nflverse-data VS nflfastR

nah, i scraped pbp data going back to 1999 and compiled the data myself

It took way too long for me to find this when I did a similar exercise, but if you ever decide to include snap counts, I'd start here. https://github.com/nflverse/nflverse-data/releases/tag/snap_counts

I just opened up play-by-play data, and while I just looked at regular season data (some of these guys weren't in POs, so it doesn't matter), and only had 6 guys with 50+ catches of 2 or less air yards - including McCaffery, who's not on here. So....I'm a but confused. But whatever, I lowered the requirement to 30 receptions (to include everyone who got mentioned here), and here's the resulting table, sorted by Success Rate on such receptions:

PBP participation has exactly what you are looking for.

It might not be the exact same EPA model, but nflfastR EPA data is free on github: https://github.com/nflverse/nflverse-data/releases/tag/pbp

also check out the various other things available here: https://github.com/nflverse/nflverse-data/releases

Check out this repo, I'm pretty sure you can find what you need for data: https://github.com/nflverse/nflverse-data

nflfastr has an extensive library of PlayByPlay, Roster, and Gamelog data that you can access programmatically via R.

For more on all this, here's the source of my data: {nflfastR}

Source: {nflfastR} play-by-play and win probability models.

It's been superseded by nflfastR at least for the play by play stuff. You can also download the data directly, rather than using the scraping functions. https://github.com/nflverse/nfldata

nflfastR is a database with play by play data going back to 1999, it has everything you'd need to create such a metric, but you gotta write the query yourself :)

Data Source: nflfastr library: https://github.com/mrcaseb/nflfastR