Here is an example that illustrates the question I am trying to ask here. The winner of the U1700 section of the recently completed National Open, uschess.org/msa/XtblMain.php?202106200392.5, is listed as going from Unrated to 1777P12. He played 7 games in the National Open which means his US Chess rating was initialized based on 5 games from some other rating system(s). It looks like in this case his rating was initialized based on his FIDE rating as he doesn’t have any of the other five US Chess ratings. In the past, a players initialized rating would be listed as there pre-event rating on the crosstable instead of them being listed as unrated, as long as it wasn’t P0. Is this no longer going to be the case?
This is not an intentional change, but the previous practice, which apparently only affected those whose initial rating was based on a FIDE rating, wasn’t intentional, either.
In fact, we consciously avoided showing someone’s initialized pre-event rating, because in many cases it could be used to infer that person’s birthdate. Somehow this got handled differently for initializations based on a FIDE rating.
Under the new blended rating initialization formula, someone who has a FIDE rating and any other type of rating, for example an online blitz rating, would have a blended rating based on all of the available ratings information we have.
When computing a blended initialization, we don’t record in the database how many different ratings were used to initialize it, so it would be difficult to restore the previous practice of showing a pre-event rating (in this case, I think it would have been 1515/5) when only a FIDE rating is involved.
Any change to this would have to come through channels and might involve significant programming time and database changes.
FWIW, this player’s rating went up to 1969 after the World Open.
I am not sure what, exactly, a 1420 FIDE rating (inactive since 2017) meant before the pandemic, much less after it.
According to our records (we capture the FRL monthly, that’s what we use to find someone’s FIDE Rating as of the start of an event), his rating went from 1545 to 1420 in February of 2020, which appears to be the last time he played in a regular-rated FIDE event.
He also has a FIDE Rapid rating of 1621 from 2017. We don’t currently use FIDE Rapid or FIDE Blitz ratings in the blending process, though, and we have no historical database of them, so adding them to the blending formula would require building and maintaining one, among other things.
In any event, he appears to be stronger than either of those ratings at this time, so our ratings system is playing catch-up.
I overlooked the rating change in February 2020.
Interestingly, the player in question participated in an U-14 event last year. The other players in that tournament were born in 2006, 2007 and 2008. However, this player was born in 1998, at least according to FIDE. Looking further, I notice a lower rated player from the same country with only a different middle initial who was born in 2007. Perhaps the arbiters used the incorrect player ID for the tournament in February 2020.
Would a player rated 1545 FIDE be eligible to enter an U1700 or U1800 USCF section? The CCA website converts 1545 FIDE to 1750 USCF and 1420 FIDE to 1650 USCF. National Open was not a CCA event, but World Open was one. Granted, this player scored 8.0 out of 9 in U1800, so even the 1750 rating seems terribly low.
Disclaimer: I realize this potential discrepancy is not the responsibility of the directors in Las Vegas or Philadelphia.
Michael Aigner
The USCF adult conversion formula converts 1420 to about 1515 and 1545 to around 1632. The youth conversion formula would produce slightly higher numbers, but the blending formula does not say that we should use the youth conversion formula, much less what the age cutoff would be.
I tend to agree with your suggestion that the 2020 event might have had the wrong player, but I’m not sure that’s something our FIDE team can even take to FIDE for consideration. But the more players you have in the database (FIDE or US Chess), the greater the chance of an ID confusion due to similar names.
Perhaps this is why that event is not included in the latest rating supplement.
Alex Relyea
I’m not sure what you mean, as of yesterday afternoon that 2020 event was still showing on his FIDE profile.
It is, but it’s not a part of his rating.
Alex Relyea
Looks to me like FIDE still has his rating at 1420, which is what it dropped to in the February 2020 list. Is there somewhere else I should be looking at instead of ratings.fide.com/profile/13310429
I’m extremely embarrassed. I got confused somehow. You’re right, of course.
Alex Relyea
Is the rating system document located at glicko.net/ratings/rating.system.pdf going to be updated to include the new way player ratings are initialized? The above link is given in several places on the US Chess website so it would be good if the document was current.
I emailed Mark Glickman and he updated the rating system document located at glicko.net/ratings/rating.system.pdf.