Collegiate cross country computer rankings and simulations (LACCTiC)

@Washed Up Grad Student. Thanks Bijan, have been following your work on this for a few years. I've also done some of my own tfrrs scraping using Beautiful Soup in Python. Just curious, I've heard that tfrrs can get very territorial about scraping their website, I've even seen a cease and desist letter before (Colin Carroll had a bad experience, see link below). Did you get a license from tfrrs to use their data or is this the wild west at this point? I would definitely not be a whistleblower, I love this stuff and think the data should be widely accessible to all. They should even have an API to make it easier IMO.

oops forgot to post this

https://colindcarroll.com/project/cross_country/

Is there a way to go back see progress over time? I'd love to keep track of our conference ranking week-to-week as the season progresses with the culmination at the conference/region meet.

Also, no Oregon showing on the D1 men's rankings?

The Big West has a couple of teams transitioning into Div I from DII. You picked up CSU Bakersfield but missed UCSD.

I follow two of the top BigTen women's teams closely. One of their runners posted on her Facebook page months that she is a grad student at Colorado. I assume she may be running for them now. Another runner you listed highly on the same team did not run for the school last season as she graduated and is no longer at any school. Another runner a little further down on the list is no longer at the school. I assume she graduated. She is now living and running on her own in a southern state according to her Strava activities.

I appreciate your work and understand you cannot keep track of all the runners regarding their status but just looking at this one team with a #3 and #7 runner from your list no longer on this top 10 national team will affect the comparison to others teams. I suspect there are top athletes listed on other teams who are no longer on the teams.

If you tell me who I can try to fix it.

Earlier in this thread people were complaining about seniors being not listed, so I went with a more conservative guess - essentially leaving everyone in. My plan is to set all seniors who have not competed by the end of next weekend to "inactive."

-Bijan

Just wanted to give you a big thumbs up for this. I have done similar analysis, albeit on a much, much simpler and smaller basis for my wife who is a high school coach. I found that normalizing the data really helped the kids gauge their improvement better. Too often kids were getting 'PRs' on courses that were obviously fast or upset for running slow on tough courses.

This is great stuff. If I were to make one suggestion it would be to have the ability to plot data graphically. It's an easier way for an athlete / coach to see trends in the data.

I have updated the weighting method on the website. The new system attempts to use gaussian mixture models to detect "tempo" or "easy" races and remove them. For example, Conner Mantz's recent race is now detected as a tempo and no longer counted.

Keep the feedback coming!

So far, this change seems to give somewhat strange results for some runners. Henry Jaques in particular stands out in D3, with 100% weight given to a great 5000m time in 2020. Cooper Teare and Alex Masai stand out in D1 for the same reason. If you're separating tempo/all-out efforts into two or more groups, maybe outlier performances are being treated as the latter, with typical races labelled as tempos?

Thanks for pointing that out. I was playing around with things a bit still and noticed the same thing you did. I think the results are a bit more reasonable now (only considering xc races when removing outliers).

Not sure how you track transfers, but Connor Weaver on the Utah State roster is a transfer from BYU. I’d think you could link the records and mover his times.

Same with Simone Plourde at Utah. Transfer from BYU.

This is great, incredibly rapid, and useful, but also inscrutable. It would be useful to include some more information to contextualize the numbers.

Am I misreading this in thinking that if you are relatively better at 8k/10k than 5000m, your TIC 5000m equivalent score will be lower than your actual best 5000m time? I'm looking at Nico Young's page. It has his best 5000m time of 13:24 on the track but then xc scores of TIC equivalents in the 13:10s (13:15 for NCAA xc champs and 13:18 for an in-season meet).

I handle transfers based on results - if someone runs a result under a new team it is updated. If a new profile was created for them, this is a bit more difficult. I fixed that were mentioned.

It is correct to read faster 5k conversions as being "better" at xc. A TiC equivalent is what we would expect someone's 5k PB from the pervious season to be based on what they ran at a race. So if your PB is 15:00 but you run a TiC equivalent of 14:40, you're finishing the way a 14:40 guy would have finished.

Does it count for short courses? I ran at the Stump Invitational last week and although it was cool to see a lot of fast times, the 8k ended up being around ~300m short. The 6k was also like ~200m short. My 5k prediction is bonkers because of it, which is cool, but I don't think I can do that on the track right now.

hmmmmmmmmm wrote:

Does it count for short courses? I ran at the Stump Invitational last week and although it was cool to see a lot of fast times, the 8k ended up being around ~300m short. The 6k was also like ~200m short. My 5k prediction is bonkers because of it, which is cool, but I don't think I can do that on the track right now.

Most of the conversions at the Stump Invitational seem to be around what people have run before - maybe a bit on the quicker side. The winner "broke 23" but it converted to around 14:01.

The method is distance-independent. It compares you to the people around you and what they have run on other courses/track PRs. This can be unstable if the race doesn't have a ton of previously evaluated runners, or a large group run slower than usual.

Have you raced a 5k in the new shoes? I know everyone is sick of hearing it but there is about a 12-15 second difference between 5ks in 2020 and 2021... You can see it in the TiC conversions.

Do you mean vaporfly or dragonfly? My school is too poor to afford those lol. Plus they like us to be in adidas if we can. My teammate ran 14:34 last year and the TiC thing gave him a 14:17 or something for the meet. That is more than just a small difference. We were both more fit in the spring than right now currently. Do the shoes even give you an advantage on the grass?

I don't know if they help in grass, but my point is the times are being converted to what the average person would have run in the previous season.

If runner A ran a 15:00 5k in normal shoes and runner B ran a 5k in 14:48 in dragonflys, they are similarly fit. If they then go both tie in XC, both times will be given a TiC of around 14:48 (if most peoples PRs are in dragonflys), but the person with the 15:00 PB might be confused.

I'll look more deeply into the score of that meet. I'm guessing its probably biased a bit by not having enough comparisons and will correct itself in a week or so (this happened last week as well for a few races). This should get better as the season goes on.

After looking into it, the race is a bit strange:

William Fallini-Haas has a pr of 14:41, but he beat a few very good people:

1) Ed Kiolbasa - 14:08

2) Daniel Maneloveg: 14:17

3) Brian Shulz: 14:41, but William beat him by over 30 seconds

The trend seems to continue down the list. Maybe you and your friend were the only ones running hard and everyone else was chilling - or maybe you guys are fitter than you realize?