Glicko Rating System

[Animiral]

I was just browsing a bit on lichess.org. According to their FAQ, they use the Glicko2 rating system.

Frequently Asked Questions • lichess.org

This reminded me of an old rating complaint thread here that I posted in.
Ranked games - different Lvl Players - NO handicap:

@Animiral

On a slightly related note, the Wikipedia article on the Elo system has some really interesting bits. Specifically:

From a modern perspective, Elo’s simplifying assumptions are not necessary because computing power is inexpensive and widely available.
Moreover, even within the simplified model, more efficient estimation techniques are well known.

And then this:

Subsequent statistical tests have suggested that chess performance is almost certainly not distributed as a normal distribution, as weaker players have greater winning chances than Elo’s model predicts.

If this is true, maybe it would be interesting to try out some variations on the rating system and see if there is a noticeable difference.
Are any of you mathematically enclined enough to explain some of those ‘well known’ better estimation techniques?

That thread died shortly afterwards.

Now, lichess is like the OGS of chess: It’s a shiny, new, Web 2.0 style, open-source chess platform written in Scala, and it is gaining in popularity. The fact that they decided to use Glicko2 probably means that someone knew what they were doing and it’s a good choice for a competitive game platform today.

More info on Glicko:

Glicko rating system

Welcome to Glicko ratings (contains implementations in R, Python, Javascript, PHP and Microsoft Excel)
“Example of the Glicko-2 system”

To me, it sounds like a simple enough improvement on Elo.

I’m not saying that OGS should change its rating system to use Glicko right now. But as some people, myself included, were probably wondering what the proper alternatives to Elo were, now we know

[anoek]

We certainly looked at a number of ranking systems (including Glicko), however we decided to go with the EGF modifications on ELO since it gave us an established way of accounting for handicaps in ranked games.

[kamoro]

I’ve been toying with rating systems for a while now. One of my main annoyances with ELO is picking a good K-factor. You have to compromise between accuracy and adjusting quickly to differences between rank and strength.

By the way, on OGS, is the K-factor for corresponence games different from the K-factor for live games? I would think they should be, since correspondence games don’t bring in new data as quickly, and thus warrent a higher K-factor.

[anoek]

We use the EGF’s K factor guidelines, which basically scales with rank. I’ve questioned it a bit myself… but I figure they did a fair amount of thinking on good numbers.

We don’t use different K factors for game speed, but we do for board size (9x9 counts less than 13x13 which counts less than 19x19). For correspondence games, it does get a little funky when you have really long games where people change ranks in the middle of a game, but such is life. You make up for the slowness of by virtue of the fact that folks tend to play a lot at the same time, so you can still progress through your ranks at a (relatively) good pace.

[flovo]

This thread is outdated since Jul 2017. At this point the OGS rating system changed to glicko2. The rating system changed in many aspects. All info in this thread about Elo on OGS is obsolete.

OGS has a new Glicko-2 based rating system! [2017]

tl;dr We’re switching to a Glicko-2 rating system, which should be great for a lot of reasons. All of your past ranked games have been accounted for during this transition. Hello OGS! For the past several years we’ve been using a rating and ranking system based on Elo developed by the European Go Federation. It has served us reasonably well, but there have been some shortcomings when applying that system in an online setting. Most of the shortcomings can be traced back to the fact that the sy…