My friends C. and C. recently raised the following question:

Do late-round NFL draft picks from __Power 5 schools__ tend to perform better than late-round picks from small schools?

To answer this, I used rvest and __Pro-Football-Reference__ to look at NFL Draft classes from 2000-2021. Using PFR's __"Weighted Approximate Value"__ statistic, I got the following hideous table:

This table shows the average Weighted AV ('value') of players picked in each round, separated by Power 5 players ('major'=1) and non-Power 5 players ('major'=0). 'N' is just the number of players in each category, i.e. there were 210 non-Power 5 picks in the 4th Round from 2000-2021.

It turns out that non-Power 5 players have actually outperformed their powerhouse peers in each of Rounds 4-7 (our arbitrary cutoff for "late rounds") over this time period (also arbitrary).

This difference, however, isn't deemed significant by t-tests* or models**, so the (boring, sorry!) conclusion here is that there's no discernible advantage to drafting late-round players from small schools.

*Because we're doing 4 separate t-tests (one for each round), we're running the risk of p-hacking, so the fact that nothing comes up as significant even without correcting for the number of tests we're conducting speaks to how weak the evidence is that there's a significant advantage.

**Neither a linear model nor a GAM with 'major' as an indicator variable seemed to think Power 5 status was a significant predictor of Weighted AV.

## Comentarios