A cross-sectional study of NFL tight ends

Building upon our previous investigation of what defines a premier NFL tight end, the next logical step is to look for trends in active tight ends similar to what we saw in the hall of fame tight ends.

To this end, I used the season statistics combined with the NFL depth charts from the 2014 NFL to compile a list of starting NFL tight ends.  This list was approximately 40 players, as some teams field a 2 TE set while others had a mid-season injury at the position.  I struggled a little with what statistics would be most associate with "NFL success", but I ultimately chose to use the receptions, yardage and TD average per season across their career to capture usage, production, and scoring.

It seemed unfair to include rookies in this comparison, as it's widely acknowledged that the change from a college tight end to a professional is a difficult transition.  After excluding 3 rookies, we were left with a sample of 37 active tight ends.  I collected their career statistics along with combine numbers and filled in some of the gaps in the combine data using their college pro-day numbers (when available).  Here's a statistical summary;

Variable	N	Mean	Std Dev	Min	Max
Reception/Season	37	37.9	17.2	8.0	78.6
Yards/Season	37	437.8	215.1	103.6	950.4
TD/Season	37	3.7	2.3	0.3	10.8
DashTime (s)	37	4.69	0.15	4.38	4.91
Height (in)	37	76 3/4	1 1/2	73	80
Weight (lbs)	37	253.6	7.8	236.0	270.0
Bench (reps)	34	21.0	4.5	12.0	33.0
Vertical (in)	36	34 3/5	3	30	42
BroadJump (in)	32	115 2/3	4 1/4	110	128
ShuttleTime (s)	33	4.39	0.17	4.03	4.84
ConeTime (s)	33	7.12	0.20	6.82	7.67

The logical question at this juncture is "Do any of the combine measurements correlated with NFL success (average receptions, yards, or TDs per season).  The answer: "Yes, but only weakly":

Correlations	Dash	Hgt	Wgt	Bench	Vert	Broad	Shuttle	Cone
Receptions	-0.23	0.14	0.09	0.02	0.15	0.28	0.01	-0.22
Yards	-0.30*	0.13	0.08	0.10	0.21	0.33*	-0.02	-0.23
TDs	-0.21	0.17	0.17	0.00	0.22	0.23	-0.01	-0.22

It looks like there may be a weak correlation between dash and cone times, as well as the broad jump and vertical jump.  The asterisk (*) denotes a marginal (p < 0.10) statistical significance.  However, we observed in our case series report on hall of fame tight ends that those men were characterized by a remarkable athleticism for their size.  To this end, it might be reasonable to adjust for height and/or weight in these numbers.  As I had a short stint as a physics teacher, I know the laws that govern the collision of bodies, so I tried to adjust in a manner that would have some physical interpretation that was meaningful:

Adj_Dash = Weight/Dash ~ Mass * Velocity = Average Momentum over 40 yard dash

Adj_Vert = Vertical+Height = High Pointing Potential

Adj_Broad = Weight*Broad ~ Mass * Velocity^2 ~ Explosive Kinetic Energy

Adj_Cone = Height/Cone

In adjusted cone, it may be arbitrary whether we choose to adjust for height or weight.  However, I felt like taller individuals (at the same weight), would have a harder time changing directions due to their increased size.

	Inst_KE	Avg_p	High_Pt	Agility	Composite
SEASON_REC	0.27	0.29*	0.22	0.28	0.38†
SEASON_YDS	0.31*	0.35†	0.28	0.28	0.42†
SEASON_TD	0.28	0.34†	0.29*	0.30*	0.38†

Again, the asterisk (*) denotes a p-value less than 0.10, and the dagger (†) denotes a p-value less than 0.05.  However, it's worth noting that for any correlation greater than 0.25, the p-value is less than 0.15.  The Composite variable is simply the product of the 4 adjusted individual scales, which may suggest that there is some interaction between these 4 variables.  For a final note, I'll leave the reader with the scatter plot matrix to illustrate the strength of linear association that was detailed in the chart above.  I threw in the standard scoring fantasy points and PPR scoring points just to make it 5 x 5.

And so we can see a weakly linear relationship, but significant relationship between Receptions, Yards, and Touchdowns with the composite variable.