Projecting NFL Stars Based on College Stats

Written By: Toby Junker

    With the NFL season just wrapping up, we, as football fans, shift our focus from guys who have dominated the league over the course of the past 5 months, to guys who could be dominating it 5 years from now. It may seem far off at the moment, but in just over 3 months, NFL Commissioner Roger Goodell will be standing at the podium announcing the 1st pick of the 2019 NFL Draft as the best and brightest prospects from college football join the pro ranks.

    Between now and then there will be endless hype that builds for some prospects while anonymous scouts take shots at other guys. While the East-West Shrine game and Senior Bowl have already occured, teams still have the NFL combine, pro days, and personnel interviews to try and figure out which players to select come late April. Teams must assess how they view these draft prospects, who fits in with their scheme and who projects well at the next level.

    Figuring out who to draft has been one of the most difficult things for NFL franchises. One way that I thought it would be interesting to try and tackle this conundrum is by looking at the statistical careers of collegiate players, find former collegiate players that have had similar builds and careers and projecting the former players NFL career as the current players.

    Last spring I took “Fundamentals of Programing and Computer Science,” the introductory computer science course offered at Carnegie Mellon. For my end-of-semester term project, I built a program that attempts to do exactly this: project NFL careers by similarities in college careers. The program begins by prompting the user to select a position group:  QB, RB, WR, TE, DL, LB, or DB (offensive line was excluded as it was too difficult to acquire a variety of statistics).

The program then begins to ask for basic information about a prospect so that it doesn’t project someone with a completely different build or profile. It first asks the projected round a prospect is supposed to be drafted in. Any final projected must have been drafted in a round +/- 1 round of the prospect’s projection. Next the program prompts the user to input a height and weight for the prospect. Players who are more the 10 cm or 10 kg away from the prospect are also eliminated for the final projection. The final piece of basic information that is needed is whether the prospect played in a major or minor conference.

At this point the program asks for three different stats from the prospects college career; these vary based on position group. For instance, if you are looking at a QB prospect it will ask you for completion percentage, TD:INT ratio, and games played.

For RB’s the stats are carries per game, yards per attempt, touchdowns per game. WR’s and TE’s are catches per game, yards per catch, and touchdowns per catch. DL and LB’s are tackles per game, sacks per game, and forced fumbles per game. DB’s are tackles per game, INT’s per game, and forced fumbles per game.

For each position there was a weighting of stats according to what weighting matched the most players to their actual NFL projection. The program then compares the stats of the prospect you entered against every player who’s played college football to find the most similar statistical career. It then projects the player who had the most similar college career as the prospect’s expected NFL projection, projects the player who’s NFL stats are most similar to 1.1 times the expected stats as the prospect’s ceiling, and projects the player who’s NFL stats are most similar to 0.9 times the expected stats as the prospect’s floor. In the hypothetical instance from the example screenshots above, the program returned Joe Theismann as the expected player, Carson Palmer as the ceiling, and Daryle Lamonica as the floor.

I used the beautifulSoup package in python to web scrape all of the statistics for this project. All statistics are current up until the end of the 2017-2018 NFL season. All college football stats are courtesy of and all NFL stats are courtesy of If you have more questions about how the program works feel free to check out this video demonstration ( or drop a comment down below!

    Now, returning to this year’s NFL draft, I thought it would be a fun exercise to see how some of this year’s top prospects from each position group do in my program. For players that didn’t get much of a chance to play early on in their career, I used their more recent seasons to project their NFL outlook.

Dwayne Haskins, QB, Ohio State

    Up first we have the Ohio State product Dwayne Haskins. Many draft experts have him as the top ranked QB in this years class and as such he’s projected to go in the top 10 picks in almost every mock draft. Haskins had one of the most favorable NFL projections as his expected outcome is Cam Newton, his ceiling is Kurt Warner, and his floor is Chris Miller. While NFL fans would be sorely disappointed in Haskins falls to his floor, if he were to be as good as Cam Newton or Kurt Warner I’m sure whatever fanbase he ends up with would be very pleased.

Joshua Jacobs, RB, Alabama (Jr year stats)

    Second on our list of prospects is Josh Jacobs. The Alabama running back played a large role in the Crimson Tide returning to yet another college football championship and is the only running back in this year’s draft that is currently projected to go in the first round. Unfortunately for those RB needy teams, my program didn’t love Josh Jacobs outlook as it projected Roland Hooks as his NFL comparison, Warren McVea as his ceiling, and Jeff Smith as his floor.

AJ Brown, WR, Ole Miss

    The WR that I decided to project out was AJ Brown. Seeing as their hasn’t been a clear WR 1 that has emerged in this draft class, I went with a guy that seemed to be pretty high up in everyone’s rankings. Brown had a prolific college career at Ole Miss and drew a comparison to one of the most talented WR’s in the NFL. His NFL career projection is Sammy Watkins with a ceiling of John Jefferson and a floor of Oronde Gadsden. While many have been disappointed with Watkins career thus far, Brown could be a second coming of the talented WR.

TJ Hockenson, TE, Iowa

    Rounding out our offensive prospects we have this years top TE TJ Hockenson. Hockenson took a big statistical leap this past season as he doubled his TD production and over doubled his receptions and yards from the previous season. Unfortunately for Hockenson, my program didn’t seem to be impressed as it projected Gavin Escobar for his expected career and ceiling with Chip Glass as his floor.

Nick Bosa, DL, Ohio State

    First up on the defensive side of the ball we have projected #1 overall pick Nick Bosa. Bosa is hailed as one of the best DL prospects, someone that can completely change any defensive he’s put into. Many in the real world are comparing him to older brother Joey Bosa (who plays for the Los Angeles Chargers) but my program was pretty stuck on a different (less impressive) comparison. It was the only time where my program projected the same player for floor, expected, and ceiling. Here’s hoping Nick Bosa outperforms Jarvis Moss.

Devin White, LB, LSU (So & Jr stats)

    Next up on defense we have perhaps the prospect who received the most favorable comparison, Devin White. The LSU stud had a projected floor of Jerrell Freeman with an expected career and ceiling of Patrick Willis. If White comes anywhere near the level of Willis, whatever team that ends up with him will have their middle linebacker problem solved for years to come. And even if he only reaches the heights of Jerrell Freeman, his team will have a reliable and dependable MLB for the next decade.

Greedy Williams, CB, LSU

    Rounding out our projections we have top ranked CB Greedy Williams. The LSU standout had an amazing Freshman season and although his encore wasn’t as impressive, he still felt comfortable enough to enter the NFL draft. My program isn’t too sure what to make of Williams as it projects Nate Allen as his NFL comparison with Lance Schulters as his ceiling and Marcus Coleman as his floor.

    Wrapping it up, don’t worry if your favorite teams projected draft pick doesn’t have the career trajectory that you were hoping for. There are a lot of things that go into the success a prospect can have a the next level that aren’t taken into account here. However, if you think my program managed to make a really good comparison for any of the guys above (or I guess if you vehemently disagree with one of the comparisons) be sure to get down in the comments below and let me know your thoughts!

One thought on “Projecting NFL Stars Based on College Stats

Leave a Reply