Of Combines and Correlations – Part 2


Last week, I introduced Arrowhead Addict readers to Joe Landers’ approach to forecasting NFL success, or rather I should say the likelihood thereof, of a NFL draft prospect based on that prospect’s NFL Combine performance. I then used his methodology, first straight up, and then later factoring in a weight adjustment of my own, to assess the likelihood that the Chiefs first round draft pick, Dontari Poe, would be a successful player in the NFL.

To recap, Landers compiled physical attribute scores from 2005 to 2008 and developed peer averages for each attribute at each position. If a prospect performed above average at his position on a Combine metric, he scored an EPA (Exceeded Peer Average) which Landers uses an indicator of future NFL success. Landers’ primary definition of success was whether the prospect made first team in the 2008 season.

There was at least one reasoned criticism about what I did last week in analyzing Dontari Poe. In order to get more of what I felt would be an apples to apples comparison, I added my own tweak to Poe’s assessment by focusing only on Combine scores for DTs who weighed 330 lbs or more. For the record, Poe himself weighed in at 346 lbs and I felt, due to the laws of physics, it was important to make this distinction and try to draw a line somewhere because the prototypical, ideal nose tackle is generally in the 330 plus weight range. The criticism to my approach was that rather than focusing on weight, I should “redo” my analysis by compiling and correlating scores of true Nose Tackles rather than just DT “fatties” without regard to whether they played NT or 43DT. Two non-fatty NT examples were provided – Jay Ratliff (292 lbs) and Barry Cofield (304 lbs).

I’ll address this criticism first generally and then specifically.

To ignore weight as a factor is to ignore physics as a factor.  Because WRs, DBs, and to a lesser extent RBs and LBs, are typically at the lighter end of the NFL weight scale, they tend to be the fastest players on the field. At the other end of the weight scale are OL and DL “uglies” who tend to be the slowest of all players. The reason for this is simple physics. What’s more, a player who lacks speed, particularly linemen, can more than compensate for this with the right combination of power and quickness (noting that quickness is not the same as speed).

At a more specific level, not all NTs are created equal.  Jay Ratliff, for example, is not really used to anchor the middle, eat blockers, and/or collapse the pocket. Rather, he tries to use his combination of size and speed to slide through a gap and he typically comes in on passing downs for that purpose rather than play as a 3 down NT. Similarly, you will rarely see Barry Cofield, when he’s healthy, take on multiple blockers, and when he does, he routinely gets crushed.  In other words, neither of these guys are the type of NT that a guy like Romeo Crennel would want to feature in his 3-4 scheme. Dontari Poe, for his part, looks to possess an ideal combination of size and athleticism to ultimately be a 3 down NT.

Enough rehashing.  Ready for some more Landers-type assessment of our other draft picks?

Jeff Allen, OG*

Landers’ OG peer averages and 1st teamer EPA percentages:

Short Shuttle: 4.72 sec; 53% of 1st teamers scored an EPA

Three Cone: 7.84 sec; 65% of 1st teamers scored an EPA

Vertical Jump: 28.45”; 53% of 1st teamers scored an EPA

Broad Jump: 101”; 47% of 1st teamers scored an EPA

Forty yard dash: 5.31 sec; 53% of 1st teamers scored an EPA

Bench Reps: 25.34; 41% of 1st teamers scored an EPA

*  I say OG, rather than OT, because guard is the position that Allen is expected to play for the Chiefs. Besides, scoring EPAs at the OT position is more challenging than those at the OG position.

In the case of OGs, as was also the case with DTs, the most important test to score an EPA on is the 3 cone drill, wherein 2/3rds of 2008 1st teamers did so within the 2005-2008 study period. Lander’s goes on to conclude the greatest correlation indicator for OG success is for a prospect to score an EPA on the 3 cone and then at least 3 EPAs beyond that.

Jeff Allen’s (6’4”, 307 lbs) Combine Numbers:

Short Shuttle: 4.9 sec

Three Cone: 8.01 sec

Vertical Jump: 27.5”

Broad Jump: 102”; EPA

Forty yard dash: 5.28 sec; EPA

Bench Reps: 26; EPA

As you can see, while Allen scored 3 EPAs, he nevertheless undershot on the leading indicator, the 3 Cone drill. What I take away from this is that, statistically speaking, the likelihood of Allen becoming a starting OG is not as great as would the case had he scored at least .17 seconds better on the 3 cone. Whether Jeff Allen proves himself to be a statistical exception, and there are always are some, remains to be seen.

Next week, the Chiefs 3rd round pick, OT Donald Stephenson.

That’s my Double Take.

What do you take away from this, Addicts? Does Joe Landers’ forecasting model raise any new questions for you about Allen’s future as a Chief? How soon, if at all, do you expect to see Jeff Allen replacing Ryan Lilja as a starter?