Buffalo Bills analytics-based NFL Draft predictions: How have I done?

A look back at my history of data-driven draft picks

When you’re a writer creating content for the Buffalo Bills, it can be tough to craft meaningful content for draft season when you don’t watch college football. So seven years ago I landed on the idea to tap into my behavioral science background and run analytics on pre-draft time/energy committed by the front office to make predictions. While it’s been fun, in the past I’ve expressed skepticism that I’d ever hone in on the exact player at the top of the board.

Then Keon Coleman happened. Rather than take my victory lap and be done with it, I felt this was a good opportunity to look back at my entire history with the premise and see how I’ve done.

Methodology

I’d like to keep this as clean as possible, so what I’ve done below is gone back and pulled just my predictions out of each year’s articles and compared them to the actual draft class. Links to each article are included. For ranking sake, I’ll give one of these labels to every prediction:

• Accurate — My prediction came to fruition.
• Plausible — It’s possible this would have occurred if another team hadn’t screwed it up for Buffalo.
• Off the mark — The prediction did not occur.
• Kind of — I’ll try to avoid this, but this is for when the thing I predicted sort of occurred.

2018

The year that started the whole project! You might (correctly) guess that the idea began as an attempt to see if I could predict which quarterback the Bills would be taking. I had thoughts beyond that as well, so let’s see how I did.

Targeted positions

• Quarterback — Accurate (Josh Allen).
• Receiver — Accurate (Ray-Ray McCloud and Austin Proehl).
• Linebacker — Accurate (Tremaine Edmunds).
• Offensive Tackle — Off the mark (Wyatt Teller is close but not quite).
• Defensive end — Off the mark (Harrison Phillips is also close but not quite).
• Offensive skill positions — Kind of (they took two receivers which is mixed in with my earlier prediction).
• Slot corner — Accurate. It wasn’t in my bulleted list of positions, but in the article itself I noted the spike in corners and suggested that perhaps Buffalo wasn’t satisfied with Vontae Davis in the slot. They picked Taron Johnson, so maybe this can count as “Super Accurate.”

Specific players

• Rashaad Penny — Plausible. I didn’t specifically name him as a prediction, but the article noted he was a rarity in that Buffalo met with him more than once and could have been a target. The Bills selected Allen and Edmunds in Round 1 and likely wouldn’t have been able to trade back into the first round to select Penny (pick 27).
• Frank Ragnow — Plausible. This is similar to Penny, with Ragnow being taken 20th overall.
• Baker Mayfield — Off the mark. I could give myself a boost and note that he was gone long before Buffalo could have taken him, but there’s no reason to think general manager Brandon Beane and head coach Sean McDermott have been fibbing about Allen being their guy. Mayfield had the most visits with Buffalo in 2018 and maybe he was their Plan B if he fell.

2019

Brandon Beane’s second draft for the Buffalo Bills was a bit tougher to figure out a top-of-the-board target thanks to Josh Allen being added the year before. In addition to my main article linked above, we did a Buffalo Rumblings contributor article where I doubled down on my top target.

Targeted positions

• Wide receiver — Off the mark.
• Edge-rusher — Off the mark.
• Not a center or kicker — Accurate.
• Guards, DTs, CBs — Kind of. Ed Oliver was their first pick and they did draft a corner in Jaquan Johnson. I’m only giving myself partial credit because in no way did I emphasize any of these positions, let alone defensive tackle.

Specific players

• D.K Metcalf — Off the mark. Metcalf was available for Buffalo three times and the Bills passed. Interestingly, I remarked on how often they met with Ole Miss players. Dawson Knox doesn’t save this prediction for me though as I noted how few tight ends they had met with and felt this wasn’t a priority area.
• A.J. Brown — Off the mark. See above except Brown was only available for Buffalo’s first two picks.

2020

For this season and the next, I actually forgot to do this pre-draft so I didn’t have official predictions, but rather wrote an article on what I believe I would have said. I tried to stay true to the concept and was fine predicting answers I already knew were wrong. This was also the COVID year and data was wonky to say the least.

Targeted positions

• Wide receiver — Kind of. While Gabe Davis and Isaiah Hodgins were selected, I identified that analytics would have called this position the top target and Davis was Buffalo’s third pick.
• Defensive lineman (preference to defensive end) — Accurate (A.J. Epenesa).
• Running back — Accurate (Zack Moss).

Specific players

• Denzel Mims — Off the mark. Mims was taken a few spots after Epenesa. Buffalo could have taken Mims but didn’t.

2021

As noted, this was the other season when I did it after the fact. Draft visits for this year were also impacted by COVID. This year was notable as the usual sources of data had precisely zero Buffalo Bills draftees as pre-draft visits. Let’s check in anyway though.

Targeted positions

• Wide receiver — Off the mark. Buffalo took Marquez Stevenson in round six, but the data had this as a higher priority.
• Running back — Off the mark.
• Safety — Off the mark.

Specific players

• Asante Samuel, Jr. — Off the mark. Buffalo let Samuel pass right by.

2022

There were two big positives this season. I remembered to write the article before the draft, and there was better data than 2021, which was a terrible year for this exercise. Did more data lead to better results?

Targeted positions

• Corner — Accurate. I had a different corner as my target (see below) but I felt corner was the pick in Round 1 and Kaiir Elam made this come true.
• Tight end — Off the mark.
• Running back — Accurate. I felt a tight end or receiver would be the pick in Round 2, but James Cook was their guy making the above bullet off the mark, but this one spot on.
• Linebacker — Accurate, and Terrel Bernard seems to be a steal.
• Wide receiver — Kind of. I felt it would be earlier, but Khalil Shakir makes this mostly right just in Round 5.

Specific players

• Trent McDuffie — Plausible. McDuffie was taken by the Kansas City Chiefs just a couple spots ahead of Elam. It’s possible McDuffie was the target and Elam was Plan B.
• Jelani Woods — Off the mark. Buffalo could have taken Woods instead of Cook and passed.
• Rachaad White — Off the mark. The Bills rolled with Bernard two picks ahead of White and already had their running back selected with Cook.
• Chad Muma — Off the mark. Buffalo chose Cook over Muma. It’s possible they felt he may fall, but I can’t give myself credit for that.

2023

I don’t recall looking back at my picks from last year, so this should be a fun exercise for me too.

Targeted positions

• Wide receiver — Plausible. See below on specific players.
• Offensive weapon (some preference on TE) — Kind of. Dalton Kincaid was their first selection, but I felt they’d add two players between TE and RB and they didn’t.
• Defensive back (preference on safety) — Off the mark. While I could take credit for Alex Austin, I’m not ready to pat myself on the back for a seventh round flier.
• Offensive line — Accurate (O’Cyrus Torrence).
• Not a linebacker — Off the mark. Dorian Williams in the third round is a decent investment.

Specific players

• Jordan Addison (Plan A) — Plausible. Addison and the next two players seemed to have a lot of smoke for Buffalo but were gone during a wild run on receivers just before Buffalo picked. It’s possible Buffalo could have traded up for one of these guys, but the Bills had limited draft capital last season to move up with. We’ll never know if they preferred one of these guys to Kincaid or not.
• Zay Flowers (Plan B) — Plausible. See above.
• Jaxon Smith-Njigba (if he falls) — Plausible. See above.
• Jalin Hyatt (if he falls) — Off the mark. Hyatt was the rare receiver to fall in the 2023 draft. Buffalo had a couple chances to get him if they liked him but passed.

2024

And that brings us to now! We know I nailed one pick, but let’s see how well I did overall.

Targeted positions

• Wide receiver — Accurate (Keon Coleman).
• Defensive back (preference on corner) — Kind of. Cole Bishop was their second choice but I was leaning more toward corner, not safety.
• Defensive line — Accurate (DeWayne Carter).
• Overall a defensive skewed draft — Off the mark. Exactly half the draft was defense and half offense.

Specific players

• Keon Coleman — Accurate. I got one right finally!
• M.J. Devonshire — Off the mark. Buffalo literally had 10 chances to draft Devonshire, and passed on him. It’s possible they were hoping to land him as an undrafted free agent, but there was a ton of analytics smoke on Devonshire and Buffalo passed over-and-over.

The Verdict

I’m going to take a short victory lap on Keon Coleman because overall I still think draft volatility makes it very tough to pinpoint a specific player. The Dalton Kincaid year is a prime example. I scatter-gunned a number of wide receiver names and of the many tossed out, only one can objectively be called wrong.

“By position” I think has better accuracy, with overall good results in predicting the usage of draft picks. Going forward, I may include aggregate data for all years. I don’t know if you noticed while reading, but wide receiver was listed as a targeted position in every single year I’ve done this. Most years were labeled as “kind of” right, with only the most recent draft resulting in an “accurate” label.

For my final thoughts, I’d want to compare this to draft guru predictions, as I’m not convinced I fare any better than other ways to predict the draft. Ultimately, I think this concept is a good supplement to other sources and is at least worth the yearly write-up.