FIDE Ratings are going to start from 1400 ! What's your opinion ?

I can understand that switching to GLICKO is a daunting task - it's a different algorithm, and the consequences are probably hard to predict. And then there's the logistic problem of how to organize the switch.

Would everyone just start at 1500 and let the algorithm do its magic over a couple of months, maybe using ELO and GLICKO in parallel for some time? Or could one use the players' last FIDE/ELO ratings as a proxy for the starting rating in GLICKO?

With all the data that the FIDE statisticians have, it would be at least interesting to do a simulation how that switch-over scenario could play out - similar to the simulations that Jeff Sonas did for his proposed "compression".

Glicko is Elo with a better formula for the k-factor. No one with a stabilized rating should experience big changes.

im a noob im 11 and i got 1150 fide I almost never play fide games

@sheckley666 said in #22:
> Glicko is Elo with a better formula for the k-factor. No one with a stabilized rating should experience big changes.

I'm not sure that's true. Yes, the principle is similar, but Glicko's Rating Deviation (RD) works quite differently than the K-Factor in ELO. For example, RD is meant to also increase during periods of inactivity, and one would need to find parameters for this time-dependent increment that the OTB community would agree to - that's more a political than a statistical question.

And then there's the question of starting values and how similar (or not) the absolute rating values are between the two systems. Absolute values of the ratings are pretty meaningless in both Glicko and ELO, in principle, BUT FIDE uses absolute ELO numbers as thresholds for various master norms. So would these norms need to be adapted, and if so, how?

There are lots of other devils hiding in the details for sure.

@Panagrellus said in #21:
> I can understand that switching to GLICKO is a daunting task - it's a different algorithm, and the consequences are probably hard to predict. And then there's the logistic problem of how to organize the switch.
>
> Would everyone just start at 1500 and let the algorithm do its magic over a couple of months, maybe using ELO and GLICKO in parallel for some time? Or could one use the players' last FIDE/ELO ratings as a proxy for the starting rating in GLICKO?
>
> With all the data that the FIDE statisticians have, it would be at least interesting to do a simulation how that switch-over scenario could play out - similar to the simulations that Jeff Sonas did for his proposed "compression".

I'd expect it would be quite straightforward to "replay" several years of past games using the Glicko algorithm, so that you can basically assume Elos from 10+ years ago work as glicko values, and then replay all the results in the database from then to the current time to achieve current glicko values. Would probably be more accurate than either just switching current Elo as Glicko, or starting all over again from 1500 (which would probably cause a civil war)

This seems to make sense from technical and theoretical point of view but it could bring tons of complaints of the "you got it wrong, X had lower ELO than me and now he has more, that's unfair" type. Just see how difficult it is for some people here to understand why they "don't get full points refunded" when cheating is addressed. There are also some technical details like the choice initial RD values and initial ratings for new players who were introduced during the recalculated interval but it shouldn't be that hard to sort those out.

I asked fide if they will make the changes at the ratings, they said "Once approved, the changes will be communicated to the public in December 2023 "

Any more news from fide whatsoever? They really should make a public statement

I'm pretty sure that when I began rated chess, oh so long ago, they started at 1400.
But it's been so long. So very long. Nurse? Can you bring me a chocolate?

Ok, let's be serious. A 1500 rating was an achievement when I was young. Seriously, it was. Snicker away.

Any day now