporter
    While it is true that kills are not everything, you can't deny that winning engagements is very closely couppled to winning the game if nothing is completely wrong with the strategic/support side of the game. If you have 2 people that can take out the whole enemy team by themselves, they can just walk into your base, kill all defenders and kill all the eggs / ips. If you watch competetive games you'll notice that one of the biggest reasonst for a team to loose is loosing too many engagements.

    Of course not everyone percieved as a 'pubstomper' is nearly that good, so this luckily doesn't happen too often. Many people who get insane k/d just don't care about the team/win and just run around picking off people running around alone. Don't do that and pursue a team goal and your safe and still have a good chance to win. If they don't get enough com support the available ammo also limits the damage they can do.
  • PaLaGiPaLaGi Join Date: 2008-01-03 Member: 63331Members, Constellation
    xen32 wrote: »
    Grimfang wrote: »
    The soloplayer against structures is only good, if the rest of the team can keep the enemy occupied. I often take the role as well, and while it can distract the enemy team, and help on some level. If you are good at killing enemy players you probably contribute more in the long run, since you deny enemy players map control. But ofcourse everything helps, so contribute where you can :)

    I unno, we had a game of all good players 10v10, formed teams so one had best shooters in our community (those kind of people you'd think are cheaters if see for first time) vs team of people who focus more on macro level of the game.
    And we (2nd team) won matches on both sides.
    Dem shooters consume tons of meds and ammo instead of building, all go lerk and fade, never go gorge or buy GL, so they end up with no means to kill enemy structures, they will never sacrifice themselves for greater good, and killing certain structures is the only way to win the game. Sure killing enemies would help achieve that. But your team can win without a single kill and vice versa - you can have full team of 100:0 and game won't end.

    Having a few shooters on your team is a great advantage of course, but you can't overdose KD, as it will get you nowhere.

    Ya seriously screw that guy that just solo'd 4 skulks in mezzanine so you could solo cap 2 rts unharrased. That guy is overrated. You are the real hero building those RTs because "he won't build". Your ELO should basically be the same.
  • Ghosthree3Ghosthree3 Join Date: 2010-02-13 Member: 70557Members, Reinforced - Supporter
    PaLaGi wrote: »
    xen32 wrote: »
    Grimfang wrote: »
    The soloplayer against structures is only good, if the rest of the team can keep the enemy occupied. I often take the role as well, and while it can distract the enemy team, and help on some level. If you are good at killing enemy players you probably contribute more in the long run, since you deny enemy players map control. But ofcourse everything helps, so contribute where you can :)

    I unno, we had a game of all good players 10v10, formed teams so one had best shooters in our community (those kind of people you'd think are cheaters if see for first time) vs team of people who focus more on macro level of the game.
    And we (2nd team) won matches on both sides.
    Dem shooters consume tons of meds and ammo instead of building, all go lerk and fade, never go gorge or buy GL, so they end up with no means to kill enemy structures, they will never sacrifice themselves for greater good, and killing certain structures is the only way to win the game. Sure killing enemies would help achieve that. But your team can win without a single kill and vice versa - you can have full team of 100:0 and game won't end.

    Having a few shooters on your team is a great advantage of course, but you can't overdose KD, as it will get you nowhere.

    Ya seriously screw that guy that just solo'd 4 skulks in mezzanine so you could solo cap 2 rts unharrased. That guy is overrated. You are the real hero building those RTs because "he won't build". Your ELO should basically be the same.

    Lol, mind you, I do find that even if I do go ahead and kill a bunch of skulks (4 is a bit too many for me right now but 3 can happen). Somehow even though I pushed and did it alone all my capper teams die behind me when they were in groups of 4 and we have no res... So then I have to go cap :(
  • dePARAdePARA Join Date: 2011-04-29 Member: 96321Members, Squad Five Blue
    @xen32

    Lets arrange a 10v10.
    Your "Builder- an Structurekillerteam" against my "Shooterteam".
    Then lets see who is going to win all the games.
  • xen32xen32 Join Date: 2012-10-18 Member: 162676Members, Reinforced - Supporter
    dePARA wrote: »
    @xen32

    Lets arrange a 10v10.
    Your "Builder- an Structurekillerteam" against my "Shooterteam".
    Then lets see who is going to win all the games.

    It just happened once that we could split like that, I don't have actual team :(

    Tho I've been playing on a server with you last week (HBZ de Para, right?), if I recall correctly, you weren't focused on maintaining your K/D, you've been doing some other stuff and ratio wasn't 'Y U NO DIE'.
    I'm talking about Run'N'Gun'N'EARN FIRST PLACE IN STATS type, because 'skill' counter only counts kills.
  • dePARAdePARA Join Date: 2011-04-29 Member: 96321Members, Squad Five Blue
    The real "sharpshooters" doing stuff also:
    They killing 3 skulks, building some stuff, killing another 3 skulks, saving 2 rts, killing 2 skulks,.....

    Or you have a group of 2-3 "sharpshooter" marines going straight for the startharvesters, upgrades-/eggs.
    These players arent ramboing, they are doing an very important job called "Area denial".
    And these 3 can become unstoppable in combination with an half decent com against average pub players.

    Losing nearly every engagement no matter what side=losing the game.
    If 4 skulks cant kill 2 marines again and again these aliens going to lose.

    And thats why you have to split the sharphooters.
    Im talking about sharpshooter with gamesense here. These are the players with an high Elo score typically.

    These "Rambos" with gamesense carrying a team without them you cant expand as com.
    To much of them in one team=stomp.

  • GhoulofGSG9GhoulofGSG9 Join Date: 2013-03-31 Member: 184566Members, Super Administrators, Forum Admins, Forum Moderators, NS2 Developer, NS2 Playtester, Squad Five Blue, Squad Five Silver, Reinforced - Supporter, WC 2013 - Supporter, Pistachionauts
    edited November 2013
    Ah just wanted to add shine doesn't shuffle all player by elo because it happened in the past that by this way teams stayed the same for like 3 rounds. It shuffles the rest by another method (i use scorebased at my server) to avoid that teams stay the same.

    @depara So maybe change that as it has not to be KDR based. In fact its also KDAR based as assist count as half kill!!
Sign In or Register to comment.