My fascinating study of the NFL draft

[cdcox]

All the trade down talk got me thinking if there was a objective way to analyze this. So I came up with a methodology to assess the value of picks.

I basically looked at every draft from 2007 to 1993 (15 years worth of data). I considered picks 1, 5, 10, 15, 28 and 29. To rate the players, I used Pro-Football Reference's Approximate Value, which is an attempt to rate a player's contribution to his team. I compared approximate value per year for each pick: the higher the value the better the picks turned out to be. I then calculated the ratio of (draft value points)/(approximate value per year). The lower this ratio, the better the value of the pick. Here are the results (first entry is pick, second is draft value points, third is approximate value per year, fourth is draft value points required per unit of average value per year).


Pick Draft Value Points AVPY DVP/AVPY
1 3000 8.7 343
5 1700 7.5 228
10 1300 6.0 216
15 1050 5.4 193
28 660 6.2 105
29 640 5.5 115


For picks 1 through 15, there is a trend where the players drafted earlier ended up having better careers. But at pick 28 a funning thing happened in which players picked at 28 were better than those picked at 15 and those picked at 10. Looking at the list of players, Trevor Pryce and Derrick Brooks were two really good players picked at 28. But even removing those players, would only drop the AVPY to 4.75. So I analyzed round 29 and it turned out to yield at least as good of players as those drafted at 15. What's going on? Teams that draft later are better. They probably make better picks than those picking at 15. But this artifact can be overcome by picking as well as the good teams. The other factor is that good teams surround those 28 and 29 picks with better talent, so the players have a better chance of making an impact.

The last column is really interesting. It is the number of draft value points that a team needs to spend in order to get a unit of NFL production for each pick. You get a lot more value by picking late than you do picking early.

For example, drafting at pick 5 has the same number of draft value points as picking at 15 plus either 28 or 29. Picking at 5 you can expect to get 7.5 points of NFL production per year (Peyton Manning averages about 20 points of NFL production per year). But if you were instead to pick at 15 and 28 or 29 (approximately the same number of draft value points) you could expect about 11 points of production.

One final caveat. It is better to concentrate NFL production in fewer players. For example, I'd rather have one player contributing 10 points of production than two players contributing 5. The reason for this is that with the player contributing 10, you can put another player on the field. If he contributes anything, that pair will be better than the pair contributing 5 each.

I do think that the draft value chart might be flawed.

What does this mean for trading down? I think I'd rather have two players around 6 AVPY would be better than one contributing 7.5. Does that mean always trade down? No. If you really believe in a player at 5 you should take them. But if you are ambivalent, trading down makes a whole lot of sense.