Ranked Elo Overhaul

Leave your suggestions about the game here!

Re: Elo Rework

Postby Joacgroso » Sat Jan 20, 2018 6:03 pm

I think conqueror fits more for a war game. Champion por master would be better.
Sorry if I have spelling mistakes. English is not my main language.
I added a petition to my achievement rework!
Plese come and vote to support it!
If you have any suggestions related to achievements, please post them on my thread so we can suggest them together!
User avatar
Joacgroso
Arsonist
Arsonist
 
Posts: 2665
Joined: Mon May 02, 2016 6:21 pm
Location: Argentina

Re: Elo Rework

Postby Flake » Sat Jan 20, 2018 6:46 pm

Joacgroso wrote:I think conqueror fits more for a war game. Champion por master would be better.

Agreed. I've now changed Master to Expert and Conqueror to Master.

So we now have Bronze, Silver, Gold, Platinum, Diamond, Expert, Master in that order.
Flake
Consort
Consort
 
Posts: 1081
Joined: Tue Oct 28, 2014 8:34 am
Location: England, UK

Re: Elo Rework

Postby Jackparrot » Sat Jan 20, 2018 6:50 pm

You say some where in the post that there should only be a limited amt of games that your ELO is based off. Last I saw it was 200 or so, i would change it to something like 5-50 As you can be a good player, but have a temporary losing streak, and that would ruin everything
I am not as active as I used to be, although I should be somewhat active.
Jackparrot
Sheriff
Sheriff
 
Posts: 579
Joined: Thu Mar 30, 2017 10:16 am

Re: Elo Rework

Postby Flake » Sat Jan 20, 2018 7:06 pm

Jackparrot wrote:You say some where in the post that there should only be a limited amt of games that your ELO is based off. Last I saw it was 200 or so, i would change it to something like 5-50 As you can be a good player, but have a temporary losing streak, and that would ruin everything

There is now no such limit on the amount of games considered, since in this version of the rework I feel it isn't very difficult to catch up to the higher ranked Tiers if you are truly at that skill level, even if you do have a losing streak in Placement games by chance.
Flake
Consort
Consort
 
Posts: 1081
Joined: Tue Oct 28, 2014 8:34 am
Location: England, UK

Re: Elo Rework

Postby Flake » Sat Jan 20, 2018 7:08 pm

A poll has now been created, please Vote! The Poll question is "Do you think this Elo Rework should replace the current Elo system?".
Flake
Consort
Consort
 
Posts: 1081
Joined: Tue Oct 28, 2014 8:34 am
Location: England, UK

Re: Elo Rework

Postby Joacgroso » Sat Jan 20, 2018 7:22 pm

Flake wrote:
Joacgroso wrote:I think conqueror fits more for a war game. Champion por master would be better.

Agreed. I've now changed Master to Expert and Conqueror to Master.

So we now have Bronze, Silver, Gold, Platinum, Diamond, Expert, Master in that order.

I really like that names. I voted yes on the poll.
Sorry if I have spelling mistakes. English is not my main language.
I added a petition to my achievement rework!
Plese come and vote to support it!
If you have any suggestions related to achievements, please post them on my thread so we can suggest them together!
User avatar
Joacgroso
Arsonist
Arsonist
 
Posts: 2665
Joined: Mon May 02, 2016 6:21 pm
Location: Argentina

Re: Elo Rework

Postby Flake » Sat Jan 20, 2018 7:23 pm

Joacgroso wrote:
Flake wrote:
Joacgroso wrote:I think conqueror fits more for a war game. Champion por master would be better.

Agreed. I've now changed Master to Expert and Conqueror to Master.

So we now have Bronze, Silver, Gold, Platinum, Diamond, Expert, Master in that order.

I really like that names. I voted yes on the poll.

Thanks for the suggestions and thanks for voting!! :-)
Flake
Consort
Consort
 
Posts: 1081
Joined: Tue Oct 28, 2014 8:34 am
Location: England, UK

Re: Elo Rework

Postby Jackparrot » Sat Jan 20, 2018 8:04 pm

Also, not only does ELO need to be re-worked, so does the constraints to get into ranked, currently it is very easy to get into ranked, play 50 RP games and done. I would make it more strict, such as winning with every role atleast once. please include this in your guide
I am not as active as I used to be, although I should be somewhat active.
Jackparrot
Sheriff
Sheriff
 
Posts: 579
Joined: Thu Mar 30, 2017 10:16 am

Re: Elo Rework

Postby Joacgroso » Sat Jan 20, 2018 8:06 pm

People should not be forced to win with roles that don't spawn in ranked.
Sorry if I have spelling mistakes. English is not my main language.
I added a petition to my achievement rework!
Plese come and vote to support it!
If you have any suggestions related to achievements, please post them on my thread so we can suggest them together!
User avatar
Joacgroso
Arsonist
Arsonist
 
Posts: 2665
Joined: Mon May 02, 2016 6:21 pm
Location: Argentina

Re: Elo Rework

Postby Jackparrot » Sat Jan 20, 2018 8:09 pm

Joacgroso wrote:People should not be forced to win with roles that don't spawn in ranked.


Ok then with all roles that appear in ranked
I am not as active as I used to be, although I should be somewhat active.
Jackparrot
Sheriff
Sheriff
 
Posts: 579
Joined: Thu Mar 30, 2017 10:16 am

Re: Elo Rework

Postby Joacgroso » Sat Jan 20, 2018 8:14 pm

I think increasing the amount of games would be better. Forcing people to buy scrolls in order to win with every role wouldn't be fair. Also, people could get their wins in unbalanced gamemodes, like all any.
Sorry if I have spelling mistakes. English is not my main language.
I added a petition to my achievement rework!
Plese come and vote to support it!
If you have any suggestions related to achievements, please post them on my thread so we can suggest them together!
User avatar
Joacgroso
Arsonist
Arsonist
 
Posts: 2665
Joined: Mon May 02, 2016 6:21 pm
Location: Argentina

Re: Elo Rework

Postby wozearly » Sat Jan 20, 2018 9:00 pm

Flake wrote:Thank you SO much for the analysis!

Yes, the main idea of this rework is to be entirely immune to grinding. Both changing factors other than win rate, which are the opposing elo & team elo and increase/decrease in elo change due to faction, eventually balance out and become 0 after many games, leaving win rate as the only factor having influenced the person's Elo. At the same time, these two factors are quite influential with a lower amount of games, which I think should be the case.

I think the main issue I'm having with this is I'm unsure about the extent to which the two factors, opposing elo & team elo and increase/decrease in elo change due to faction, have an impact on the Elo change. I'm starting to think that increase/decrease in elo change due to faction should have more of an impact on Elo change, as you said, but I'm honestly not too sure. As such, perhaps K * (A% - F%)/(3A%) would be more appropriate than the current K * (A% - F%)/(4A%)?

Thanks again!


Yeah, there are two key aspects of team vs opponent ELO that might benefit from a sense-check:

1) To what extent does excluding your own ELO lead to odd situations (and how bad is including it)
2) How much importance should the calculation attribute to ELO of team/opponents versus defying your expected winrate

My lingering concern about effectively ignoring your own vs opponent ELO when you're a solo faction is that while this makes sense over time, assuming you typically play games within your tier (or adjacent tiers), in individual games it may create some strange situations. For example, a lower ELO NK taking on a good Town and Mafia receives no greater reward, and no blunted loss, compared to a high ELO NK facing off against people at the bottom end of the tier below.

While in NK's case you do typically need a dose of fortune to get a win, you are heavily influenced by how well the team factions work with themselves and against each other.

That said, it is also counterintuitive where repeatedly being the highest ELO player on a team is actually advantageous to you in terms of ELO gain, and being the lowest ELO player sees you (in effect) being punished for having good teammates. There may be other ways to handle team ELO that don't require excluding your own results, though. I'll plug some bits in and see.

Similarly, I'll run the same setup through 3*A and see what the effect appears to be.
wozearly
Veteran
Veteran
 
Posts: 491
Joined: Wed Dec 28, 2016 6:48 am

Re: Elo Rework (UNOFFICIAL, POLL INSIDE)

Postby BasicFourLife » Sun Jan 21, 2018 6:38 am

281 edits WTF.

I edit my posts a lot, but even I don't have that many.
If Shapesifter13 tells you that your suggestion will be considered, it never will.
If you’re making a post, thinking you’re gonna impact the game, you won’t.
If you make an another fucking thread about the poor state of the game, you should move on.
If you suggest that Firefighter be added to the game or the fact that Retributionist should die when they resurrect, I will personally murder you instead.

FM / TG: 5 - 10 - 8
BasicFourLife
[Forum Mafia XVI] Winner
[Forum Mafia XVI] Winner
 
Posts: 1535
Joined: Sat Dec 17, 2016 6:47 am
Location: UTC +2/+3 - Somewhere hidden

Re: Elo Rework

Postby Jackparrot » Sun Jan 21, 2018 7:12 am

Joacgroso wrote:I think increasing the amount of games would be better. Forcing people to buy scrolls in order to win with every role wouldn't be fair. Also, people could get their wins in unbalanced gamemodes, like all any.


They still get to learn how the role works in all any, people dont have to buy scrolls to get the role they want, they can get it normally.
I am not as active as I used to be, although I should be somewhat active.
Jackparrot
Sheriff
Sheriff
 
Posts: 579
Joined: Thu Mar 30, 2017 10:16 am

Re: Elo Rework

Postby Flake » Sun Jan 21, 2018 7:33 am

wozearly wrote:
Flake wrote:Thank you SO much for the analysis!

Yes, the main idea of this rework is to be entirely immune to grinding. Both changing factors other than win rate, which are the opposing elo & team elo and increase/decrease in elo change due to faction, eventually balance out and become 0 after many games, leaving win rate as the only factor having influenced the person's Elo. At the same time, these two factors are quite influential with a lower amount of games, which I think should be the case.

I think the main issue I'm having with this is I'm unsure about the extent to which the two factors, opposing elo & team elo and increase/decrease in elo change due to faction, have an impact on the Elo change. I'm starting to think that increase/decrease in elo change due to faction should have more of an impact on Elo change, as you said, but I'm honestly not too sure. As such, perhaps K * (A% - F%)/(3A%) would be more appropriate than the current K * (A% - F%)/(4A%)?

Thanks again!


Yeah, there are two key aspects of team vs opponent ELO that might benefit from a sense-check:

1) To what extent does excluding your own ELO lead to odd situations (and how bad is including it)
2) How much importance should the calculation attribute to ELO of team/opponents versus defying your expected winrate

My lingering concern about effectively ignoring your own vs opponent ELO when you're a solo faction is that while this makes sense over time, assuming you typically play games within your tier (or adjacent tiers), in individual games it may create some strange situations. For example, a lower ELO NK taking on a good Town and Mafia receives no greater reward, and no blunted loss, compared to a high ELO NK facing off against people at the bottom end of the tier below.

While in NK's case you do typically need a dose of fortune to get a win, you are heavily influenced by how well the team factions work with themselves and against each other.

That said, it is also counterintuitive where repeatedly being the highest ELO player on a team is actually advantageous to you in terms of ELO gain, and being the lowest ELO player sees you (in effect) being punished for having good teammates. There may be other ways to handle team ELO that don't require excluding your own results, though. I'll plug some bits in and see.

Similarly, I'll run the same setup through 3*A and see what the effect appears to be.


Yeah, I honestly wish I could make it so that the player's Elo is considered, but I can't see how I could do it while still maintaining balance. As an extreme example, say you are in Master Tier and are at a significantly higher Elo than the majority of players in Master. Since you play against Diamond, Expert and Master Tiers, the average player Elo will be significantly lower than yours. As such, after many games, you will lose quite a significant amount of Elo and be dragged towards the average Elo within Diamond, Expert and Master simply for having a higher Elo. The same extreme example could be applied to Bronze, where having a significantly lower Elo than everyone else means you are dragged up easily simply for having a lower Elo. The same effect happens in every tier but to a lesser extent.

I don't think Elo gain/loss should be centripetal, that is you shouldn't be thrown towards the centre Elo simply for having a lower/higher Elo than the people you play against. At the same time, I feel like your own Elo should be considered, especially for NE/NK so that you gain/lose more depending on a higher/lower Elo. You really have to choose one or the other, and I think the former is more important in the long run.

Thanks so much again!!
Last edited by Flake on Sun Jan 21, 2018 7:35 am, edited 1 time in total.
Flake
Consort
Consort
 
Posts: 1081
Joined: Tue Oct 28, 2014 8:34 am
Location: England, UK

Re: Elo Rework (UNOFFICIAL, POLL INSIDE)

Postby Flake » Sun Jan 21, 2018 7:34 am

BasicFourLife wrote:281 edits WTF.

I edit my posts a lot, but even I don't have that many.

Yeah, I keep making very minor edits and then submitting, should probably stop doing that!
Flake
Consort
Consort
 
Posts: 1081
Joined: Tue Oct 28, 2014 8:34 am
Location: England, UK

Re: Elo Rework (UNOFFICIAL, POLL INSIDE)

Postby Shilster » Mon Jan 22, 2018 5:49 am

Amazing and very well made

Needs more attention, you should advertise this in the discord if you haven't
-




It matters not who you are, Death awaits you.

-
User avatar
Shilster
Werewolf
Werewolf
 
Posts: 3483
Joined: Mon Apr 03, 2017 9:59 pm
Location: (GMT+7)

Re: Elo Rework (UNOFFICIAL, POLL INSIDE)

Postby Flake » Mon Jan 22, 2018 10:00 am

Shilster wrote:Amazing and very well made

Needs more attention, you should advertise this in the discord if you haven't

Thank you! I actually haven't thought of doing that! Thanks for the suggestion :)
Flake
Consort
Consort
 
Posts: 1081
Joined: Tue Oct 28, 2014 8:34 am
Location: England, UK

Re: Elo Rework (UNOFFICIAL, POLL INSIDE)

Postby Joacgroso » Mon Jan 22, 2018 5:54 pm

It would also be nice if players were able to know the formula after a ranked game, and know all the values behind their elo gain/loss.
Also, when you say winrates, do you mean general winrates? It may be nice if you applied winrates from their elo bracket instead of from all ranked games, although I'm not sure about that.
Sorry if I have spelling mistakes. English is not my main language.
I added a petition to my achievement rework!
Plese come and vote to support it!
If you have any suggestions related to achievements, please post them on my thread so we can suggest them together!
User avatar
Joacgroso
Arsonist
Arsonist
 
Posts: 2665
Joined: Mon May 02, 2016 6:21 pm
Location: Argentina

Re: Elo Rework (UNOFFICIAL, POLL INSIDE)

Postby Flake » Mon Jan 22, 2018 6:17 pm

Joacgroso wrote:It would also be nice if players were able to know the formula after a ranked game, and know all the values behind their elo gain/loss.
Also, when you say winrates, do you mean general winrates? It may be nice if you applied winrates from their elo bracket instead of from all ranked games, although I'm not sure about that.

As for your first point, it might be interesting and informative to show the three affecting factors on the player's win rate, and I don't really see a reason not to do so. So I support that suggestion.

As for your second point, average winrate and faction winrates are taken from the player's Elo bracket (±1 tier), which is stated in Section 2:

• "A%" (Average Tier Win Percentage) refers to the average win rate of players in your current Tier, give or take one tier, for all games they have played in those Tiers. For example, if a player is in Platinum and the average win rate of all players in Gold, Platinum and Diamond Tiers for all games all players have been in Gold, Platinum and Diamond for is 60%, A% = 0.6.

• "F%" (Average Faction Win Percentage) refers to the average win rate, of any given faction, of players in your current Tier, give or take one tier, for all games they have played in those Tiers. For example, if the player is Bronze and the average Mafia win rate of all players in Bronze and Silver Tiers for all games all players have been in Bronze and Silver for is 40%, F% = 0.4.
Flake
Consort
Consort
 
Posts: 1081
Joined: Tue Oct 28, 2014 8:34 am
Location: England, UK

Re: Elo Rework (UNOFFICIAL, POLL INSIDE)

Postby Flake » Wed Jan 24, 2018 8:11 am

Bump
Flake
Consort
Consort
 
Posts: 1081
Joined: Tue Oct 28, 2014 8:34 am
Location: England, UK

Re: Elo Rework

Postby wozearly » Wed Jan 24, 2018 5:51 pm

Flake wrote:Yeah, I honestly wish I could make it so that the player's Elo is considered, but I can't see how I could do it while still maintaining balance. As an extreme example, say you are in Master Tier and are at a significantly higher Elo than the majority of players in Master. Since you play against Diamond, Expert and Master Tiers, the average player Elo will be significantly lower than yours. As such, after many games, you will lose quite a significant amount of Elo and be dragged towards the average Elo within Diamond, Expert and Master simply for having a higher Elo. The same extreme example could be applied to Bronze, where having a significantly lower Elo than everyone else means you are dragged up easily simply for having a lower Elo. The same effect happens in every tier but to a lesser extent.


Well, assuming that ELO is actually a relatively fair indication of the player's skill based on observable past performance, that dragging effect is arguably intentional.

If you are a better-than-average player facing off against worse-than-average opponents, you should be coming out with a better-than-average win rate - otherwise your observed performance is not in line with your current ELO, and your ELO should drop as a result. The same is true even if your own team are made up of worse-than-average players (although the impact would be much smaller for the high ELO player in this case, because the team's ELO is being averaged)

The most critical factor in this is that the importance of the ELO calculation modifying for self-vs-opponent ELO difference is correctly balanced against the importance of typical winrates for the factions. In a nutshell, to what extent would we expect a better player to overachieve in the long run against weaker opponents? How much influence do we think their skill has on average winrates, compared to luck of the draw and natural balancing in Ranked (ie, doesn't matter how good you are, you're not going to get a super-high winrate as NK in the long-term)?

This is where factors like 4*A vs 3*A would typically come into their own, as they effectively set the ratio of importance between the two factors. What's difficult is getting a feel for what is 'right' without access to the same stats that BMG have at their disposal - this is probably something the devs themselves would need to take a view on.
wozearly
Veteran
Veteran
 
Posts: 491
Joined: Wed Dec 28, 2016 6:48 am

Re: Elo Rework (POLL INSIDE)

Postby Joacgroso » Thu Jan 25, 2018 9:44 pm

Another suggestion: what about taking rolelists into account? Mafia shouldn't be punished too much for losing against a town with ret, 2 meds and a trans, but it could be more punished for a list with 3 RT spies, for example. Things like that could make the formula even more balanced.
Sorry if I have spelling mistakes. English is not my main language.
I added a petition to my achievement rework!
Plese come and vote to support it!
If you have any suggestions related to achievements, please post them on my thread so we can suggest them together!
User avatar
Joacgroso
Arsonist
Arsonist
 
Posts: 2665
Joined: Mon May 02, 2016 6:21 pm
Location: Argentina

Re: Elo Rework (POLL INSIDE)

Postby BasicFourLife » Fri Jan 26, 2018 6:37 am

Joacgroso wrote:Another suggestion: what about taking rolelists into account? Mafia shouldn't be punished too much for losing against a town with ret, 2 meds and a trans, but it could be more punished for a list with 3 RT spies, for example. Things like that could make the formula even more balanced.

Um, this could be fixed by BMG by making the role balanced?
If this was implemented it would mean that BMG is saying that their game is broken and that they are not fixing it, and instead implementing this.

Think about this Joac.
If Shapesifter13 tells you that your suggestion will be considered, it never will.
If you’re making a post, thinking you’re gonna impact the game, you won’t.
If you make an another fucking thread about the poor state of the game, you should move on.
If you suggest that Firefighter be added to the game or the fact that Retributionist should die when they resurrect, I will personally murder you instead.

FM / TG: 5 - 10 - 8
BasicFourLife
[Forum Mafia XVI] Winner
[Forum Mafia XVI] Winner
 
Posts: 1535
Joined: Sat Dec 17, 2016 6:47 am
Location: UTC +2/+3 - Somewhere hidden

Re: Elo Rework (POLL INSIDE)

Postby Joacgroso » Fri Jan 26, 2018 8:57 am

It would still improve the elo sistem. The only way of balancing that would be implementing the different, secret rolelists that were suggested before. Otherwise they would have to change everything.
Another thing: WWs shouldn't lose too many elo in games like tp=bg TI=spies TS=med TK=Vet RTs=spies/meds, where they couldn't kill anyone in a rampage. Maf should also have a different winrate if NK=arso and NE=jester instead of NK=SK NE=Witch. That should also be taken in account.
Sorry if I have spelling mistakes. English is not my main language.
I added a petition to my achievement rework!
Plese come and vote to support it!
If you have any suggestions related to achievements, please post them on my thread so we can suggest them together!
User avatar
Joacgroso
Arsonist
Arsonist
 
Posts: 2665
Joined: Mon May 02, 2016 6:21 pm
Location: Argentina

PreviousNext

Return to Suggestions

Who is online

Users browsing this forum: orangeandblack5 and 1 guest