NOT KNOWING YOUR HEADS FROM YOUR TAILS
Pigskinrevolution.com
In one of our meetings with NFL teams regarding our ZEUS™ software, an Assistant GM (whose identity will remain
hidden) told us that since historical data showed that there was no general advantage to receiving the kickoff in
overtime, a team dominated by its defensive prowess should choose to kick off if they won the toss. As we document
here, he was only wrong on both his facts and his proposed strategy.
In the recently released 2006 Pro Football Prospectus, contributor Jim Armstrong laid out the facts regarding kicking
and receiving in overtime. For the first twenty years of NFL overtime (1974-1993), the receiving team went 95-93-13
(50.5%) in the 201 overtime games. Starting with the 1994 season, the receiving team’s record in the twelve
subsequent years was 106-70-3 (60.1%) in 179 games. Armstrong pointed out that the location of the NFL kickoff was
moved from the 35 yard line back to the 30 yard line starting with the 1994 season and his conclusion was that this was
the cause of the change in success for the receiving team. Unlike Mr. "Assistant GM", Jim Armstrong at least got the
facts right. However, his conclusion was erroneous.
ZEUS™ is the perfect tool to study the value of receiving the kickoff in overtime. Unlike NFL history where only 380
data points are available, ZEUS can run 1,000,000 overtime games in less than a minute. With ZEUS’s customization
feature, we can (and in fact did) change the characteristics of both teams, looking into Assistant GM’s contention that a
team with a (relatively speaking) much stronger defense than offense should opt to kick off.
Assuming the kickoff is from the 30 yard line between typical, equal teams the receiving team wins 57.0% of the time
which is a 14% advantage in receiving vs. kicking. Adjusting for varying strengths of offenses and defenses for both
teams, the advantage can drop to as low as 8%, but it never switches in favor of the kicking team. With the possible
exception of hurricane force winds in the face of the receiving team, choosing to kick in the overtime is never the
correct call in the NFL. And as you can see, it’s not even close.
It is also not difficult to tell ZEUS to move the location of the kickoff, in this case back to the pre-1994 35 yard line. Yes,
this did reduce the winning chances of the receiving team (showing that Armstrong was qualitatively correct), but only
by 2.5%. Prior to 1994 ZEUS says the receiving team should have won 54.5% of the time.
One question remains: if it wasn’t the kickoff location that led to the nearly 50-50 success rate for the pre-1994
receiving teams, what was it? Here we only need to apply statistics. The fact is that experiments almost never agree
100% with theory because of the limited number of experimental trials that can be performed. For the 201 pre-1994
overtime games, the single standard deviation (a measure of the expected random fluctuations in the results) is 7.0
games. The prediction (based upon 54.5% winning chances) is that the receiving team should have won 109.5
games. They actually won 101.5 games. (Note that, similar to hockey, we count a tie as ½ win and ½ loss for this
analysis.) The discrepancy between actual wins and predicted wins was -8.0. Divide this by the standard deviation
(7.0) and we see that the experiment disagreed with theory by 1.1 standard deviations. According to statisticians this is
not an unusual random discrepancy. Thus the 1974-1993 overtime data is consistent with ZEUS’s prediction within
statistical uncertainty.
So what exactly is the break even field position for the receiving team to start their drive in overtime? ZEUS says the
18 yard line. Any deeper and the offensive team is an underdog and any further ahead they are a favorite. Therefore
even a guaranteed touchback by the kicking team proves to be disadvantageous (48.2% GWC for the kicking team
with their opponent beginning the drive on the 20 yard line).