I have McCabe ahead of Valby...Roe as well. They are solid hill runners. If Valby can run the hills, she makes this very interesting.
Will Valby go out extremely fast...I don't think so...I believe her and Tuohy will sit at the front together. Tuohy knows she can run a low 19 pace on a hilly course, does Valby? Mccabe, Chmiel, Mercy and Roe can run hills very well. Of this top group, Valby has the least experience running this type of course and front running against this talent.
Maybe Valby tries to push hard to the first hill, but I don;t think Tuohy or McCabe let her get much of a gap. And then we will see how things stand.
Regardless...it will be a banger of a race!!
Speaking of LACCTiC, this is how the individuals are predicted for the NCAA finals:
Besides the problem with Chelangat losing to Touhy at Paine, the lack of any Nutty performances factoring into the individuals 5k rating raised questions. That meet was closest to the NCAA field and Touhy won handily in the last k. She’s also undefeated against the best competition across the season.
“The steeper hills are at about 1500, 3500 and 4500 and before and after each there are downhills. I think the challenge is from 2000 (low point) to 3000 with the long uphill, repeated at 5000.”
What I did an inadequate job of describing (having been there this year) is a course that is essentially constructed in three sections — the long downhill-to-flat first ~1K; the ups and downs of the ~2d K, and the extended uphill of the third K. If one successfully addresses the second section, they are rewarded with long grind of the third before there is substantial recovery.
Tuohy is the only runner in the NCAA to beat Chelangat ever time they ran against each other in 2022 other than maybe Roe (I think Roe beat her twice, not sure if Roe ever lost to Chelangat in a 2022 race). NCAA indoors, outdoors, Joe Piane etc. It is certainly possible that Chelangat has gotten faster than Touhy over the summer; she is 25 or 26 and possibly nearing her peak, Touhy is only 20. Chelangat is also an excellent XC runner. But how likely is it?
LACCTiC is all based on relative time performances. If conditions affect everyone equally, it should handle them pretty well.
By the way, the scores have been changing around a bit this weekend. Sorry if thats been confusing for anyone - I've been trying to tweak some outlier removal to make the scoring system a bit more robust.
A bunch of runners now only have the conference championship score factoring for them now, at 100% (eg Van Camp, Markezich). What is the reason for that, where others have 2 or more? Thanks
I admire the effort. Not shading it at all. The work is appreciated. But compare the data we have in college with the speed rating data. In high school runners would be running every week. Thousands of comparisons. The same courses would be run year after year after year. A sense of how a course runs (a course par) would be obtained. Tactics would be straight forward. And I do not recall the variability in course length and conditions, with courses being constantly reconfigured. Over time Bill developed a very reliable rating. Yes it had the footlocker effect, but you had a good deal of confidence in the rating system.
In college runners race maybe 3-4 times total, often less. The courses change in conference and regionals every year. The race tactics can difer radically. It is just hard to get something like a SR in college distance running.
“The steeper hills are at about 1500, 3500 and 4500 and before and after each there are downhills. I think the challenge is from 2000 (low point) to 3000 with the long uphill, repeated at 5000.”
What I did an inadequate job of describing (having been there this year) is a course that is essentially constructed in three sections — the long downhill-to-flat first ~1K; the ups and downs of the ~2d K, and the extended uphill of the third K. If one successfully addresses the second section, they are rewarded with long grind of the third before there is substantial recovery.
I do believe the short steep hills into a long slow hill is part that gets overlooked...as you mention, a definite lack of recovery time for an extended period. At this point speed is largely irrelevant...it's all about the hill strength/endurance, and then who can recover the best to do it again.
Chelangat being 15:06 and Tuohy 15:19 only makes sense if you believe the Bama runners were not trying AT ALL at Piane, and Chelangat should have beaten Touhy beaten 12 seconds instead.
Touhy broke Kelati's 2019 course record by 12 seconds in that race, and that was the season Kelati won.
Cook didn't look as strong at Big 12 as she did in her opener... she might be out of the top five battle in my opinion. McCabe, Tuohy, and Valby remain undefeated thus far, with Valby and Tuohy having much wider margins of victory. If the race is bunched through 5k McCabe could spoil with her incredible kick, but I don't think Valby will leave it that late after seeing her dominate from the front at SECs. Roe always in it but has yet to have a win yet on the season. Chelangat looks good, but Tuohy and Valby might prove too strong. Chmiel needs more recognition as she is mixing it up with Tuohy off of a major injury, and could keep gaining fitness.
1&2. Tuohy/Valby tossup, but on the hilly OSU course I give it to Tuohy
LACCTiC is all based on relative time performances. If conditions affect everyone equally, it should handle them pretty well.
By the way, the scores have been changing around a bit this weekend. Sorry if thats been confusing for anyone - I've been trying to tweak some outlier removal to make the scoring system a bit more robust.
A bunch of runners now only have the conference championship score factoring for them now, at 100% (eg Van Camp, Markezich). What is the reason for that, where others have 2 or more? Thanks
The current system averages the best performance (i.e. "PR") with the most important performance (i.e. race with the highest importance rating). Importance ratings are scored based on the diversity of the runners (inter-regional is generally more important), and their score is double if they are conference, regionals, or nationals.
If your most important race and your best performance line up, then your score is just based on that performance.
Not sure if this is the best way to go, but its what I felt made sense.
“If conditions affect everyone equally, it should handle them pretty well.“
But your programmed assumption is that conditions (often weeks apart) are the same. No?
No, I don't assume the conditions are the same. If two races weeks apart on the same course run differently, people will run different times, which will mean the races will be scored differently.
LACCTiC does what you do when you are suspicious about a course. Maybe you see some times from Paul Short that don't make sense - maybe your rival ran 23:50. Say he ran 24:40 the week before in a race where you ran 24:35. So you think, "I probably would have run 23:45ish at Paul short." But of course your rival could have just tanked last week, so it's better to use a few runners as references. Do this for every single runner on a huge scale and you can see how you can start to score races. Then add track PRs as another race and adjust everything to that. Thats it (plus tons of tweaks to filter outliers).
“No, I don't assume the conditions are the same. If two races weeks apart on the same course run differently, people will run different times…”
Then how do you account for different conditions when it’s two (or more) different runners on the same course weeks apart? How do you know the course ‘runs differently’ and to what do you attribute those differences? Simply because the times are different? Maybe it’s earlier in the season (people don’t wanna peak in September); maybe the lead dog ran a different race strategy and everyone tagged along; maybe the weather was a factor; there are many potential reasons.
Istr I asked something like this before and my recollection is that you used time comparisons as a proxy for course degree of difficulty. So, similarly, I’m curious about this. I like your product but I wanna understand how the outcomes are derived.
“No, I don't assume the conditions are the same. If two races weeks apart on the same course run differently, people will run different times…”
Then how do you account for different conditions when it’s two (or more) different runners on the same course weeks apart? How do you know the course ‘runs differently’ and to what do you attribute those differences? Simply because the times are different? Maybe it’s earlier in the season (people don’t wanna peak in September); maybe the lead dog ran a different race strategy and everyone tagged along; maybe the weather was a factor; there are many potential reasons.
Istr I asked something like this before and my recollection is that you used time comparisons as a proxy for course degree of difficulty. So, similarly, I’m curious about this. I like your product but I wanna understand how the outcomes are derived.
As far as the algorithm is concerned those two races on the same course are different races and could run differently - its all based on how people perform relative to their ratings before the race. If not all runners are uniformly affected by something then we will have problems -- anything that affects all runners the same will be accounted for.
Because everything is based on relative times, if the entire NCAA gets 5% faster, lacctic sees the courses running 5% faster. So there is a bit of a drift across time: late season performances can be slightly harder to score high at since everyone is running better. This drift is pretty difficult to correct - I don't have any good ideas for it yet (other than maybe interpolating based on expected improvement from outdoor track to indoor track).
I note for all those who think the Oxford course was short, that there was an SEC Preview on t here same course some time before the Championship. ‘Short course’ doesn’t seemed to have been in-play for that event. Did they change it between events? Not likely. I also note that the further one goes down the list of finishers at the Championship, the less delta we see between previous bests and their Championship time; in fact, a couple of the Ole Miss finishers who ran in the Preview were actually slower in the Championship. Perhaps ‘the rabbit won the race’, and those closest to that rabbit were pulled along more rapidly in the chase.
I can't disagree with you on this...the course 'looks' short but who knows. Maybe, maybe not. I guess it really doesn't matter. In the end, she ran an incredibly fast time....3.05 pace for 6km. That is about the same pace she ran the 5000 m outdoors. So it is possible....especially since the course was almost dead flat, and conditions were good. Same for Mercy.
In the end....all it does is add a little drama and excitement for Nats!!
I note for all those who think the Oxford course was short, that there was an SEC Preview on t here same course some time before the Championship. ‘Short course’ doesn’t seemed to have been in-play for that event. Did they change it between events? Not likely. I also note that the further one goes down the list of finishers at the Championship, the less delta we see between previous bests and their Championship time; in fact, a couple of the Ole Miss finishers who ran in the Preview were actually slower in the Championship. Perhaps ‘the rabbit won the race’, and those closest to that rabbit were pulled along more rapidly in the chase.
That still doesn't explain how the splits showed the top 40 W finishers ALL accelerated a lot between 2.12 and 3 K and ALL then slowed a lot again. By ALL I mean each and everyone without exception. I stopped looking after top 40. Runners who had settled into 80-82 second 400 pace in the 2nd K suddenly ran at 76-78 for a 1K stretch and then went back to 82 or so. Assuming just the 2.12 or 3K split is off and nothing else does not fix it - just moves the problem to another split.
The top finisher at the Preview was a minute faster at the SEC race. The second place finisher about 30 seconds faster. The Preview was not a lot of runners.