Sign up Login
Store
Players Online: 5.9k
Info
Rules Staff Discord
Support Suggestions Forums Client
Forums Developer Updates Special Developer Log #4 - Moving away from ELO and S13
Subscribe
Special Developer Log #4 - Moving away from ELO and S13

Posted on 12 July 2016 - 07:43 PM

The new kb is defiantly smooth and my hits seem to register. The only problem I see is the actual distance you get knocked back. For example it is extremely difficult to get a combo in the speed ladders as both players take an insane level of kb, which resets the fight after the first hit trade.
 0
PM Link

Posted on 12 July 2016 - 08:29 PM

BaneOfArthropods wrote

SmellyPenguin wrote...


It is not elo. Elo is always zero-sum, this rating is not, which is exactly why this rating is so irrelevant.


This rating isn't irrelevant because it's your rating lol. It's literally what your rating is.
 0
PM Link

Posted on 12 July 2016 - 08:33 PM

I really like that this change is taking place, it solves a lot of problems I have with elo (being scared to get drained and fairness in who I fight) and really takes a step in the right direction. It's nice to see that Minecraft is becoming more and more like competitive games such as League of Legends in terms of ranking.

My only suggestion is to either change how "diamond" and "emerald" are placed on the boards. Emerald is more rare than diamond, therefore it should be placed higher than Diamond, correct? I don't really care too much (definitely not reaching either) but I think it makes more sense logistically.

Regardless, I'm really excited to see this in action soon! Amazing job developers, you all continue to impress me with your knowledge and ideas.
 0
PM Link

Posted on 12 July 2016 - 08:33 PM

Oremitron wrote

Why can't you just show our raw number value alongside our rank? Noone gives a shit because a few idiots which make up about 1% will be toxic about numbers, just /ignore them ezpz. Or just make bragging or making fun of others for their stats mutable. Make it be like Gold III - 1804 or something so we can more accurately see where we stand, and also know how close or how for we are from getting promoted or demoted. Also so we can compare with people ion the same rank as us.


When we first released Elo ratings, nobody knew how good an 1800 was or how good a 2200 was. People just played and played until eventually people understand how good a player with some Elo rating was. In the new system the same standards will be set. Players will get shuffled around in their divisions until they are put into the proper division and the same standards will be set. People will be able to know how good someone is by looking at their division and seeing that they are Silver 3 or Gold 1 or Plat (Emerald) 4.
 0
PM Link

Posted on 12 July 2016 - 08:34 PM

https://beta.badlion.net/forum/thread/126189

@SmellyPenguin this would help allot, atleast I think so.
 0
PM Link

Posted on 12 July 2016 - 08:47 PM

SmellyPenguin wrote



This rating isn't irrelevant because it's your rating lol. It's literally what your rating is.

IMO, any rating system that is not zero-sum is irrelevant.
 0
PM Link

Posted on 12 July 2016 - 08:51 PM

BaneOfArthropods wrote

SmellyPenguin wrote...


IMO, any rating system that is not zero-sum is irrelevant.


That's a pretty bold statement. I suggest you do some research on the topic so you can enlighten yourself.
 0
PM Link

Posted on 12 July 2016 - 09:02 PM

BaneOfArthropods wrote

SmellyPenguin wrote...


IMO, any rating system that is not zero-sum is irrelevant.


So basically every system used in every relevant video game nowadays?

topkek
 0
PM Link

Last edited on 12 July 2016 - 09:15 PM by Monasuko

SmellyPenguin wrote

BaneOfArthropods wrote...



That's a pretty bold statement. I suggest you do some research on the topic so you can enlighten yourself.
It my opinion, if your rating is decided by some random program no one knows if it's right or fair. The beauty of Elo is the simplicity. Sure you may die to a hacker or two (usually when people call hacks the person wasn't hacking, he just got a nice combo) but the system is always fair.
 0
PM Link

Posted on 12 July 2016 - 09:14 PM

Archybot wrote

BaneOfArthropods wrote...



So basically every system used in every relevant video game nowadays?

topkek
Yeah, pretty much. If you want a true measure of skill, it has to be zero-sum. If you want to know how much better a Gold player is than say a Bronze player you need a straight forward system, like Elo. With Elo you can tell a 2000 rated player (assuming he has peaked) is roughly 31x as skilled as an average 1400 player, because he must win 31 times to every loss to maintain his Elo against that player. That's why Elo works so well, because you can tell how much better you are than the average player. It's not just some random medal.
 0
PM Link

Posted on 12 July 2016 - 09:36 PM

MaccariTA wrote

HIT DETECTION IS SHIT IN SEASON 13
ur non donor what u talkin about
 0
PM Link

Posted on 12 July 2016 - 10:15 PM

Tyguy95 wrote

I just find it annoying how the kits that already take forever to queue in (in terms of nobody playing) i.e Iron, UHC, Diamond, Soup, etc will take even longer now with the new "Kohi-like" queuing system…. anyway to help fix this problem?



True
 0
PM Link

Posted on 12 July 2016 - 11:36 PM

I just want to throw in a huge thank-you to Archy for working so hard on making a better, more friendly and usable ranking system for Badlion and for taking the time to write up this novel of a post all about it. I feel much more informed about how the elo system, this new rank system and the KB/"hit detection" all function now. Thanks so much, @Archybot. I assume I speak for a large part of Badlion in my saying these things.
 0
PM Link

Posted on 13 July 2016 - 12:05 AM

BaneOfArthropods wrote

Archybot wrote...

Yeah, pretty much. If you want a true measure of skill, it has to be zero-sum. If you want to know how much better a Gold player is than say a Bronze player you need a straight forward system, like Elo. With Elo you can tell a 2000 rated player (assuming he has peaked) is roughly 31x as skilled as an average 1400 player, because he must win 31 times to every loss to maintain his Elo against that player. That's why Elo works so well, because you can tell how much better you are than the average player. It's not just some random medal.


Basically your argument is "Elo is a lot simpler to understand and therefore it is the best rating system and every other rating system sucks".
 0
PM Link

Last edited on 13 July 2016 - 12:19 AM by Monasuko

SmellyPenguin wrote


Basically your argument is "Elo is a lot simpler to understand and therefore it is the best rating system and every other rating system sucks".

How am I supposed to understand the new ranking system if there is no way to tell a) how far I am from ranking up or b) how big a skill gap there is between the ranks. With Elo you can determine both of these things with 1rst and 2nd grade math (respectively).

Don't tell me I'm an idiot for not understanding an obtuse and opaque system.
 0
PM Link

Posted on 13 July 2016 - 12:41 AM

BaneOfArthropods wrote

SmellyPenguin wrote...


How am I supposed to understand the new ranking system if there is no way to tell a) how far I am from ranking up or b) how big a skill gap there is between the ranks. With Elo you can determine both of these things with 1rst and 2nd grade math (respectively).

Don't tell me I'm an idiot for not understanding an obtuse and opaque system.


You can tell how far you are from ranking up/down as you play the game more and more. You'll eventually notice a pattern in the system. The public decides the skill gap w/ the ranks the same way they did for Elo.
 0
PM Link

Posted on 13 July 2016 - 01:00 AM

SmellyPenguin wrote


You can tell how far you are from ranking up/down as you play the game more and more. You'll eventually notice a pattern in the system. The public decides the skill gap w/ the ranks the same way they did for Elo.
But that's not what happened with Elo. People didn't "just decide" a 2k player was 31x as good as an average 1400 player. We figured that out from the Elo itself (if a 2k dies to a 1400 he loses 31, but if he wins he gains just 1). This tells us that a 2k player must be 31x as good as a 1400 if he is going to keep his Elo up.

There is no parallel with the new ranking system. You gain rating. You lose rating. But nowhere in that do you ever find if Diamond III is twice as good as Emerald I or ten times as good. Do you see what I mean by opaque?
 0
PM Link

Posted on 13 July 2016 - 01:24 AM

BaneOfArthropods wrote

SmellyPenguin wrote...


IMO, any rating system that is not zero-sum is irrelevant.


Irrelevant to what?

BaneOfArthropods wrote

SmellyPenguin wrote...

It my opinion, if your rating is decided by some random program no one knows if it's right or fair. The beauty of Elo is the simplicity. Sure you may die to a hacker or two (usually when people call hacks the person wasn't hacking, he just got a nice combo) but the system is always fair.


At what point is the rank implementation 'random'? Glicko-esque rating reliability factors are deterministic.

BaneOfArthropods wrote

Archybot wrote...

Yeah, pretty much. If you want a true measure of skill, it has to be zero-sum. If you want to know how much better a Gold player is than say a Bronze player you need a straight forward system, like Elo. With Elo you can tell a 2000 rated player (assuming he has peaked) is roughly 31x as skilled as an average 1400 player, because he must win 31 times to every loss to maintain his Elo against that player. That's why Elo works so well, because you can tell how much better you are than the average player. It's not just some random medal.


Oh boy.

  1. Skill is not perfectly quantifiable via a single number. Rating systems yield approximates. All rating systems, Elo included, have a degree of inaccuracy. There is no true measure of skill, that's *exactly* why we develop rating systems.
  2. Differing playstyles and diverse skillsets don't cleanly fit into your "I am 16x better than _" world. See Dota2, LoL; players with the same rating can have incomparable skillsets.
  3. Ranked playerbase distribution data allows you to compare yourself to the community. If players in Master-tier are in the 99th or 100th percentile, you can make a good guess.

BaneOfArthropods wrote

SmellyPenguin wrote...


How am I supposed to understand the new ranking system if there is no way to tell a) how far I am from ranking up or b) how big a skill gap there is between the ranks. With Elo you can determine both of these things with 1rst and 2nd grade math (respectively).

Don't tell me I'm an idiot for not understanding an obtuse and opaque system.


See above re: skill.

You actually bring up a good point though. LoL and CS:GO (and other games) have suffered from the ranked system having hidden factors. Of course, rankings aren't actually decided randomly, but sometimes it can feel like they are.
 0
PM Link

Posted on 13 July 2016 - 01:53 AM

BaneOfArthropods wrote

We figured that out from the Elo itself (if a 2k dies to a 1400 he loses 31, but if he wins he gains just 1). This tells us that a 2k player must be 31x as good as a 1400 if he is going to keep his Elo up.


LOL WTF?
 1
PM Link

Posted on 13 July 2016 - 02:00 AM

How will the leaderboard look with this system? Will it show all the people with the same rank as equal or will it still order them by exact skill?
 0
PM Link