User rating/ranking (ELO system)

Comments: 88

• 12 Oct, '19
  Tiffany

  Could be fun. Similar to the Catan app. Although, it may encourage some users to try to "game" or cheat the system by creating fake accounts to play against.

• 14 Oct, '19
  Esqarrouth Admin

  Any ideas how the cheating could be prevented?

• 16 Oct, '19
  Jesse

  Make it mandatory that all players in a game need to be logged in for it to count as a rated game. It will not prevent the issue completely but that is the best that can be done as long as the site is free of cost.

• 29 Oct, '19
  Aubrie

  Maybe come up with ways to detect accounts that are throwing games (one example could be if they do multiple trades of 3+ cards for 1)
  
  And then deduct 1000 elo points from them, so future games won’t count.
  If they complain, review some of their games and see if it was shady. It should be pretty obvious, eg if they play 1 way against the public, but completely different against 1 opponent, against whom they play many times.

• 04 Nov, '19
  Matttyyy

  Adds some sort of competition and motivation to keep playing and winning. Makes it fun even when you play against some lackluster competition.

• 14 Jan, '20
  Jay

  require login for all users

• 14 Jan, '20
  player Merged

  At some point you might reach a level where it's hard to find enough similarly experienced players online without playing tournaments, I'm not sure this is a bad thing.
  
  Perhaps each level requires even more games to get to the next level (ie, maybe level 1 requires 1 game to get to level 2. But Level 10 requires 10 wins to get to the next level.
  
  This feature also implies an option to filter games to similarly ranked players, or to play an unranked game (just fun, right!?)

• 14 Jan, '20
  player Merged

  Oh. And perhaps there is a ceiling for unpaid players. IE... get to rank 5 only if you haven't subscribed.

• 14 Jan, '20
  Connor Merged

  I am not one who would agree with putting a cap on people who dont pay money. But possibly you meant to just have a verified account with email, in which i'd agree and say that would be good.

• 15 Jan, '20
  player

  I'd prefer a ranking system, just to know the players skill, as opposed to a rating system. I just posted a description of that.
  
  The ELO system sucks, and is subject to abuse. If someone thinks they won't win, they start abusing the lowest rank player to ensure they get 2nd place, just to keep their ELO up. It tends to promote abusive/anti-social play.
  
  Also, it's kind of meaningless when a random lucky win by a terrible or low ranked player drops your ELO like a rock.

• 15 Jan, '20
  player

  (High level, the ranking system I suggest (as a start) is you have to win X games against players of similar or higher rank to be bumped up a level. Start at level 1, win some games against level 1+ players, get bumped to level 2. Then you have to win a number of games against level 2+ players to get bumped to level 3.
  
  I suggest making it win X of games for your level. So Level 1 requires 1 win against Level 1+ players. Level 10 requires 10 wins against level 10+ players.)

• 16 Jan, '20
  player Merged

  Conner, what is your opinion of the approach to ranking? Or tweaks you think would be productive.

• 28 Jan, '20
  DA

  Some sort of monitoring system for players that seem to win excessively? Are they always playing with the same players, obvious game throwing trades, fast win times? Of course the ban hammer should drop if accused players are caught.
  
  Hope you guys can come up with something because it would make the game that much more enticing! Keep up the good work!

• 16 Mar, '20
  Gemma

  was working, now isn't??

• 21 Mar, '20
  PotatoRiot

  I really like players proposal! This would prevent abuse starting from lever 2 (well one could farm accounts to get to a higher level with the main account but well at least the top 10 players one would see in their history if they only played with the same accounts over and over again...)

• 15 Apr, '20
  LTAlex

  Also to prevent cheating you can create a calibrating system for new accounts and get fewer points for playing against lower-rated players.

• 24 Apr, '20
  OnTheBrinkMN

  Way back when, when I played Catan on Sea3D (S3Dconnector), they had a pretty good ratingsystem, with a possibility to report abuses and stuff. Players with same IP-adresses and using fake accounts to boost their rating would be blocked (for a certain period) eventually if someone reported it. Worked fine then (more than 10 years ago), so can't see why it shouldn't work now

• 27 Apr, '20
  player Merged

  No, ELO sucks as a way of calculating rankings and it wasn't designed for this style of game (it was designed for chess where there was no element of randomness, two players, and no bots to abuse/game). I do agree some kind of ranking system would be useful and have proposed a loose idea for one, basically that you can only increase in rank, and that is done by winning a certain number of games against equal or higher ranked opponents

• 27 Apr, '20
  player Merged

  Some of the biggest flaws of ELO is it gives way too big a ELO boost (or ELO hit) for a lucky win by an unskilled opponent, or encourages people to play for 2nd by abusing low position players, or encourages people to drive a player to quit and then game the bot to win (or let the bot win) to gain ELO.

• 11 May, '20
  Chris Merged

  I think this would be a fantastic addition.
  
  First, ELO would have to be coded into the game, and some thought placed into that - would you see opponents' ELO before the game starts? Would you be able to create a table with minimum and maximum ELO, etc. etc.
  BoardGameArena would be good to compare against, they have it pretty well down pat if you ask me.
  
  Currently a player's profile shows Wins, Games, Points, Win %, Pts/game, and Karma. [CONTINUED]

• 11 May, '20
  Chris Merged

  With the addition of ELO, allowing sorting by different columns, perhaps a new "season" that starts every 3 or 4 months, and separate leaderboards for "vs humans only" and "All games", it would be all set.

• 12 May, '20
  player Merged

  In the long run ELO works. Gotta learn how to math.

• 13 May, '20
  Sasha Wilde Admin

  "Rank [ELO] system with ranked gameplay" (suggested by LukeFish on 2020-04-25), including upvotes (3) and comments (3), was merged into this suggestion.

• 13 May, '20
  Sasha Wilde Admin

  "Add a Leader board on the home page" (suggested by Kenpachi on 2020-04-12), including upvotes (4) and comments (2), was merged into this suggestion.

• 13 May, '20
  Sasha Wilde Admin

  "Enhancement: Player ranking based on games won against similarly ranked players" (suggested by player on 2020-01-14), including upvotes (2) and comments (4), was merged into this suggestion.

• 13 May, '20
  Sasha Wilde Admin

  "Ranking" (suggested by Natasha on 2020-02-28), including upvotes (1) and comments (0), was merged into this suggestion.

• 16 May, '20
  Elma Houghton Merged

  to rate somebody during the game,, ie if they drop out, are horrid, or do not play well.. i know its dubious.. but maybe as a personal rating as well as the rating we have for how many games we play.

• 16 May, '20
  Valeria Merged

  Like this idea... o u can select o kick someone based on that. Is very anoying when people retired bcs they dont like the numbers o for the karma. If u could rate them when game is finish even if they leave before..

• 16 May, '20
  player Merged

  I understand the intent, but surely sure this system would be equally misused. I've played more than enough games where people are clearly playing with their friends and as such just target the 3rd or 4th player. How do you think they'll rank the 4th person who complains or quits due to overly aggressive and bullying players!?

• 22 May, '20
  Zak

  I think the rank/rating system could work by making a rank queue that you have to queue into in order to prevent people from botting and winning their games. If there isn’t a queue for ranked then there is no way to actually play with people that are your rank/skill, nor is there a way to not prevent cheating

• 27 May, '20
  Raj Merged

  Similar to chess, you could implement an Elo system where players can increase / decrease their Elo as they play more games. It is a good way to track your improvement and progress and would incentivize more people to play alone vs. with their friends.

• 29 May, '20
  player Merged

  ELO is not suitable for Catan. I agree there should be a ranked system where if you win X games against similarly (or higher) ranked players your rank goes up, but having it drop drastically because of bad dice or poor play by other unskilled players (who screw up a game) isn't reasonable. There's to much luck and variables for ELO to be accurate (unlike chess which is 1-on-1 and entirely skill based)

• 29 May, '20
  player Merged

  And honestly I don't think your rank should ever drop, you don't lose experience or skill from losing a game. You just aren't worthy of a higher ranking (yet)

• 31 May, '20
  Puddnhead

  ELO would be great. The risk of people gaming the system to improve their ELO doesn't matter much to me. The main point is to protect the low-rated players who are new to the game from having to be paired with experts. It's no fun to try out a new game and then lose every game.

• 26 Jun, '20
  Donna Merged

  no comment

• 01 Jul, '20
  NFOGhost Merged

  I was looking for the same. I would like to know who is strong so I can either play against them or spectate them.

• 02 Jul, '20
  Jared

  Since there already is a huge tournament presence through Discord and Instagram for Colonist and other Catan apps, creating a ranked system and tournaments through Colonist would put it ahead of many other competitors such as Catan Universe and Catan VR

• 24 Jul, '20
  Troy

  I like this idea, but also like the idea of just general personal stats. Win pct, average place, win streak, etc.

• 24 Jul, '20
  Troy

  I like the idea of a rating system, but it sounds a little complex/could be controversial, probably not high priority. At the least a leaderboard would be really cool. Giving the user the option to sort by wins, points, points per game, win pct, games played, games finished, finished pct, etc. Being able to filter on daily, weekly, monthly.

• 24 Jul, '20
  Troy Merged

  LOVE the idea of a leaderboard. Would be really cool. Giving the user the option to sort by wins, points, points per game, win pct, games played, games finished, finished pct, etc. Being able to filter on daily, weekly, monthly stats.

• 24 Jul, '20
  Troy Merged

  Also, you should be able to filter on game rules. If I am looking at a win pct leaderboard, I am only curious about win percentages of games with the same number of players. Player A with a win pct of 45% that only plays 2 person games is not comparable to Player B with win pct of 35% that only plays 4 player games. Same goes for points per game.

• 30 Jul, '20
  Austin Admin

  "Create a ranked ladder where you can climb up the ranks by playing similarly ranked players" (suggested by Raj on 2020-05-27), including upvotes (6) and comments (2), was merged into this suggestion.

• 15 Aug, '20
  JUAN Admin

  "Leaderboard of wins and points" (suggested by Matt on 2020-05-21), including upvotes (4) and comments (3), was merged into this suggestion.

• 15 Aug, '20
  JUAN Admin

  "Earn points for wins. Lose points to quit a game" (suggested by Donna on 2020-06-26), including upvotes (1) and comments (1), was merged into this suggestion.

• 18 Aug, '20
  JUAN Admin

  "give someone a rating based on their play" (suggested by Elma Houghton on 2020-05-16), including upvotes (4) and comments (2), was merged into this suggestion.

• 26 Aug, '20
  Shikamaru

  It would be nice to include a "power score" statistic in each player's profile. Win rate does not tell whole story because the "odds" for players of equivalent skill winning a 1v1 game (50%) > odds of winning a 1v4 game (25%). Power score = Overall winrate divided by "overall expected winrate." Expected winrate = (#2 person games)(1/2) + (#3 person games)(1/3) + (#4 person games)(1/4).... I'd also suggest removing games where 50% or more of the players are bots from the calculation.

• 26 Aug, '20
  Shikamaru Merged

  A player's win rate is not a great measure of that player's skill. That is because your odds of winning a 2 person game are higher than your odds of winning a 4 person game. Therefore two players of equivalent skill are likely to have much different win rates if one player tends to play 3 person games and the other tends to play 4 person games. A"power score" can normalize a player's "actual win rate" by that player's "expected win rate." Expected win rate = (# of 2 person games played)(1/2)

• 26 Aug, '20
  Shikamaru Merged

  As a follow on: there are many ways to translate the above calculation into a user-friendly number. One way: the "power score" could be indexed to 100 for ease of interpretation. For example, assume a player exclusively plays 4-person games, and that player has a 25% winrate. That player's power score could be 100 (i.e. 25%/25% x 100). Under this system, if a player plays exclusively 4-person games and has a 40% winrate, then that player would have a power score of 160 (i.e. 40%/25% x 100).

• 26 Aug, '20
  john Merged

  This seems rather confusing, and tells you nothing about the skill of their opponents. I'd prefer a ranking system, where someone's rank (starting at zero) only increases [by 1] after X wins against similar or higher ranked players. If you beat someone of a lower rank, that's meaningless as you should have won. [Rank should never decrease, as losing a game doesn't decrease your skill]. This would allow filtering by skill. [Some refinement needed to prevent friend circles from gaming it]

• 26 Aug, '20
  john Merged

  At the very least, 2 player vs 3 player vs 4 player are different games, so I don't see how you can lump them together under a single rating.

• 26 Aug, '20
  Shikamaru Merged

  I agree that a ranking system would be best. However, I'm suggesting a simple addition to a player's profile that normalizes their winrate to account for the fact that it is easier to win a 2-person game than a 4-person game. A ranking system like the one you suggest is a major development effort. The "power score" I suggest is a simple metric that can be added to a player's profile. It is a much easier "ask" from developers and would serve as a useful way of assessing player skill for now

• 26 Aug, '20
  JUAN Admin

  "Add "power score" to each player's profile" (suggested by Shikamaru on 2020-08-26), including upvotes (5) and comments (4), was merged into this suggestion.

• 28 Aug, '20
  JUAN Admin

  "Super league" (suggested by Halit on 2020-04-18), including upvotes (1) and comments (0), was merged into this suggestion.

• 09 Sep, '20
  Anton

  For ELO system: Different rankings for 4-player games, 3-player games, and 2-player games.
  
  The goal here is that people want to be matched with people of similar skill.
  
  Alternative to ELO: You could ask players if they want to queue in “normal” or “experienced”. Slap on some requirements such as, at least X amount of wins. You could also do a survey at the end of the game “rate this game :( , :/ , :)” and look at difference in player stats to see which variables are off in :( games

• 13 Sep, '20
  tmblwd

  Definite support for ELO (maximum change per game could easily be reduced for Catan to limit the random swings element).
  I also play AOE2 that uses the first 10 games you play to put you at roughly the correct rank. For these matches larger changes are allowed.
  
  Ranked queues are great! Try to match with those at same level and increase window depending on how many people in the queue.
  
  I don't care for leader tables, but for me the ranking and queing are very important developments.

• 29 Sep, '20
  JUAN Admin

  "Implement an Elo System for Ranked Play" (suggested by Joshua Migneco on 2020-09-29), including upvotes (1) and comments (0), was merged into this suggestion.

• 06 Oct, '20
  Ollie

  Definitely needs an ELO system. Would play and spend much more money.

• 16 Oct, '20
  Konidias

  Simple way to curb cheating would be to charge a small one time fee to unlocked ranked matches. ($5 or something). This would let you play against other ranked players and the $5 charge would reduce the amount of people trolling/cheating as getting caught would result in them losing money. This has always been a decent deterrent in online games.

• 17 Oct, '20
  Stroom

  Glicko2 seems like the best option.
  Usable out of the box, it seems. Have to add some database on your side.
  https://www.npmjs.com/package/glicko2

• 16 Nov, '20
  Rafaeldomar

  The only thing that makes my friends not play only on Colonist is the fact that there is no elo system. We love everything else about the site, but it's such a major thing that's missing.
  
  Also, what if the karma system wasn't be capped out at 20? If it went to infinity and beyond it'd be a cool little additional way to incentivise people to play, (other than the much needed elo system).

• 17 Nov, '20
  The Giver

  1. Charge a decent amount to "unlock" ELO mode. This discourages people from wanting to play if they're spending money.
  2. Create a separate page for ELO mode, where instead of inviting players to a lobby and joining a lobby, there's a big queue button and you get put in a game automatically with other players based on your preferences. Do NOT allow players to invite friends or other players to the lobby.
  3. Increase your karma to 100, and set it so that you can match with others based on karma.

• 25 Nov, '20
  JUAN Admin

  "ELO" (suggested by shishamoo on 2020-11-25), including upvotes (1) and comments (0), was merged into this suggestion.

• 13 Dec, '20
  SillyWords

  ELO is fine. In a multi player game with n players you have to have (n-1)! factorial results e.g. players ABC and D play, so that generates a result set for in AvsB AvsC AvsD BvsC BvsD and CvsD.
  
  To prevent rating inflation via fake accounts, two factors have to be added in to the ELO system
  a) one, a factor which is in every implementatio based on the uncertainty of the rank because of the number of games. So for example a new player gets a starting rank of 1000, but a win over a new player is not worth as much as an established player with 20+ games who also has a rank of
  b) a factor indicating how many different opponents a player has played against. This prevents a player from generating three dummy accounts and just playing again and again against the same opponents.
  
  So in order to game the system, you have to create multiple accounts, rank them up by playing diverse people, then play them all again once the dummy accounts have achieved full ranking potential.

• 31 Dec, '20
  JUAN Admin

  "Add star system for good players ( friendly players )" (suggested by Gus p on 2020-12-23), including upvotes (1) and comments (0), was merged into this suggestion.

• 06 Jan
  Samuel Garcia

  1. Create a specific gamemode for ELO games. Basically ranked matches, where instead of inviting people to a lobby, you enter a queue to be matched with players with similar ELO to yours.
  
  2. Have a minimum numbers of games played and karma ratio to be able to play ranked matches.
  
  3. Make the highest rank be special one with limited people in it. Sort of like in League of Legends, where challenger rank only has 200 players and they have to fight every month to keep their spot. That makes it cooler to differentiate the "pros" from the rest.

• 09 Jan
  Alistair H

  I feel that an elo system is good, because in a ranking system, it is only for the people who play tons of games. I play colonist.io sometimes, but I don't play it all the time, so an elo system might be better, so people who don't play the game too, too much can still be ranked high.

• 12 Jan
  Cyril

  it is a must go. it is very limiting not to have the level of opponents and no challenge to go

• 01 Feb
  JUAN Admin

  "Ladder system and matchmaking" (suggested by Oleg on 2021-01-31), including upvotes (1) and comments (0), was merged into this suggestion.

• 19 Mar
  Jonathan

  I dislike this idea. I already go into a game, and sometimes somebody will check my stats then write in the chat about my stats and everyone should block me. Then it is 3 against 1. Makes it not fun for me, and if I leave the game I lose karma. This elo rating, will make it worse for players like me. I quit playing on Catan Universe because of the way people on there team up so much.

• 31 Mar
  Sollylogger Merged

  How about adding a "Strength of Schedule" rating in addition to the %won in the player profile. This would give you a rating of how good the opponents are that the person is playing. To calculate, at the end of the game, take the win% of all of the opponents. Add these three numbers, call it Game-SS. Then for the last 100 games, total all the Game-SS numbers and divide by the number of games. This would give an average "Strength of Schedule" number. You would have to adjust for Bot games- I suggest excluding them. And also adjust 2,3,5,6 person games to normalize the data to be comparable to the 4 person game.

• 05 Apr
  JUAN Admin

  "elo" (suggested by BennyHK on 2021-04-02), including upvotes (1) and comments (0), was merged into this suggestion.

• 05 Apr
  JUAN Admin

  "rating players" (suggested by Sollylogger on 2021-03-31), including upvotes (1) and comments (0), was merged into this suggestion.

• 06 Apr
  Glimfeather

  A lot af colonist.io's succes is that it is available - there are always games in the lobby, and everyone plays with everyone. Beware that new players might feel the community isn't as accesible with a prominent elo system. A leaderboard sounds fine - or even a competitive track - but let's be very careful when changing fundamental parts of the colonist.io community. Hope the thoughts make some sense :)

• 02 May
  roger

  Please implement this! It gets tiresome to play with people with poor strategies. Some sort of ranking or tier system would do wonders.

• 29 May
  aoaoaoao

  @Glimfeather simple, treat all guest players as the starting ELO (assuming 1200 based). This is what other sites (ex: lichess.org) that use an ELO system does.

• 29 May
  aoaoaoao

  btw the only way an elo system would work is if you added a queuing system. Users joined rooms should not be ELO based games

• 08 Jun
  timi

  maybe you could also rate players by how often they get muted/blocked/reported, and how often they get friended (once that feature is added). then you could have a much more rigorous karma system and ideally you can filter based on karma so people who are jerks don't even get to come into your room. then you just need a way to make it a nice experience for new players who come in with no karma

• 15 Jun
  Jesse

  How come we do not have this yet? Win % is only taking into account the last 100 games. But it would be great if it counts all the games.
  
  ELO kind of system would be much better. Where the points also can be taken into account even if you lose. Obviously, the main reward is for those who win.

• 16 Jun
  ACCC

  Elo systems means everything for a competitive online game.

• 08 Jul
  Jesse

  Absolutely. I am waiting for this for a very long time. You could also give the players an option if they want to include all of their game history in ELO rating or start afresh. I definitely want my entire game history.
  
  Also, CHEATING is not really possible(or can be very much minimised). I have a really good idea on how this can be achieved.
  
  The default ELO for new accounts can be 0, rather than a minimum value. If all 0 rated players play, then the ELO of the winner should go up. After that, you can never increase your ELO by playing against all 0 rated players. Your ELO will only go up if you win against a non 0 player. And the ELO increase or decrease should be proportional to the ELO of the players in that game. This will motivate players to get a good ELO and keep playing the game. I know, it would motivate me to play more and get me more addicted probably. LOL!
  
  Also, if there are players who cheat with multiple accounts, then others can report and action can be taken, right?

• 11 Jul
  Susan

  Yes this seems like a nice idea. I think it will work. Can we have this sometime soon?

• 27 Jul
  Rubicund

  I did not spend time reading through 80+ comments in this thread so my idea might have already been said. Nonetheless, to deal with the boosting problem, make the option to play rank open only for players who have played 200+ fully completed non-ranked games (No bots matches, leaved games do not count, serial games consisting of the same groups of people stop counting after the first game of that day and would reset the following day--this would make any attempt to boost extremely strenuous).
  
  Instead of investing in an expensive authentication system, simply require accounts be made using mainstream email services, as the creation of such accounts require users to go through phone verification before users can access the service.
  
  In ranked games, I think queue (match making designated by a system that attempts to match the user with NEW players of similar elo as user), in other words, single-player match making, would work best to fight against boosting. Let me know what you think!

• 06 Aug
  Yoda

  This was suggested nearly two years ago and promised as a benefit to paying players. When will this feature arrive?

• 06 Aug
  JUAN Admin

  Currently being worked on.

• 24 Aug
  Yod

  Thanks, Juan. Would you please give an estimate on when the ranking system is expected to be in place?

• 24 Aug
  JUAN Admin

  We are working full time on this feature but it is really complex and has too many parts. There is a lot to design regarding UI/UX, as well as architecture design. Then coding, testing, fixing, getting feedback, and optimizing. That's why we can't give estimates about new features' release dates. Hopefully it will be ready before the end of the year, but can't promise anything.
  
  Become a playtester in Discord to help us release new updates and features without bugs! While testing them before anyone else ;)
  https://discord.gg/Nh7dKuEb8X

• 31 Aug
  Yod

  ELO should be lightweight on both UI and UX. This is algorithmic stuff that is mainly in the background and can tie into profile viewing/game sorting without a heavy lift on the design side.
  
  Releasing a light version to test with the most important feature(s) would get you there within weeks, not months or years such as it has been.
  
  Once you have the parameters of the algo, the coding and release of this particular thing is iterative and can be launched in pieces, which will give you better feedback than launching the whole thing at once anyway.
  
  Maybe the delays could be mitigated with a more agile approach.

• today
  Anonymous

  This ELO ranking system would be amazing and has been on the list for two years already. Why is it not done? (Same with the mobile app, but I think this is more important and fun).