Lets Beef - Battle Rap Forums - Elo Ranking System + Matchmaking

Lets Beef - Battle Rap Forums (https://www.letsbeef.com/forums/index.php)

- General Talk (https://www.letsbeef.com/forums/forumdisplay.php?f=80)

- - Elo Ranking System + Matchmaking (https://www.letsbeef.com/forums/showthread.php?t=160589)

Elo Ranking System + Matchmaking

This has come up before, and I'm sure a lot of you gamers and chess players already know what this Elo system is, but I think this would be a dope addition to LB as its one of the only ways to have an actual skill based ranking system

For those of you that don't know what the Elo ranking system is, it's a ranking system where you get placed in a tier, and the criteria for earning that tier is winning/losing.

The tiers are Bronze, Silver, Gold, Platinum, and Diamond for example.

Everyone starts with a blank rank, & each person does 10 qualifying battles that'll place them in a tier.

If you win 1 out of 10 battles you're placed in Bronze
If you win 3 out of 10 battles you're placed in Silver
If you win 5 out of 10 battles you're placed in Gold (since it's the middle)
If you win 7 out of 10 battles you're placed in Platinum
If you win 9 out of 10 battles you're placed in Diamond

And you can add tiers higher than Diamond. If you win 10/10 your rank is 'Elite' or something

So now you're thinking "What if a scrub battles nothing but scrubs, wins 10/10 battles and gets the highest rank?" Well that's where matchmaking comes into play. I was wondering how LB could incorporate this, then I realized we've done blind battles on here which is pretty much matchmaking done manually. Matchmaking makes it so your 10 opponents are random, and it'll match you based on your skill once everyone qualifies

So let's say everyone qualifies and the breakdown is this

LB has:

3 Bronze battler
5 Silver battlers
10 Gold battlers
6 Platinum battlers
3 Diamond battlers

If you're a Platinum battler and you use the matchmaking system, it'll only put you against other Platinum battlers. If there aren't any active platinum battlers it'll place you against Diamond or Gold battlers, but never Bronze or Silver since the skill gap is too wide. I'm sure you can code an if/then/else statement so that you're not matched against inactive battlers

ie; If ___username___ hasn't battled in ____days_____ then bla
ie; If ___username___ HAS battled in ___days___ then "We've found a possible match"

also you lose your rank if you keep losing battles, and your rank goes up if you keep winning. So someone that qualified in the Platinum tier can ascend to the Diamond tier if they win a few battles. But if they constantly lose, they'll go down to Gold.

@X what you think about this? @Nicholas @Rant

Pointless because you don't have the userbase to pull it off plus this isn't a game where raw mechanical skill, knowledge on the latest metas from balance patches and partying up vs solos take president over knowing who you're VSing. Theres a reason people here prefer to pre arrange battles\have 3 days notice from tourneys ect. Plus most people prefer to pick their opponents spontainiously to get themselves motivated rather than have ELO RNG dictate they face *insert name* For the tenth time this week because the site doesn't have the numbers to support an ELO MM.
Which ever way you want to slice it, it wont work.

Quote:

Originally Posted by Indianapolis Jones (Post 1122924)

I've actually been working on a system off and on in my spare time writing pseudo code to try and figure out a good way to do the ranking system and I think it addresses this problem.

My method involves assigning a points to users that they gain or lose according to whether they win or lose. How many points the winner takes from the loser ranges on factors from the difference in estimated skill level, current rank and w/l record. Basically it works by having a range of 0-N where N is the cap of the maximum amount of points that can be taken. A user facing another with a much lower skill level than them would gain nothing from the win but an increase in w/l. Basically it would nerf someone like @Bnas from doing his goofy shit and being able to crack the top 100 because at a certain point you gain nothing from battling people that much lower than you. Also, the more you do it the faster it nets you nothing because your w/l goes up. I'll post it below but it's unfinished and a mess at this point.

Keep in mind this is pseudocode (Written with p5.JavaScript syntax for the most part):

Code:

var maxEstimatedSkill = 10;

var ELOPoints = NumberOfRegisteredUsers;

var MaxEloChange = 32;

//W-L expressed as decimal

var winLoss = 0;

//I'm not sure how exactly to express a mathematical formula for determining exactly how to remove the number of points taken

var pointSubtraction;



var User1 = {

//All numbers in here are random. I didn't feel like writing some actual code in

//a spreadsheet to distribute the ELO points amongst the current user count

//(139,421) randomly and then sorting it. This should do well enough to express what

//I'm getting at.

    currentRank: 35,

    EstimatedSkill: 7,

    CurrentEloPoints: 15000,

    winLoss = 75.3

}



var User2 = {

    currentRank: 1000,

    EstimatedSkill: 7,

    CurrentEloPoints: 1200,

    winLoss = 25.2

}



function Battle() {

    

    Predict = {

    User1PredictionScore = user1.CurrentEloPoints + (user1.currentRank + user1.EstimatedSkill + user1.winLoss);

    User2PredictionScore = user2.CurrentEloPoints + (user2.currentRank + user2.EstimatedSkill + user2.winLoss);

      if (User1PredictionScore > User2PredictionScore) prediction = User1 wins;

      if (User2PredictionScore > User1PredictionScore) prediction = User2 wins;

      if (User1PredictionScore == User2PredictionScore) prediction = CoinFlip; //do nothing

    }

    

    DetermineWinner = {

    if (user1VoteTally > greater than user2VoteTally) {

    user1 wins} else { 

    user2 wins

     }

    }

    

    Winner = Result

    

    DetermineRankChange = {

    if Winner was predicted {

    ELOAdded = MaxEloChange - (pointSubtraction)

    }

    if Winner was not predicted {

    ELOAdded = UnderDogWin

    UnderDogWin() {

    //Determine the amount of points added to the winner based on the amount of difference between the two.

    //Probably by using the PredictionScore.

        }    

    }

    }

}

Quote:

Originally Posted by Mindless (Post 1122931)

Code:

var maxEstimatedSkill = 10;

var ELOPoints = NumberOfRegisteredUsers;

var MaxEloChange = 32;

//W-L expressed as decimal

var winLoss = 0;

//I'm not sure how exactly to express a mathematical formula for determining exactly how to remove the number of points taken

var pointSubtraction;



var User1 = {

//All numbers in here are random. I didn't feel like writing some actual code in

//a spreadsheet to distribute the ELO points amongst the current user count

//(139,421) randomly and then sorting it. This should do well enough to express what

//I'm getting at.

    currentRank: 35,

    EstimatedSkill: 7,

    CurrentEloPoints: 15000,

    winLoss = 75.3

}



var User2 = {

    currentRank: 1000,

    EstimatedSkill: 7,

    CurrentEloPoints: 1200,

    winLoss = 25.2

}



function Battle() {

    

    Predict = {

    User1PredictionScore = user1.CurrentEloPoints + (user1.currentRank + user1.EstimatedSkill + user1.winLoss);

    User2PredictionScore = user2.CurrentEloPoints + (user2.currentRank + user2.EstimatedSkill + user2.winLoss);

      if (User1PredictionScore > User2PredictionScore) prediction = User1 wins;

      if (User2PredictionScore > User1PredictionScore) prediction = User2 wins;

      if (User1PredictionScore == User2PredictionScore) prediction = CoinFlip; //do nothing

    }

    

    DetermineWinner = {

    if (user1VoteTally > greater than user2VoteTally) {

    user1 wins} else { 

    user2 wins

     }

    }

    

    Winner = Result

    

    DetermineRankChange = {

    if Winner was predicted {

    ELOAdded = MaxEloChange - (pointSubtraction)

    }

    if Winner was not predicted {

    ELOAdded = UnderDogWin

    UnderDogWin() {

    //Determine the amount of points added to the winner based on the amount of difference between the two.

    //Probably by using the PredictionScore.

        }    

    }

    }

}

Is this with an automated MM system or pick your battles in mind?

lmao.. goldmic already had a ratings system in place years ago, ive been bringing that idea up for a long time.. for each profile, a person is given a +/- followed by a designated points, corresponding to how strong the opponent ia based on opponents, and rank.. so a strong battler will see a weak battlers points as +5/-35 meaning if he keeps battling scrubs, hell only gain 5 points toward his rating, but if he loses ,he could lose 35... if a strong battler faces one thats his level, hed see +15/-15.. etc.etc. this will encourage battlers to challenge stronger opponents to move up quicker..dunno how the point system is now, but at least with this in place, you can visually see what points youd gain, or lost against certain opponents

Quote:

Originally Posted by ILLoKWENT (Post 1122936)

This sounds better than the original suggestion. I don't like the idea of forced match-ups and I don't think suggested match-ups would even be acknowledged. The problem is, what do you guys do now? Delete all previous points? That seems unfair.

I love the idea of an ELO style rating system.

Quote:

Originally Posted by Indianapolis Jones (Post 1122932)

Is this with an automated MM system or pick your battles in mind?

You would pick your own battles. It's kind of based on the way that RPG stats are gained in that you can only level up by facing opponents with stats that are equal to or higher than your own. Basically we're applying the law of diminishing returns to battling. The difference is that someone low ranked doesn't risk as much facing someone much higher than them.

Quote:

Originally Posted by ILLoKWENT (Post 1122936)

I kind of like this but I think that's too much information to give. IMO, the only information anyone should have on anyone is their current rank, w/l and ESL. If you give people too much insight into the system they figure out a way to game it and use it in ways it wasn't intended. Also, what would a weak battler see on the page of a strong battler? What's the cap on point loss?

Quote:

Originally Posted by Nicholas (Post 1122941)

Freeze the current rankings and put them somewhere and say "This is our old ranking system". Then you restart everyone at zero.

Also, forced match-ups is lame unless it's voluntary. What I mean by that is, for the general system it's bad. If it were it's own system it could be fun though. I don't see it ever happening though.

I mean...it's points based on votes based on VP which is better then just straight wins.

It's why @UNKNOWN ARTIST w/ 282 wins & 5 losses is #24 of all time (combined: 4687 points, 287 battles) vs. @The Real Deal w/ 470 wins & 309 losses is #68 of all time (combined 3475 points, 779 battles).

With a difference of 1,212 points and 492 battles you can clearly see why battling shitty opponents doesn't mean you're better.

I'm not sure what my point was but basically it's working as well as it could be.

Side note: If you guys are collectively tired of Bnas being #1 of all time you can battle him, spit the tooley hard and lower his points and with enough big battles (lots of voters) he could eventually has less points then Mayneak and get bumped down. Of course this only works if he stopped battling or battles less scrubs and more high caliber opponents.

Or we could just do what we shoulda done the thousandth time we perma banned him.... Perma ban him. And erase his account for good measure. Just fking do it and stick to it ffs. He's owed no right/goodwill.