📊Regional Bias in CFB Recruit Rankings

▫️NE most underrated region
▫️South & West appropriately ranked
▫️Rural + Minor Metro consistently underrated
▫️BIG adjustment for value:
▫️#300 Rk Rural South recruit = #100 rk in Exp Value
▫️#100 Rk NE Minor Metro = Borderline 5 Star

The map view of this is interesting. Generally appears that Southern Urban Areas are Overrated. Conversely, Urban Areas up North are Underrated.

📊Over/Under-Rated Metro Area HS Recruits

▫️Miami somehow underrated recruiting territory
▫️LA & SF: not THAT different
▫️Philly one of the most underappreciated recruiting territories in the US
▫️Birmingham = Saban Effect?
▫️Non South-FL is rough (JAX - ORL - TPA - NP)

👀

📊Predictiveness of Official Visits pt 2

Recruiting Services Underrate Recruits who:
▫️Take 4 OVs
▫️Commit to teams that performed well in FPI the previous year (or coach & development effect!)
▫️Visited many high-FPI teams

📊Official Visit Data Analytics

▫️Recruit Rank appears to already account for OVs (flat DVOE)
▫️OV Quality calc'd by Bradley-Terry methods
▫️4 OVs could be signal, but will need to look at more data
▫️OV Quality metrics are flat
▫️TLDR: don't use OVs to adjust recruit ranks

Working on developing my own HS recruit rankings using ML methods

Draft Probability predictions for 2019 Left and 2024 Right

This is version 0.1, but overall like where we're going

Working on a ML model to predict future draft value for HS recruits

It’s a tall order since the draft is a 1/15 filter and the DV curve is exponential

But we know there is predictive info (like pos normalized ht/wt!)

Feature engineering is gonna be key here

📊UF 2023 Roster Vizualization

DEF
▫️Serious DL weight
▫️Great size on FR K. James & Collins
▫️Put Weston @ EDGE /s
▫️T. Mitchell is a dude

OFF
▫️Love the OL WT & HT distribution
▫️Hansen is gonna murder a DB
▫️Savaiinaea & Zip don't have the body types
▫️Nice blend of WR sizes

Insights from SIS college-level "Defenders in the Box" data:

▫️Results are intuitive
▫️4-5 man boxes limit pass game, but are vulnerable to run
▫️7-9 man boxes limit run game, but are vulnerable to the pass
▫️6 man boxes perform well vs both run & pass (goldilocks)

The first part of my data dive into the CFB transfer portal is up at @cfbnumbers.bsky.social

Most results are intuitive & the portal largely seems to function to help players find an appropriate level of competition

https://cfbnumbers.substack.com/p/statistical-properties-of-the-cfb?utm_source=url

This is probably expected, but I've never seen it visualized:

there is a clear relative age bias in highly rated HS football recruits.

📊What Factors Predict CFB Recruiting Success

▫️Part 1: Talent Availability
▫️Roughly 30% of recruiting can be explained by Location
▫️200 mi found to be most-predictive distance for Top 300 players
▫️No surprise Bama/UGA have recruiting success
▫️Deviations discussed in Part 2

My home for the next two weeks… still less toxic than a 247 board

Billy Napier's Roster Makeover

- Napier Recruits/Transfers make up 56 of 83 roster scholarships (67%)
-Napier Guys make up 204 of 265 (77%) remaining eligibility years
-Mullen Guys make up majority of players w/ 1-2 years of eligibility (17/27; 63%). Next year will be ~50/50