There is a TD Tip after rule 5C that has a note that states:
“Note 3: If an event or section(s) within an event have different schedules that merge, the rating report submission process now allows the director to include the time controls for each schedule.”
This makes it seem like you could list the time control for each schedule (and list who was in each schedule) and the programming would rate each schedule according to the time control listed for that schedule (I know this isn’t how the programming works).
Until MUIR can be changed I would guess these tournaments can be handled by keeping a copy of the pre-merge quicker sections, changing the post-merge games from those sections to non-rated results, and then rating the pre-merge sections at one time control and the reduced post-merge sections at another time control.
If there was sufficient interest in it (which is unclear), it might be something the pairing program authors would add, what I would call a linkage rather than a merge, an option where for pairing and standings purposes, including tiebreaks, two sections are treated as one merged section, but for reporting purposes they remain two separate sections, that way there’s no post-event fiddling around needed to submit the rating report.
This need not be only for events that fit into different ratings systems, we ran our city championship one year as several RR flights which seeded into two or three 2nd level RR flights, with total standing from both the prelims and finals being used to determine the champion from the championship section. But because this took about 3 months overall, we submitted the first set of RRs for rating long before the second set of RRs was over.
How does that note not accurately describe the MUIR setup? Can you not put in the time control by round/schedule when you are submitting a tournament for rating like you could prior to MUIR?
No, at this time the MUIR time control parser only supports one time control, it gets confused if there is more than one. Whether that will be changed at some point is unclear, A few years back the office permitted TDs to just put in the slowest time control rather than enter a lot of data that, practically speaking, is not used for much.
I looked at the May 2026 ratings for all players with a status of ‘A’ and a rating in both the regular and quick time controls. There are just under 82000. Then I looked at the players in each block of 10000: 1-10000, 10001-20000, 20001-30000, etc., and saw how similar the blocks were if you sorted by regular rating or if you sorted by quick rating. The blocks are incredibly similar, with 70-90% overlap across the board:
00001-10000: 8543 common players whether ordered by quick or regular 10001-20000: 7028 20001-30000: 7050 30001-40000: 7613 40001-50000: 8174 50001-60000: 8689 60001-70000: 9121 70001-80000: 9470
Even if you look at regular and blitz (just under 20000 players here):
00001-10000: 8769 common players whether ordered by blitz or regular 10001-19529: 8298
The fact that so many players seem to be in the same ranking tranches regardless of what rating system is looked at seems to question those reasons.
As far as I can see, the best players at regular are the best players at quick and the best players at blitz, the average the same, and the bad the same.
That is, I don’t see any compelling reason for multiple rating systems based on looking at where the players seem to shake out.
I’m not sure if I still have the graphs somewhere, but some years ago, working with Mark Glickman, the ratings chair, I created several sets of graphs showing the expected performance curve and the actual performance. As I recall I divided the data into several sets by rating and by rating system/time control.
In every subset, the lower rated player generally did somewhat better than the expected performance. That’s understandable, especially in younger and lower rated players, because they’re probably still improving rapidly. (That’s why we have a bonus factor, too.)
Looking at Blitz, Quick-only, dual rated events and regular-only events as separate subsets, the faster the time control the better the lower rated player did. This was more true when looking at games between lower rated players (under 1600 or 1800, I think) than in games among A players through Masters.
The conclusion we drew was that faster time controls led to more upset wins and draws. I can think of multiple reasons for this.
Ok, this does feel intuitively correct. And yet the rating system doesn’t seem to account for this at all, as the win probability of the rating formula (“Standard winning expectancy,” We) is the same for regular, quick, and blitz?
Yeah, this did trigger some discussion as to whether the expected performance curve should be different for faster time controls, but the RC didn’t feel that was necessary. I’d have to go back to Elo’s book to see how he originally derived that curve, but the PhDs on the committee were talking over my head at that point.
It did seem to justify keeping quick-only and blitz games separate from dual/regular rated events. Dual rating, you may recall, added games that were regular rated to the pool for quick rating, so it did not directly increase the number of regular-rated games.
Groups of 10,000 are fairly large. If you granularize into groups of 1000 or 100-point rating bands, you would see more differences. (groups of 1000 would better check relative strengths rather than expecting ratings to match up between regular and quick).
Interestingly enough, if I compare the 5744 players who had an established blitz rating and an established regular rating and were active in 2025, I find the correlation coefficient is 0.923.
But if I divide these players up by 100 point rating intervals, the correlation coefficient approaches 0 for most ratings classes.
If I do the same thing for regular and quick ratings (N=33769), the overall correlation coefficient is .985 and when I look at 100 point rating intervals, the correlation coefficient is in the 0.75 range for players rated under 200 and that goes down almost lineally as ratings go up, so by the time you get to the 2000 players the correlation coefficient is 0.206.
I think what this means is that for the regular/blitz comparison the numbers were so small that the differences were essentially random. I’m not sure why the regular/quick numbers start out high and drop down, perhaps it’s because lower rated players are generally younger and more likely to play in dual-rated events, where the dual-rated games are affecting both the regular and the quick rating pretty much equally.
