EP Ladder Model
EP Ladder Model
With the recent announcement of the Escape New World Championship, its scale and use of the EP ladder to determine the 64 players who will compete for the top 8, the ESOC team would like to open a short discussion regarding the model that the ESOC EP ladder is using and propose a new model.
Currently, the EP ladder uses the Elo rating system. Read more about it here. The Elo system has been used in Age of Empires rating dating back to the MS Gaming Zone days and has traveled with the franchise through time. The consistency of seeing the Elo type rating is great, but since the launch of the original Elo ladders for Age of Empires, improved models have come along such as Glicko. As I have watched the top of the ladder recently, with the advent of 2019 Year of the Turk and TT takeover, I have noticed that the rankings are very stale in terms of points gained and the current model is not reactive well to activity, in fact, considering you start at 1600, one could argue it rewards cherry picking games and being inactive. To fix this, we propose going to the Glicko model. Glicko type models are widely used across esports, in games such as CSGO, many Chess servers and federations, TF2, Microsoft's TrueSkill system, and others). Due to the additional factors in the Glicko system, the ratings are much reactive than Elo models. All factors considering, it should provide a better climate and ultimately more accurate ladder for future use.
Currently, the EP ladder uses the Elo rating system. Read more about it here. The Elo system has been used in Age of Empires rating dating back to the MS Gaming Zone days and has traveled with the franchise through time. The consistency of seeing the Elo type rating is great, but since the launch of the original Elo ladders for Age of Empires, improved models have come along such as Glicko. As I have watched the top of the ladder recently, with the advent of 2019 Year of the Turk and TT takeover, I have noticed that the rankings are very stale in terms of points gained and the current model is not reactive well to activity, in fact, considering you start at 1600, one could argue it rewards cherry picking games and being inactive. To fix this, we propose going to the Glicko model. Glicko type models are widely used across esports, in games such as CSGO, many Chess servers and federations, TF2, Microsoft's TrueSkill system, and others). Due to the additional factors in the Glicko system, the ratings are much reactive than Elo models. All factors considering, it should provide a better climate and ultimately more accurate ladder for future use.
mad cuz bad
-
- Musketeer
- Posts: 98
- Joined: Jul 22, 2017
- ESO: 2ndLastKnight
Re: EP Ladder Model
Yes, as an avid chess player, Glicko is the standard across pretty much all chess sites. It also has the benefit that your rating fluctuates a lot when few games are played, and gets more stable over time, so old players without ESOC ratings who want to shoot up the rating list can do that pretty quickly. Most online chess tournaments also have a requirement that your rating is non-provisional (i.e., over 20 games played), to prevent new accounts with super high ratings from gaming the system.
Re: EP Ladder Model
I think inactivity is the most important thing to try to fix.
- [Armag] diarouga
- Ninja
- Posts: 12710
- Joined: Feb 26, 2015
- ESO: diarouga
- Location: France
Re: EP Ladder Model
Well of course you want to fix inactivity but the issue I see with a "reactive elo" is that you can lose 20 ranks on a bad streak or get 20 with a good streak. Also smurfing is an issue, anyone could get top 5 by getting free wins from someone else, and I actually know some guys who got free wins to have a better seeding in tourneys.
I think that the qualification brackets should be made very carefully because the New Year Classic was a huge mess regarding that.
I think that the qualification brackets should be made very carefully because the New Year Classic was a huge mess regarding that.
Re: EP Ladder Model
I don't think ranks are stale at all. You can go up and down easily because the EP ladder is relatively new. Only reason why it may be stale is because players just don't play, for which no ranking system provides a solution.
At the end, current EP rankings are fairly accurate so I don't see the issue in the first place. Also 64 is a big number. Even if someone is left out, surely he weren't someone who would make it to the top8.
At the end, current EP rankings are fairly accurate so I don't see the issue in the first place. Also 64 is a big number. Even if someone is left out, surely he weren't someone who would make it to the top8.
- [Armag] diarouga
- Ninja
- Posts: 12710
- Joined: Feb 26, 2015
- ESO: diarouga
- Location: France
Re: EP Ladder Model
Garja wrote:I don't think ranks are stale at all. You can go up and down easily because the EP ladder is relatively new. Only reason why it may be stale is because players just don't play, for which no ranking system provides a solution.
Yea, I agree. Although you see some weird things from time to time (like turk or tabben getting #1), the elo ladder is quite accurate and the top 12 is the real top 12.
At the end, current EP rankings are fairly accurate so I don't see the issue in the first place. Also 64 is a big number. Even if someone is left out, surely he weren't someone who would make it to the top8.
Yes, of course, but you seem to forget that the qualifiers seedings will use the elo ladder, so being 4th or 9th isn't exactly the same thing, and as I said, the New Year Classic bracket was shit.
Re: EP Ladder Model
[Armag] diarouga wrote:Well of course you want to fix inactivity but the issue I see with a "reactive elo" is that you can lose 20 ranks on a bad streak or get 20 with a good streak. Also smurfing is an issue, anyone could get top 5 by getting free wins from someone else, and I actually know some guys who got free wins to have a better seeding in tourneys.
I think that the qualification brackets should be made very carefully because the New Year Classic was a huge mess regarding that.
We are going to monitor this.
mad cuz bad
- [Armag] diarouga
- Ninja
- Posts: 12710
- Joined: Feb 26, 2015
- ESO: diarouga
- Location: France
Re: EP Ladder Model
n0el wrote:[Armag] diarouga wrote:Well of course you want to fix inactivity but the issue I see with a "reactive elo" is that you can lose 20 ranks on a bad streak or get 20 with a good streak. Also smurfing is an issue, anyone could get top 5 by getting free wins from someone else, and I actually know some guys who got free wins to have a better seeding in tourneys.
I think that the qualification brackets should be made very carefully because the New Year Classic was a huge mess regarding that.
We are going to monitor this.
Also bashing noobs and dodging is not fair either.
Ideally we would want a system like in Starcraft 2 where the only rated games are the QS games so you have to play against everybody, but that's not realistic.
Re: EP Ladder Model
QS only would be ideal but that's in general for EP. For this specific tourney seed all you want is to fix a baseline of activity like 20 games or so played in the last month before the qualifiers.
Re: EP Ladder Model
Don't know much about Glicko--it's probably better--but IMO the main reason the rankings may seem stale is just that there's not a lot of incentive to play rated games. People 80-100 elo below yourself are just not worth playing, period, because you can win 10 games and gain almost nothing and lose 1 game and lose as much as you won from 5 games. There's also no rating decay, which means once you've got a decent elo and you're happy with it there's just no point in playing more, and allows players like Raphael and H2O to just sit at the top of the ladder for all time, despite having not played a single rated game in over half a year.
It would be nice if there was some kind of combined elo from RE and EP as well, maybe the RE games can have like 50% of the impact on your rank that a rated EP game would have or something, just because of how volatile they are with the cheating on QS and the maps, etc. But in general it would just be nice to be able to impact your rank via QS.
It would be nice if there was some kind of combined elo from RE and EP as well, maybe the RE games can have like 50% of the impact on your rank that a rated EP game would have or something, just because of how volatile they are with the cheating on QS and the maps, etc. But in general it would just be nice to be able to impact your rank via QS.
Re: EP Ladder Model
what exactly does glicko do differently?
Re: EP Ladder Model
If the max points won/lost was capped at like 15-20, i think what we have now would be better, tho that might not be possible.
- edeholland
- ESOC Community Team
- Posts: 5033
- Joined: Feb 11, 2015
- ESO: edeholland
- GameRanger ID: 4053888
- Clan: ESOC
Re: EP Ladder Model
Cometk wrote:what exactly does glicko do differently?
Gotcha fam http://www.glicko.net/glicko/glicko2.pdf https://www.englishchess.org.uk/wp-cont ... Glicko.pdf
TL;DR: Glicko takes inactivity and rustiness into account, which is good for a the current ladders where the ranking is relatively stagnant.
Re: EP Ladder Model
From what I undestand glicko doesn't solve the problem of inactivity. All it does is claim to provide a more accurate ranking just because it considers a range of values rather than a precise number.
How do you use one player's rating range then? What value do you pick for seeding? Top of its range? Bottom?
How do you use one player's rating range then? What value do you pick for seeding? Top of its range? Bottom?
-
- Musketeer
- Posts: 98
- Joined: Jul 22, 2017
- ESO: 2ndLastKnight
Re: EP Ladder Model
Garja wrote:From what I undestand glicko doesn't solve the problem of inactivity. All it does is claim to provide a more accurate ranking just because it considers a range of values rather than a precise number.
How do you use one player's rating range then? What value do you pick for seeding? Top of its range? Bottom?
You usually just use the average. No rating system can solve the problem of people not playing, of course, but Glicko has the advantage in that you can arrive at your "true" rating in less games.
Re: EP Ladder Model
charlemango wrote: You usually just use the average. No rating system can solve the problem of people not playing, of course, but Glicko has the advantage in that you can arrive at your "true" rating in less games.
Doesn't the PR system native to ESO do both of those things (address inactivity and lets players reach true rating in fewer games)?
somppukunkku wrote:This is not a fucking discogame.
- edeholland
- ESOC Community Team
- Posts: 5033
- Joined: Feb 11, 2015
- ESO: edeholland
- GameRanger ID: 4053888
- Clan: ESOC
Re: EP Ladder Model
PR might be good if it wasn't shared between RE and EP.Darwin_ wrote:charlemango wrote: You usually just use the average. No rating system can solve the problem of people not playing, of course, but Glicko has the advantage in that you can arrive at your "true" rating in less games.
Doesn't the PR system native to ESO do both of those things (address inactivity and lets players reach true rating in fewer games)?
Re: EP Ladder Model
You should use elo for specific case, in esoc cause player pool is not big, and game conditions/oppenent is choosen. Your examples of games have large player pool and oppenents are random. Dont know anything about Gicko, just wanted to emphasize you should consider situation in order to say better or worse. And imo reason elo is fluctating a lot is simply player base is not big, (everybody knows everyone, no suprise effect, and you can choose MU you want). In this condition, kynesie can pick indonesia and play water all day and get 2000 elo.
Re: EP Ladder Model
Imo anybody who's made it to semis in an esoc tourney and tries to sign up should get automatic placement in at least the ro64
Re: EP Ladder Model
edeholland wrote:PR might be good if it wasn't shared between RE and EP.Darwin_ wrote:charlemango wrote: You usually just use the average. No rating system can solve the problem of people not playing, of course, but Glicko has the advantage in that you can arrive at your "true" rating in less games.
Doesn't the PR system native to ESO do both of those things (address inactivity and lets players reach true rating in fewer games)?
I think PR is still linked to civ win rates and can be manipulated with low level cities. But I got your point.
Elo+decaying system would be great if the purpose is to use it for seeding. Basically any Elo + activity incentive.
Re: EP Ladder Model
But there is also the component of PR that lets new players, as well as players who have been inactive for awhile and decayed, level up super quickly (I believe I reached pr 15 on my main account in 2 games) which I think is very important.
somppukunkku wrote:This is not a fucking discogame.
Who is online
Users browsing this forum: No registered users and 27 guests