SHARK Rookie Model

Want to know what the single best way is to measure the value of NFL Draft prospects? Too bad. There isn’t one. And that’s exactly why we built the SHARK Model.

SHARK stands for all the components that go into our prospect grading … 

• Scouting: Film grades from several trusted, experienced sources.
• Historical production: What the player actually did on the field in college, adjusted for situation and competition level.
• Athleticism: The player’s Combine (or other pre-draft event) testing, compared with the historical data at his position.
• Relative draft capital: How early you get drafted matters historically. So we weigh that, relative to position.
• Kappa: This is the statistical measure of agreement between multiple factors. Our Agreement Factor measures how well a player’s inputs agree with each other. High vs. low isn’t necessarily good/bad in this case but rather a signal that one or more categories disagrees with the others.

Those components create a SHARK Model score for each prospect, which we can use to compare players by position, across positions, and even across draft classes.

Those Model scores don’t necessarily determine the order of our rookie rankings. But they play a key role in building those rankings.

About SHARK Model

Want to know what the single best way is to measure the value of NFL Draft prospects? Too bad. There isn’t one. And that’s exactly why we built the SHARK Model.

SHARK stands for all the components that go into our prospect grading … 

• Scouting: Film grades from several trusted, experienced sources.
• Historical production: What the player actually did on the field in college, adjusted for situation and competition level.
• Athleticism: The player’s Combine (or other pre-draft event) testing, compared with the historical data at his position.
• Relative draft capital: How early you get drafted matters historically. So we weigh that, relative to position.
• Kappa: This is the statistical measure of agreement between multiple factors. Our Agreement Factor measures how well a player’s inputs agree with each other. High vs. low isn’t necessarily good/bad in this case but rather a signal that one or more categories disagrees with the others.

Those components create a SHARK Model score for each prospect, which we can use to compare players by position, across positions, and even across draft classes.

Those Model scores don’t necessarily determine the order of our rookie rankings. But they play a key role in building those rankings.