As the draft quickly approaches, even the casual NFL fans have familiarized themselves with the top prospects. Mock drafts, like this awesome one, prepare us for the various possibilities that might occur. Among the commentary I’ve found especially interesting over the past few years is the comparison of specific prospects to past or current players.
“Prospect X is very similar to Player Y,” says Random Draft Analyst Z.
I think there’s some value in comparing prospects to past and current players, but it can be done better. Mock Draftable has a great database and player comparison tool based on each prospect’s Scouting Combine and Pro Day workout results since 1999. For Eric Ebron, we get something that looks like this:
This is great because the list of most similar players includes some good players. It’s a more accurate version of the same comparison, giving us an idea of how the prospect might project to be in the NFL. If a player is really similar to a few great past or current players, then it’s rational to think that they’ll be pretty good themselves.
That kind of comparison is interesting, but workout measurable and times are just a small part of player evaluation. What if we include college receiving statistics to the comparisons?
I compiled the college receiving statistics of all players drafted since 2006 and combined them with the Combine and Pro Day workout results to get percentile ranks of those players (like Mock Draftable does in their comparisons). I then applied the various percentile ranks of this year’s prospects to that dataset to find the most similar players.
The categories to determine similarity were height, weight, 40-yard dash time, vertical jump, broad jump, 20-yard short shuttle, college receptions, college receiving yards, college yards per reception, and college receiving touchdowns. The categories were chosen because each of the measurable statistics (Combine and Pro Day numbers) were available for over 75% of historical prospects and standard college receiving statistics were relatively easy to track down (with some exceptions of certain non-FBS players). If no data was available for a given category, that category was not included in the similarity test.
To take it a step further, I wanted to see how the similarity to certain players might impact a prospect’s chances of becoming a starter. I used the framework of the Bayes Theorem to estimate the probability that each prospect would be a starter given the start rate of the ten most similar drafted players. (Ten was an arbitrary choice, but I figured it was a good start.)
The results were interesting and can be found below. Keep in mind that of the 4,871 games that the tight ends drafted since 2006 have played, 54% of them have been starts. The chances that a tight end drafted since 2006 goes on to become a primary starter (regardless of round chosen) 33% of the time. A prospect worth drafting would have an estimated probability to start greater than that 33%.
The order is probably a little off from what you were thinking it should be, but keep in mind these are based on the start rate of similar players. This ranking does not include any film study, interviews, or other important aspects of the evaluation. It does, however, give use a good idea about a prospect’s strengths and weaknesses.
For example, Eric Ebron caught a lot of passes for a lot of yards but just an average number of touchdowns. Ebron’s Combine and Pro Day workouts were a bit inconsistent (fast 40 yard dash and good broad jump but a weak vertical leap and no short shuttle workout). Ebron also is a little smaller than the average drafted tight end at 6’4″ and 254 pounds. The key here is that the similarities can point to how a player can fit into an offense and how they could be best utilized.
Let’s look deeper into the top five and their similarities. The accompanying graphs compare the prospect’s data points (black fill) to their five most similar drafted players.
Marcel Jensen: Fresno State
Similarities (in order): Leonard Pope, Anthony McCoy, Kyle Rudolph, Greg Olsen, Anthony Fasano, Brandon Pettigrew, Joe Klopfenstein, Rob Gronkowski, Martellus Bennett, Marcedes Lewis
Jace Amaro: Texas Tech
Similarities (in order): Rob Gronkowski, Brandon Pettigrew, Luke Stocker, Clark Harris, Tony Scheffler, Dennis Pitta, Anthony Fasano, Jermaine Gresham, Kyle Rudolph, Rob Housler
C.J. Fiedorowicz: Iowa
Similarities (in order): Luke Stocker, John Phillips, Brandon Pettigrew, Michael Egnew, Martin Rucker, Kyle Rudolph, Tony Scheffler, Tom Santi, Brent Celek, John Carlson
Eric Ebron: North Carolina
Similarities (in order): Fred Davis, Cornelius Ingram, Rob Housler, Ed Dickson, Jared Cook, Gary Barnidge, Rob Gronkowski, Vernon Davis, Coby Fleener, Luke Stocker
Crockett Gillmore: Colorado State
Similarities (in order): Kyle Rudolph, Zach Miller, Anthony Fasano, Brandon Pettigrew, Luke Stocker, Rob Gronkowski, Martin Rucker, Tony Scheffler, John Carlson, Tyler Ecker
Another highly touted tight end prospect, Austin Sefarian-Jenkins, just missed the top five cut mostly because he was unable to participate in Combine and Pro Day workouts due to injury. Assuming average workout results, the Washington product would land just behind Ebron. He is conventionally sized and had elite production in his college career. Here’s how he compares, without workout results assumed.
This is just another way to compare and analyze prospects before the draft and it’s a work in progress. Whic prospect best fits into the Bills’ plans at tight end and how will the team use that position in 2014?
Wide receivers are coming next.