Massive rerate of USCF history database caused rating loss

Tim,

Thank you. I think that works to confirm that Anthony Booth 12400817 should have been entered in with a floor of 1900 effective 2012-01-01, but as Mike notes this will have to wait until membership corrections are up and running in MUIR, followed by a rerate.

Richard,

In the meantime TDs could rely on 28E, 28E1, 28E2 and 28E3 to assign a rating of 1900 to this player when they enter an event.

Good Chess,

Tim

Yes; I think the original concern was more for all the Arizonans who farmed rating points off of the 1900 floor rather then the 1800 one.

Some years ago, the EB established a rule that a player who is on their floor and plays in a match can be treated as having made a request to have their floor lowered by 100 points. It isn’t used very frequently.

It’s been a long time since I looked at the paper supplements. Did *21 indicate a floor of 2100 or did it indicate a provisional rating based on 21 games. If it was provisional then the first established rating could have been somewhere in the 2000-2099 range, which would have given a 1900 floor at that time (change the last two digits to zero and subtract 100). Later that would have been changed to 1800 (when the formula became change the last two digits to zero and subtract 200).

So if the * indicates provisional it may very well be that 1800 is the correct floor.

I think *21 meant an established rating with a peak rating of 2100-2199.

I’m not sure what the ones like *d3 ones meant anymore, though I think Tom Doan posted something about them a year or two ago. Something to do with title earned and norms towards the next title in the old norms system, which was poorly designed, I think.

1 Like

Probably not based off games considering the 1983 rating supp in Jan chess life ā€˜84 shows a 2068

Keep in mind that even in the 80’s a player’s peak rating might not ever show up in a ratings supplement. My peak rating was 1651 (in 1987, I think), but I think my peak published rating was more like 1588 because I had one great tournament where I peaked and then gave a lot of points back a couple of weeks later.

Jeff,

I believe provisional ratings had a / in the old days; i.e. /1213 is a provisional rating of 1213.

I checked my rating in that same supplement and I had a *17. That indicates a floor. In those days I recall (but have no solid evidence) that once a player passed over a 00 barrier, then that barrier was the new floor for that individual. So, if a player went over 1500 but below 1600 then their floor was 1500 or *15.

Good Chess,

Tim

Yes, in general whenever provisional ratings are shown, the game count should appear after a /

There’s an open ticket to fix some places where this is not yet happening on MUIR.

The Gold Master files should have the game count in the R_NRM_DAT and Q_NRM_DAT fields (because the ratings field is a numeric type), the tab files should have them in the ratings field.

When does Korey plan to do corrections?

I’m not sure what the ones like *d3 ones meant anymore, though I think Tom Doan posted something about them a year or two ago. Something to do with title earned and norms towards the next title in the old norms system, which was poorly designed, I think.

I thought *d# meant # (out of 10) points towards getting the 1200 norm. *D# would be 1300, c# would be 1400, etc. But I thought # could only be 2, 4, 5, 6, 7, 8 or 9. A performance that gave a norm for a level was worth 2 points for that level, 5 points for the level below and all ten points for the level below that. So *b2 would be 2 points for the 1600 norm while having already earned he 1500 level and *C5 would be 5 points toward the 1500 level and 2 points toward the 1600 level (C7 would be 7 points toward the 1500 level and still only 2 points toward the 1600 level.

Numbers 2, 4, 6 and 8 meant that many points towards that level while having already earned the level below and having nothing for the level above.

Numbers 5, 7 and 9 meant that many points towards that level, having earned the level below and having 2 points for the level above, i.e. B7 meant the 1600 level was earned, the 1700 level has 7 of the 10 points needed and the 1800 had 2 of the 10 points needed.

Those norms were (much) easier to earn. If memory serves, with a 21xx rating I think I got up to M7 (7 of 10 for the 2200 norm and 2 of 10 for the 2300 norrn) and the letters from 2500 and down (by increments of 100) were S (2500), s, M, m, X, x (2000), A, a, B, b, C (1500), c, D, d and maybe even E, e (1000), F, f, G, g, H, h.

Checking in on an update on the rating fixes.

Still working on some rerate issues. There’s also been a discussion of moving to a 10 year floating rerate window, and that would impact floor changes from more than 10 years ago.

We have a player at our local club who this affected and caused him to lose a floor and dropped his rating by nearly 100 points. He is anxious to get his old floor back. Is there an approximate timeframe I can give him?

Thanks,

I don’t currently have an ETA. If the 10 year floating rerate window proposal goes through, that would impact anyone whose floor change was more than 10 years ago, so we’d probably need a new strategy for dealing with those. Otherwise, the manual floor records are in place, we just need to be able to rerate back that far.

Any rerate news? Obviously we do manual rating adjustments, but just want to see if anything changed

We put in over 200 floors, some with effective dates all the way back to 2004. We’re not yet doing rerates to events prior than 2025, but what I did last week was to put in those same floors with effective dates of 1/1/25, those should show up in the early Tuesday morning rerate this week, they missed the cutoff for last week.

This will only help people who have been active since 1/1/25, of course.