Here in Maine, we are planning a “Triathlon of Chess” with 8 rounds of (G/3+2) Blitz followed by 4 rounds of G/15+5 Rapid and finally 2 rounds of G/30+5 Classical/Dual.
I would like to use the results of the Blitz to inform the 1st round of the Rapid and the results of the Rapid to inform the first round of the Classical/Dual event. My thought is to take the results of the Blitz and pair the first round of the Rapid using Swiss pairing rules (rather than starting from scratch and letting the software pair the first round based on ratings only). Likewise, I would like to use the results of the Rapid to pair the first round of the Classical/Duel.
I would appreciate your thoughts on this. Would you prefer this system or treating the sections as separate events with no carryover. (The organizers prefer the carryover method as that lends itself better to the “triathlon” feel of the event.)
For rating purposes, separate events or separate sections (making sure to designate what the time controls were).
For pairing, pair an additional round in the blitz, then force those pairings in the first round of the rapid. Once you have done that, you can revert the blitz to the previous round. Similarly for the rapid→classical transition.
A single section cannot have time controls that qualify for more than one ratings system with the exception of dual rated OTB games, which must be at least G/30 (MM +SS). (Dual rating was created to add more games to the quick system by extending it to G65.)
So if you want to have an event that progresses through all of the OTB ratings systems you could submit it as up to 4 sections:
a blitz section for G5-G10
a quick rated (only) section for G11-G29
a dual rated section for G30-G65
a regular rated (only) section for G66 or slower
Also, all games in a blitz section must be at the exact same time control.
Please note that if you are using the results of the blitz to pair only the first round of the rapid and then the results of the rapid to pair only the first round of the dual, then you have three different, albeit loosely linked, events of 8, 4 and 2 rounds rather than one event of 14 rounds. One way of doing that is to pair a 9th round of the blitz (never played) to get your round one rapid pairings and to pair a 5th round of the rapid (never played) to get your round one dual pairings.
Since you need to rate them separately anyway that tracks. If the blitz is actually four double-rounds (play each opponent once with White and once with Black) then it goes quicker and would essentially be only a four round Swiss.
If you want to keep all of the results from the prior time controls when making the new pairings then you will need to later split them up for rating purposes. Pair all 14 rounds and then make three copies. In one you delete rounds 9-14 and submit it for blitz rating. In another you convert the rounds 1-8 results to forfeit wins/losses and unplayed draws and then. after making a copy for the final step, you delete rounds 13-14 and submit it for rapid (quick-only) rating. In that copy with the converted rounds 1-8 results you also convert rounds 9-12 to forfeit wind/losses and unplated draws and submit it for dual rating (or regular only if you have something like a G/60;d10 time control). Pairing programs might not allow double rounds for only some of the rounds, so if the blitz are double-rounds while the others are not you would have to manually pair the second half of each double-round.
The k-factor of each section would be less than the k-factor for a 14-round Swiss but that is appropriate since each section would be in a different rating class.
For TD certification purposes the blitz would be a 4-(double) round or 8-round Swiss for a Category D blitz event (or higher category if there are 50+ actual players after ignoring any players who never played). The rapid would be a 2-(double) round or 4-round Swiss with the 2-double round not counting as any category or the 4-round counting as a category D quick event (or higher under the same conditions mentioned for blitz). The dual (or regular if the time control is long enough) would be a 2-round event that does not fall into any category.
I would probably pair the entire event as a single section for pairing purposes, but break it up into multiple sections for ratings purposes. That way you can keep colors balanced and avoid players facing the same opponent more than once.
This is where I would like to have a check box to treat a 2nd pairing against the same player be ok if they have not been paired against each other in the last N rounds.
I once ran a 12 round insanity where I only had about 16 players total so I paired as 3 - 4 round sections adding the final scores. I think I may have peeked to see if folks in same score group played each other already in prior a prior section before repeating a pairing.
A nice utility could split a 1 section swiss into N section by rounds – not too difficult for the SwissSys tournament json version. Also, not too bad to convert the DBF files as they are quite simple.
I worked on a system that had a bunch of utilities that were only in the super secret original developer documentation. Over the years and new developers every time we had certain types of request come up we thought we would have to write the utility which added about 40-60 hours to the task - so it always got rejected since the manual way was to fix was always estimated at 2-10 hours by admins. Many years later our developer was fixing something totally unrelated and found about 25 utilities/reports that the original developers had made just in case and one of them did exactly what we needed – they were not in the original specs so they did not get documented in the official space on purpose (my guess they were done with non-billed I/T resources)
If this can be accomplished after natural pairings of a group looking at it as if they have not yet played then it might not be too complex a change.
but after thinking about it some more do you need yet another parm so there can only be a rematch if it is the only way to pair the score group or perhaps it improves colors by the 80/200 or improves by dropping the lower rated player since the rematch is ok. so a bit of a rabbit hole.
Some years back WinTD assigned internal penalty points to pairing way out of score-group or pairing a rematch. Around round ten of a 16-18 player blitz tournament it started re-pairing a pair of leaders/near leaders each round when the score group difference penalty exceeded the re-pair penalty. I’m not sure if it still does that now that we’ve switched to doing double-rounds in a blitz tournament.