Cal Preps

[Crimson and Black]

Does anyone know, FACTUALLY, how reliable or good Cal Preps is with their formula for picking simply winners. Not really interested if their score differential is close, but how good are they at picking winners

 

[Spread All Day]

They say it’s supposed to be a lot more accurate after October 1 but I don’t know what formula they use

 

[Bless Em All]

I don't see it being that accurate.

 

[kingpin2010]

I remember a few years ago someone kept track of calpreps vs fantastic50 vs some other site that predicted games, and they all ended up around 75-80% on picks.

 

[BlackandGold]

From my experience, it has been fairly accurate.

 

[TroyTrojan05]

All I know is Vamps formula is the best

 

[Paladin]

I watched over several years and noticed with a couple of games under their belt they were hitting around 85% accuracy, getting better as the season rolled along. Pay attention to predicted close scores by one or two points - those were toss up games that could go either way and did.

 

[MoeBeturLojik]

I've followed calpreps for the last 15 years and do notice it is far more accurate for seeding and predicting playoff results than the Harbins. Given the algorithm, the more games played, the more accurate it gets. The things like key players out for injury, illness, suspension, etc are not something that feeds into the algorithm. The goal is to take all the games played to date, then minimize the difference between the predicted and actual margins of victory by adjusting team ratings up and down thru multiple iterations. i.e. team with a rating of 40 should beat a team with a rating of 30 by 10 pts. Then do that for every game played so far for all 16,000 or so schools.

 

[D4fan]

Vamps was about 93% accurate. As mentioned above injury is a wildcard that can not be factored into the analysis. Same for if a team plays first five games without their all-American qb.

Overall these computer driven rankings are far more accurate than even the most non biased eyeball test. We have bias as human beings and the computer does not, so it more accurately can predict ( with the exception of injury, available athletes etc.).

 

[vamp2syd]

I remember a few years ago someone kept track of calpreps vs fantastic50 vs some other site that predicted games, and they all ended up around 75-80% on picks.

That was me back when I did that stuff. Regular season anything over 80% is decent. My system, Cal-Preps, and Drew Pasteur would consistently be able to predict with a greater than 80%. It does lower a bit in the post season.