• Follow us on Twitter @buckeyeplanet and @bp_recruiting, like us on Facebook! Enjoy a post or article, recommend it to others! BP is only as strong as its community, and we only promote by word of mouth, so share away!
  • Consider registering! Fewer and higher quality ads, no emails you don't want, access to all the forums, download game torrents, private messages, polls, Sportsbook, etc. Even if you just want to lurk, there are a lot of good reasons to register!

ESPN (A bunch of Death-Spiraling maroons)

It’s a simple least squares model based on what he’s explained to me. When I’ve called him out on it he says it ‘does what it is supposed to do’ and then references least squared error. That’s when I went after him on his choice of loss functions.

Anyhow, at a minimum, for early season game predictions to be of any value, he should be considering a Bayesian model where the prior is based on things like returning players, their stats, attendance at road games and other similar things. Picking the prior should not be overly difficult as one in that position has ample historical data to work with and can use a multitude of features from a prior year to predict the following year.

It used to be that things like a Gibbs Sampling / Markov Chain Monte Carlo model was a LOT of work to pull off. In ‘95 I was involved in writing a Bayes Multinomial Probit in a matrix algebra language (SAS/IML) and it sucked... there was no direct function for a Kronecker product if I recall correctly among other things... but now, with the general availability of every model under the sun with Python and R, there’s no excuse for shit models. If the results are not good, it just means the modeler is uninformed or fucking lazy. In this case, I think it may be a combination of the two.

I mean hell, he probably doesn’t even need to go that complex. I imagine that an XGBoost model that included some features to represent past year’s history and some type of feature to represent time (number of game played in current season) would create interactions that would fairly naturally weight early season predictions to the historical data and create a more accurate prediction in weeks 1-5.

Is it really just that? I see them referencing Neural Networks, some guy's R-based package, and fellating each other over acronym games trying to be Sabermetrics for football.
I'd prefer a model that made no attempt at 'preseason rank'. That goes for computer and human polling.

Or, kept completely separate from in-season, one that's trained on recent historical finishes, returning starters and positions, nfl draftee positions and draft # (good indication of talent leaving), transfers by position and recruiting rank, and recent recruiting by position and rank.
.... but the data entry and verification would be exhaustive. Scraping NFL Draft and a composite recruiting rank would be easy part. Figuring out who disappeared from rosters and/or transferred... less fun. Not an easy way to scrape every school's athletic site - as if the schools even keep their own shit updated. Basically end up with a database of 85+ players across 130 teams...
Not worth a preseason rank imo
But would be interesting to see how much discount ML figures out for a 4* transfer vs fresh 4* recruit on both a season-by-season and over cumalative seasons.
 
Last edited:
Upvote 0
YearlyWatchfulBellsnake-small.gif
 
Upvote 0
Is it really just that? I see them referencing Neural Networks, some guy's R-based package, and fellating each other over acronym games trying to be Sabermetrics for football.
I'd prefer a model that made no attempt at 'preseason rank'. That goes for computer and human polling.

Or, kept completely separate from in-season, one that's trained on recent historical finishes, returning starters and positions, nfl draftee positions and draft # (good indication of talent leaving), transfers by position and recruiting rank, and recent recruiting by position and rank.
.... but the data entry and verification would be exhaustive. Scraping NFL Draft and a composite recruiting rank would be easy part. Figuring out who disappeared from rosters and/or transferred... less fun. Not an easy way to scrape every school's athletic site - as if the schools even keep their own shit updated. Basically end up with a database of 85+ players across 130 teams...
Not worth a preseason rank imo
But would be interesting to see how much discount ML figures out for a 4* transfer vs fresh 4* recruit on both a season-by-season and over cumalative seasons.
But espn would use their own recruiting rankings, which I believe they intentionally skew to the SEC/ACC schools. All of the metrics which they use are biased to benefit the schools they're invested in, and they use their own biased systems to feed bias into their other biased systems.

I've seen it for years now. I avoid using QBR because they created it, and obviously their FPI, player impact ratings, and Playoff predictor numbers are just crap trying to feed their narrative.
 
Upvote 0
Back
Top