2023 Draft Prospects: Data-Driven Deep Dive

[Out of the Rafters at the Q]

So, I had a gap between meetings and wasn't dragged around by the kids. I decided to start dumping the edge prospects into a mySQL database to play around with numbers like age, RAS, 10-yard split, PFF grades, etc.



I left out the guys who obviously aren't going to be there. Started at Will McDonald, who is #25 on PFF's big board.

If you have any questions, like "Hey, can you show me all the guys under 22 who have a RAS over 9.00 and PFF Pass Rush grades over 85?" feel free to post them here and I can spit out the query.

If you have other positions you want to see, just start building it out in Excel or Google Docs and people can chip in to help. It's pretty easy. Here's what the Excel Spreadsheet I built for the edge class looks like (and from there it's super simple to import into mySQL)

'

If you prefer to write it out in text, comma-delimited is my preferred format.

Also, if we want to look at other positions, we can discuss the metrics we want in the database. Perhaps for DT's, we want to see bench reps. Maybe for WR's we want to see hand size or receiving yards. That sort of thing.

I can also probably give access to a few trusted people. If you're interested in having access to the DB, PM me.

First order of business, here are all the edge players under 23 years of age with a 10-yard split under 1.65 seconds sorted by PFF's True Pass Set Pass Rush Grade.

 

[Out of the Rafters at the Q]

I'm moving on to DT's. Added some more info. Obviously still working on filling out the blank columns. Let me know what you think:

 

[Out of the Rafters at the Q]

Anyone know where to find an age for Jonah Tavai, Robert Cooper or Dion Bergan?



One thing stands out for me. Jalen Redmond. His RAS is 7.86.... because he's listed as a DE. If you change him to a DT, his RAS jumps all the way up to 9.72! That's good for second in the entire class behind only Keion White.

 

[Out of the Rafters at the Q]

Looks like Moro Ojomo is a good bet for pick 74 if we go DT



While Gervon Dexter does have the higher RAS, look at those PFF grades. They aren't even close. 84.1 to 55.2 in pass rush grade?

If we value the PFF grades, we probably have Ojomo solidly ahead of Dexter.

 

[Out of the Rafters at the Q]

I decided to do WR next.

I also decided that there's no point in selecting what data I think is useful--might as well put it all in, including Reception Perception for the receivers.

Here's what that monster is going to look like.

 

[Out of the Rafters at the Q]

Well, here's what the WR sheet looks like now:



I have to add columns for basic stats, like receptions, targets, yards, etc.

I also realized that all those Reception Perception columns to the right of the filled-in RAS ones? Yeah, that's just one of like six pages they have. I'm gonna have to build something to scrape data.

What started out this morning as a "Hey, I should put this stuff into an excel sheet so I can start guessing who the Browns will take" has now turned into a "I want to compile this year-over-year to try and analyze trends--both in terms of what the Browns like to draft and in terms of what leads to NFL success.

 

[Out of the Rafters at the Q]

I've decided to grab everything I can for every prospect, and make this a yearly thing--maybe build up some models over time.

Right now I finished my python scripts to scrape all of RAS and dump it into mysql.



Nearly 2,000 rows of goodies from scraping RAS.

Next up is scraping PFF.

After that, I'd like to scrape Reception Perception for the WR's.

I'm also going to have to pull some basic counting stats for different positions. I'm honestly not sure where the best place to go for that info is.

If there's anything you want a drill-down perspective on, feel free to toss it in this thread. Also, if there's something I'm forgetting about that would be valuable, let me know.

 

[natedagg]

I've decided to grab everything I can for every prospect, and make this a yearly thing--maybe build up some models over time.

Right now I finished my python scripts to scrape all of RAS and dump it into mysql.

View attachment 15127

Nearly 2,000 rows of goodies from scraping RAS.

Next up is scraping PFF.

After that, I'd like to scrape Reception Perception for the WR's.

I'm also going to have to pull some basic counting stats for different positions. I'm honestly not sure where the best place to go for that info is.

If there's anything you want a drill-down perspective on, feel free to toss it in this thread. Also, if there's something I'm forgetting about that would be valuable, let me know.

Can you load it into GPT somehow and ask for best prospects or regression equations or just random shit?

This is the heat btw. Much respek for all of this. I haven’t played around but love what you are doing! Apply to the Browns!

 

[Out of the Rafters at the Q]

Can you load it into GPT somehow and ask for best prospects or regression equations or just random shit?

This is the heat btw. Much respek for all of this. I haven’t played around but love what you are doing! Apply to the Browns!

I'm not a developer.

I wrote a single Python script 3 or 4 years ago. That's the extent of my Python exposure.

I wrote 7 Python scripts where if you execute the top-level one, it will scrape RAS for every single player from every single position then insert it into mysql.

I've been leveraging ChatGPT the entire way. I definitely couldn't have done this in a couple hours after everyone else goes to bed if I was on my own.

I don't think ChatGPT is the right tool for your request. ChatGPT was programmed on data and spits out answers based on that.

What you want is to train an AI on this specific data and then have it spit out an answer.

Hopefully, this dataset can get there. If I start adding NFL performance to each draft prospect, you can very quickly start to see what data points translate to NFL success.

Or, if you just wanted to chart the players the Browns drafted, you could get a pretty good predictive model of who the Browns would be targeting in the next draft.

(Also, thanks for making this a thread and not just one person talking to himself )