ELO stats



  • @Administrator could you please FINALLY explain the logic / formula behind ELO calculation PLEASE!

    Again I am not that interested in ELO itself as in matchups which are affected by ELO and which gain/loss in games are still so random for no apparent reason.

    This topic has been raised MULTIPLE times so I find this highly offensive that no reply has been given despite majority of forum readers would be interested in reply for one reason or another!
    Adding to that this should not be a hard question given you have developed the game and definitely not as time consuming to solve vs. other topics in forums like quitter problem or dice roll.

    This also might affect even a game strategy as if I am last but still can minimize ELO damage i would do that rather give up or in the worst case quit as I will not be sure if I get -5 or even -10 ELO...



  • ELO gain/loss is based on what the ELO is for the players before the game. If you have low ELO and you win, you gain more ELO than another player would. If you have high ELO and win, you get less ELO because you were expected to win anyway. If you have high ELO and lose, you lose more ELO. See ELO in Wikipedia. Showing the precise calculation in the end of the game is a bit overkill.

    Developers might have done some weird stuff with the ELO calculation since ELO is meant for 2 player games usually. I guess they call it ELO but are not actually following it as intended. The admin is not very transparent with stuff like this. If they have decided to do something their way and are not willing to change, they will not answer you.

    "I don't care about ELO" ... "I would quit if I know that I can not minimize ELO damage." Which one is it? Also quitting is the main problem with this game. You are saying that you would quit games if they don't go your way... Can you please not be a sore loser and play until the end. It's not fun to win if someone quits right before the game is actually over. Even if the system does not penalize you for that (yet).



  • @Stroom Thanks actually didnt know that ELO is such a popular thing that it actually is available in Wikipedia.

    Yes the basics you mention are clear to me - can make conclusions out of multiple games i have done, but still precise info would be nice to know. Especially if this is well known formula...

    Regarding my comment - yes that second part was intended to be more general. But still given at the moment matchmaker + my banlist (which contains ~80 people already) ensures quite good matchups I dont really care about ELO.

    At the same time if quitter issue is not solved + I fall back into ELO zone where I start to get games vs nubs I might go for quitting just to enjoy better matchups really.

    I see no point in suffering in the system which is faulty/does not work and dont want to spend time on working my way back if majority of the games which I win ELO gain is not comparable vs ELO lost in random games where usually Dice rolls like crazy.

    For example yesterday had a game where I lost by 8 points with all resources and numbers (4/5/6/8/9) while person who won placed his settlement AND city on 5/6 only which of course given the result was top 3 roll after 7... ELO lose -10, past 3 games I won total ELO gain was about 12 if i am not mistaken... its like 3:1 ration time wise and you get tripple the negative effect sometimes just because the dice.

    In my opinion suggested formula of all players chipping in for example 5 ELO in the beginning of the game and winner takes 2/3, 2nd place 1/3, loser nothing or in 3 player game and in 4 player game 1/2, 1/3, 1/6 and nothing would make more sense.

    Especially given if above said formula is true as we can't choose who we play vs - maybe I would like to play only vs people with hither ELO given that would be more challenging, in case of a win I would gain more/progress faster and in case of a loss well tough luck but not dramatic lose of ELO to make me play vs nubs.

    @Administrator would be still nice to get confirmation from you what it is and is the info available in Wiki or somewhere else the one used in the game.



  • I said in the beginning (2-3 months ago) that ELO is not a useful statistic in Catan. If you want a ranking, use something else and do not call it ELO at all. ELO is meant for 2-player games with no luck and complete information. Catan is a luck-bluff-negotiation game with partial hidden information. People will get obsessed with high ELO so they are not playing the game for fun anymore, rather to grow their ELO. Maybe creating multiple accounts to boost their ELO or abuse the system in any way they can. Just having (WORKING) player reliability statistic should be enough. In PlayCatan it showed how many of the last 100 games you quit and finished. No "karma stars" needed. A very easy percentage is good enough.

    You will probably lose half or 1/3 of the games even if you are a good player. This makes ELO really fluctuate over the course of games. Even if you are the best player in the world the ELO score in Catan can not show the real ranking if it can be +-200 for each player depending on the week.



  • Well as for my ELO i could tell for sure that +/- 200 is not the case.

    Nowadays when I have good games which mostly are played till the end I actually win more than a half (sometimes I make mistakes with placement / some games are just determined by the dice - rolling odd numbers).

    My ELO pretty much stays at the level of 1050-1070 for past few month which is a bit unfair given above said.

    Anyways this ELO thing needs some explanation from Admins and overall system fix from there, current combo unfortunately is not favorable unless they want to have a community where idiocracy rules the rankings.



  • @Stroom , that helps. thanks



  • @Stroom For once i completely agree with Stroom. It is not a game of chess or analogs. It's a game with dice.



  • To add to the topic actually today had one of my fastest wins (=easy game for me) still other players did get 7 and 6 points while I got 13.

    BUT I gained +9 ELO, while I did post a some time ago a game where I won while one of the players actually lost a point and had to deconstruct finishing on 2 points and other player got 3 or maybe 4 points = pretty much flawless victory. And then I got just +7 ELO...
    As on the negative side - yesterday I lost a game by being last and finished on 8 points in 15 point game => got -10 ELO...that game was not even close as bad as this was easy.

    Again getting back to my point that this ELO situation is way unpredictable while so many features depend on it - match-maker / ranking / some of the players are motivated by it / maybe in future it will have something to do with tournaments...



  • ELO is designed in a way that it is easier to "fall down" than climb up. Whining does not work, you have to improve your skill. The top chess players have ELO of 2800 and they probably get something like 0.5 points if they win. If they lose, ELO might drop by 10 or more. It is natural... IF it was a game with 2 players.

    ELO does not count your points, only the fact that you won or lost. All players getting 9 points means nothing. Even if it is a close game... A counter example: A player has the largest army and longest road, is at 9 points. In one turn, a player plays 4th knight, taking the largest army AND steals his road, wins the game. Now that player has 5 points. Does that mean that he was so bad? ELO does not care about what happens in the game. Honestly, it is really hard to come up with any kind of heuristic to determine how close the game was.

    Now, in Catan, people will lose about the same amount of times than they win. ELO is not suitable for that. Normal ELO for a player should be around 1200. In Catan it is somewhere around 1000. Someone who makes the decisions in USM does not really seem to know anything about maths.

    Ranking people in games like Catan or Poker is impossible. Maybe only count/display how well the player has performed in tournaments. only low ELO can show that a player is bad or a quitter. High ELO could mean that the player only plays with less skilled players and always wins, abuses the system or something like that. A skilled player who plays with people of the same skill level will stay at the same default ELO with the current system.



  • That's what I am saying why bother and apply thing like ELO (which is as far as you say more suitable for chess) here in CATAN?

    Why we can't just make the game plane and simple - all chip in, winner takes majority.

    And then tighten auto-match so that players with ELO +/- 20? get matched up ensuring all are "at the same level".

    I really see no point in the current scheme as:

    ELO is unclear, misleading
    auto-matchup ends up matching people with various skills
    some players might not take a game vs experienced players well and quit due to heavy loose
    high skill players risk having a bad luck and falling back on ELO in a major way

    Like really who gains anything at the moment? Its more like lose lose situations.

    @Administrator could you please do something starting with explaining ELO calculation PLEASE (for the 10th time).



  • If everyone "chips in" and the points are later distributed... This will definitely be abused - players create temporary accounts, play with them, get points to their accounts and then you have people in leaderboards with thousands of points.

    ELO is actually quite clear. Just math. Only the fact that they are using it for 3/4 player games is a bad idea. There is more ELO lost than won in the games so eventually people will have low ELO. It's just that ELO is not suitable for games like Catan.

    Auto-match-up just can not work in Catan. You have about 20 players waiting for a game at any given time (if lucky). There is no way you can assure that the skill levels are within +-20 ELO. The more parameters you add to the matchmaker the slower it gets OR it will just revert back to matching anyone with anyone because if there are not enough players searching for games then matchmaker is useless. Also add to that the network of blocked users... It's really stupid that the developers haven't reverted back to the lobby system.

    @Administrator Still waiting for the statistics of how many games are created each day and how many unique players you have in Catan daily. And maybe how long a player spends on average online.

    Is it really that hard to use Wikipedia - it has the calculations right there:

    https://en.wikipedia.org/wiki/Elo_rating_system

    Probably something similar is created and now they have to tweak the k-factors etc.



  • I have noticed double accounts already now some even have tipple.
    Not much you can do there...besides adding some red flag notification from admin's side for players whos Experience / ELO (and possible Karma) are suspiciously high.

    For example player having only played 10 games and ELO of lets say 150 (15 ELO gain per game which is highly unlikely).

    As for match maker / lobby well...I can't imagine how would banlist integrate lobby:
    I would not see games where at least on of blocked players is in?
    I would get warning that people from banlist are in the game?
    What happens if for example 10 quitters are online and 3 good players - who gets to play? (I would actually suggest to @Administrator in case that is possible add some negative effect to reported quitters so they have to wait longer in match-maker or at least have disadvantage in cases where match-maker has to decided whether to give a game to 3 "clean" players vs. leaving out a person who has quitters in his/her banlist).



  • In PlayCatan lobby: You can not join a game if you have been blocked by a player who is in the game. Currently you just get a warning. It would be good to automatically filter by that criteria too - hide all games which you can not join for any reason. It's a very simple check when a player wants to join a game. Much harder in a matchmaker because it has to compare all the players to each other in order to find a game. In a lobby, it is done by the players themselves.



  • Yeah exactly my point and here I guess matchmaker wins because community is still small = not many active players and if you have to filter them out (majority) due to banlist you might just get stuck in filtering...



  • Umm... The problem you describe is more related to matchmaker rather than the lobby. Lobby allows you MORE control over what games you join and what games you see. Matchmaker hides everything from you so you don't really know what the hell is happening.

    The filtering would be automatic or just 1-2 buttons to select what you want to filter out. So on your end you would only see the games which you can join. Would you rather have the matchmaker still put you in a game with people who you have blocked? With matchmaker the "stuck in filtering" would happen much more often than in lobby. I don't see how a lobby would be worse than matchmaker for online Catan.

    If you filter out a "majority" of players because your ban list is so huge then the problem is with you not other players. Why do you think the matchmaker does not have the same problem if you have a huge ban list? It is essentially doing the same filtering for you in but in case of lobby you have more control over what game you start. You can see all the pending games with their options. In case of matchmaker you have to precisely set what you want and the matchmaker will just wait until there are enough players who want to play with these options. If no-one knows that you would want to use some specific combination of options, you basically can not start the game with a matchmaker. With a lobby other players will see your options and they will more likely join much sooner.



  • 1st of all match-maker since the bug has been fixed has EVER hooked me up with players whom I have in the list.

    2nd yes it does the filtering for me and from what I see happening it might just given advantage to the fact of fixing the game over the fact to keep quitters / reported players on halt which is the thing to be fixed. Same as when you choose 3 OR 4 player games and game starts 3 player game as soon as 3 people have been found rather than waiting a bit for possible 4 player game.

    Works fine for me as long as ELO factor is explained and possibly fixed/fine-tuned.


Log in to reply