Preseason NFL kicking trends

With the recent changes to thetouch-back rules taking effect this 2016 NFL preseason, one can begin to probe the data for trends in both returner and kicker strategy.  To this end, a python script was written to scrape the play-by-play data from the 2015 and 2016 seasons for every instance of the word “kicks” or “kickoff” from the ESPN play-by-play data.  The result was 604 and 556 respective kickoffs from 2015 and 2016 preseason games.  After excluding onside kicks (9 and 6 respectively), kicks that were called back due to penalty (3 and 4 respectively), and kicks from locations other than the 35 yard line (18 and 19 respectively), we’re left with 574 and 527 standard kick-offs in 2015 and 2016 respectively:

Year	G	KOs	Onside KOs	From non-35	No Play	Standard KOs	Returns from EZ	Returns from field	Touchbacks	OOB
2015 Pre	64	604	9	18	3	574	220	87	261	6
2016 Pre	64	556	6	19	4	527	156	142	229	0

We can now compare the average starting field position (FP) following pre-season NFL kicks between 2015 and 2016:

Starting FP of "standard" KOs
Year	Kicks	Avg.FP	SD.FP	p-value
2015	574	20.8	7.3	< 0001
2016	527	23.1	7.5

It’s no surprise that the average starting field position is significantly different after increasing the touchback spot by a full five yards, but it is interesting that the difference is only 2.3 yards.  Furthermore, we can compare the field position in 2016 of kicks into and out of the end zone (EZ):

Starting FP by EZ kick location in 2016
Year	N	Avg FP	SD	p-value
Kick to EZ	382	23.3	7.3	0.364
Out of EZ	145	22.6	7.8

Though the difference is not statistically significant, a kick short of the end zone yielded 0.67 fewer yards (2 feet) this pre-season.  In preseason games, players are vying for a roster spot and are likely being instructed to return kicks at a higher percentage than we should expect in the regular season.  To this end, let’s examine the return and kick rates:

		Returner Decision						Kicker Decision
Year	Stat	Kneel	Return	p-value		Year	Stat	out of EZ	Kick to EZ	p-value
2015	freq	261	220	0.124		2015	freq	87	481	< 0.0001
	rel %	54.3%	45.7%				rel %	15.3%	84.7%
2016	freq	229	156			2016	freq	142	385
	rel %	59.5%	40.5%				rel %	26.9%	73.1%

Indeed, the 5% bump in touch-back rate may only be a difference due to random noise. However, the 11.6% increase in kicks short of the end-zone is a big enough difference to prove a systematic difference between kicker behaviors between 2015 and 2016.  As the informed reader knows, preseason is about the process of player evaluation and may not be reflective of regular season trends, so kickers and returners may do drastically different things next week, but rest assured that this data will be evaluated in a prospective manner as the season unfolds.