It seems a number of people are losing 1 rating point somewhere.
One example is Aryan Bezawada (USCF ID 30691059). His last two tournaments the rating goes 1697 –>1704 and then 1703–>1712. One point disappears
Another is Ian Rodrigues (USCF ID 17068052). His last two tournaments are 1480–>1574 and then 1573–>1578. Again a point disappears. I have a number of similar examples
I’ll look into this tomorrow, as I’m about to head to bed. It sort of sounds like a rounding issue, perhaps in how the pre-event rating is shown. The 1704 was actually 1703.86, and maybe the new crosstable program is truncating rather than rounding to get to an integer value.
So does the new calculation start from the base of 1703.00 or 1703.86?
While I defer to Mike’s superior knowledge of the system here, my understanding is that it’s used the decimal values behind the scenes for quite some time; in the MSA system you could even click back and forth to see the decimal or the integer values.
Internally, 1703.86 (and several more digits) is the post-event rating which should also be the pre-event rating when the next event is rated for that person. That’s been the case since September of 2014 when we switched from storing integer ratings as the post event rating to storing them in floating point.
But what gets displayed on MUIR is, at this point, at least, still an integer, and it is possible that 1703.86 was truncated to 1703 rather than rounded to 1704 as it was for the November ratings list.
I actually don’t have direct access to the MUIR database today because I’m in the process of setting up a new laptop and that’s one of the things that hasn’t been set up on it yet. Otherwise I could go see what the actual FP numbers are in the database and perhaps resolve the question quickly. :sigh:
Leago has confirmed that this ratings display issue is indeed caused by not rounding and does not affect the flow of ratings in floating point form from event to the next. It should be fixed for future events but it won’t be updated for existing events until we do the first rerate, hopefully next week.
Another fun question for you: Krish Ladi had two tournaments rated out of order. One of them has a base of 1718 and another has a base of 1717 US Chess
We cannot control the order in which tournaments are submitted, and when one comes in out of time sync either way we rate it, one or more tournaments are now wrong because they don’t show the linkage from one event to the next.
Here’s an example. Suppose someone plays in 3 tournaments. If the events are submitted in order, that player’s ratings would go like this:
A - 1452 → 1467
B - 1467 → 1448
C - 1448 → 1481
However, tournament B is submitted late, so the order in which the events are received is A, C then B.
This gives us:
A - 1452 → 1467
C - 1467 → 1485
So, when event B is received, which rating should we use as the pre-event rating for this player when rating event B, 1467 or 1485?
If we use 1467, then event C has the wrong starting rating and wrong ending rating compared to what they should have been, although B is correct.
If we use 1485, then event B has the wrong starting rating and wrong ending rating compared to what they should have been, and so does C.
That’s why we do rerates. The first rerate will hopefully be run next week.
My apologies: I should have framed the question better
For the player in question, his 10.27 tournament is 1743–>1718
The next three tournaments are rated out of order (which is understandable). But one of them uses 1718 as a base and two of them use 1717
According to the explanation from several days ago, it seems that the base should be 1718 or 1717, but not both.
My point is, no matter which way we do it, there will be multiple tournaments with the wrong starting rating rather than the correct rating for that event once it is rerated. It is unavoidable.