Is LACCTiC Dead?

It was ok for teams but individual it’s way off.

Wasn’t the guy that did it a math grad student at WashU? I assumed that this was a project for his studies and that he has finished his studies and moved on.

He was a post doc at Harvard/MIT. PhD from Cal Tech.

He is now a professor at Dartmouth.

engineering.dartmouth.edu

Bijan Mazaheri

@thayerschool

Causal inference; data integration; mixture models; machine learning; biological datasets

Open link

Many teams have not opened up and/or rosters not updated on TFRRS.

Blanks was always ranked in the 20 -30 range yet he won it every year.

The challenge is only rating races which really reflect the athlete's fitness and somehow ignoring less than full efforts.

Someone get Flotrack to sponsor him.

College xc really doesn't provide the meaningful data points to make precise and accurate championship predictions, with teams running so few times per season and the top people probably only running all out once at the end of the season, so there was always going to be some fuzziness in even the brightest, well-schooled former runner's mathematical model.

Right, he did start weighing the results somewhat but still picking the right races can be somewhat arbitrary and probably difficult to apply for the thousands of runners to get the best assessment.

Hey guys. Its not dead. I'm going to start updating the website again this or next week. I just taught a large-ish class this summer and have a lot of grading to do.

In defense of the individual rankings, I do capture a simulation of the race before I load results in:

https://www.lacctic.com/races/10175

The arrows show how off the simulation was. Blanks was ranked 5th and got 1st. If you only care about who is going to win, I agree human rankings are likely more accurate because top runners rarely run all-out. However, LACCTiCs goal is to accurately rank the entire race, and the entirety of NCAA. Its pretty good at predicting finishing places of people in the middle pack, especially when you consider how dense the field is (a few seconds is 20-30 places). This is also why the team predictions are so good, because humans can rarely predict how a change in the depth of the field will affect a team's performance.

In the 2024 Cross-country season LAACTIC was uncannily accurate, as much as could be expected, and more, for the top Div I Women. This was proven out by the post-season results at BYU.

The accuracy was crazy. At the time, the program was saying Valby was in sub-15min shape. And then, boom, sure enough, Valby went sub-15 soon thereafter.

Will it still rate races that will have already happened after you’ve updated it?

Here is an update on this...

TFRRS has put up new security that is blocking my requests to scrape data. Until I find a way around it, LACCTiC cannot be updated. It appears that the owners of TFRRS are not happy that I was using "their" data

Update on the update. Right after I posted that I decided to see if AI could find a way around it. We are in the clear for now. Nobody say a word

When can we expect Lacctic to be updated? Thank you for all you do

Not sure if this makes me a commie or what, but I hate the idea that race results can be kept as trade secrets or IP. USATF should mandate that all race results be open source data and tell timing sites to find other ways to make their money - like offering compelling presentations and interpretations of the data.

Bumping this, it’s an amazing site

Now Meylan has a note up that he can’t easily pull race results from a lot of places around the country any more so he won’t be speed rating nearly as many races.

Swell. This sport can’t get out of its own way.

Just another reason to "celebrate" Flotrack

flotrack.org

Direct Athletics and TFRRS Extend Agreements with the NAIA and NJCAA - FloTrack

FloSports Staff

FloSports Secures Long-Term Extensions for DirectAthletics and TFRRS to Continue to Serve as Official Results and Rankings Platform for the NAIA and NJCAA.

Open link