CFB Recruiting And Draft Evaluation Data

I spend alot of time on message boards discussing my beloved Ole Miss Rebels.

I stumbled across a question, "how well do recruiting evaluation agencies like 24/7 sports evaluate and project talent? And also being a data nerd, I figured this was the perfect question to lob into the ML and data science community.

The dataset here is an amalgam of recruiting player data from 2008-2022, NFL draft data from the same span, and player metrics from 2013-2022 (first date available). All credit to data creation and warehousing should go to Bill at CFB data.

Columns:

• ident = unique ID to player
• rank = player's recruit ranking in class per year
• name = player name (str)
• recruit_year = year the player was signed
• comp_recruit_rating = composite recruit rating (0.0-0.99)
• stars = player star rating (1-5)
• position = player's position on field
• school = player school/college
• draft_grade = assessment of NFL draft potential (continuous)
• draft_success = was player drafted in rounds 1, 2 or 3? (0, 1)
• countablePlays = usage rate, essentially how many plays each player contributed to over career
• averagePPA.all = an advanced metric, Predicted Points Added attempts to boil down how many points a player adds to team - average value over career
• totalPPA.all = same as above, but summed over player career

Let me know if you have any questions! Happy modeling!