Writers of Pro Football Prospectus 2008

Most Recent FO Features

Harris-Brady.jpg

» Week 12 DVOA Ratings

Denver remains No. 1 in the Football Outsiders DVOA ratings, but New England moves up to No. 2 and has taken over as our Super Bowl favorite.

17 Nov 2003

Guest Column: Going for Two

Guest column by Christopher Moses

The two-point conversion has been a staple of college football for several decades.  It gives a team the opportunity to make up points from missed extra point kicks, or simply to put pressure on the opposing team and coach.   There have been rare instances where, late in a game and down by one, a coach will decide to go for two in order to get the win rather than head to overtime.  On the few instances this tactic has worked, the coach has looked like a genius.  However, the repercussions of missing the two-point conversion in this situation are far greater than the benefits of a successful try.

The two-point conversion was added to the National Football League in 1994.  It has become quite evident over the course of nine years that coaches in the NFL are even more hesitant than college coaches to go for two.  Looking at numbers from 2000-2002, however, you can see that in each consecutive year, the number of fourth down attempts and 2-point conversion attempts increases:


2000 2001 2002
2-point conversion attempts 77 81 93
2-point conversion percentage 44.0% 46.9% 52.7%
Fourth down attempts 455 466 483
Fourth down conversion percentage 49.0% 45.1% 51.4%
Third down conversion percentage 38.0% 37.2% 39.4%

This table shows the totals for all teams from 2000-2002.  The lone team that I have excluded is the Houston Texans as they have only been in existence for one year (2002).

From the table, one is able to see that, in each consecutive year, the number of fourth down attempts and 2-point conversion attempts increases.  This coincides with the 2-point conversion percentage each year.  In just two years, the 2-point conversion percentage jumped 8.7%.  I feel that this is directly correlated with the number of 2-point conversions attempted.  But when I ran regression analysis for each individual team, the results did not support this thesis.  The R-square between 2-point conversions attempted and 2-point conversion percentage was extremely low, at .0141.  

(MATH EXPLANATION: R-square is a statistical term, normally interpreted as the relative amount of variance of the dependent variable y explained or accounted for by the explanatory variables x1, x2,..., xn. For example, if R-square = 0.762 we say that the explanatory variables "explains" 76.2 % of the variance of y.  A high R-square -- close to 1 -- reveals that there is a small variance between the predicted figure and the actual figure, and an R-square of 1 means that the explanatory variable explains 100% of the variance of y.)

The R-square of .0141 means that the explanatory variable (2-point conversions attempted) explains 1.41% of the variance of y (2-point conversion percentage).  One is able to see from this that there is very little correlation at all between the two variables.  The explanation that I have for this is that there is such a small sample size to choose from for each individual team.  For example, the teams that took the most 2-point conversions over the course of the three years were the Detroit Lions and the New Orleans Saints.  Neither team had much success over this span -- Detroit missed the playoffs each year, and New Orleans only made it in 2000.  There is most likely an adverse selection problem here like we may see in insurance at times.  By this I mean that the teams who are most likely to take the extra risk are the teams who have nothing to lose because of their poor record.  For example, in 2001, the Detroit Lions attempted 9 two-point conversions, more than any other team in the league that year.  They finished out of the playoffs with a 2-14 record, and of these two wins, one came in the final week of the season.  

 When comparing the 2001 Lions to teams with a similar number of two-point attempts, one can see the difference between a bottom feeder and a Super Bowl contender.  For example, in 2000, the St. Louis Rams were 9-7 and one of the top teams in the NFL.  They attempted 10 two-point conversions and completed six of them (60%).  Likewise, the Pittsburgh Steelers, in 2002 had a record of 10-5 and were the AFC North Division winners.  They attempted 8 two-point conversions and completed seven of them (87.5%).  

My next theory was that the more times a team decided to go for it on fourth down, the more likely they were to have a higher percentage completion rate on two-point conversions.  I felt this would be true because more fourth down attempts mean that the team is more accustomed to being in a high-pressure, short yardage situation.  However, when I tested this for all the teams individually, the numbers gave back basically the same results as comparing two-point attempts to two-point success.  The R-square in this case was only marginally better at .1483.

I continued to run tests like these for each consecutive year from 2000-2002, unable to find anything with significance.  Like the two tables above, I also ran tests on the total figures for the individual teams.  It was not until I decided to test the overall league for the three-year period that I was able to find something significant.  I was able to find three different variables using this method that turned out to be significant.

The first useful variable was the total amount of two-point conversion attempts for the three-year period.  Instead of measuring individual team totals, I measured the totals for the entire league in each year.   This test supports my thesis that with each progressive two-point conversion attempt, the probability of converting that attempt increases.  Here there is an R-square value of .9917.  I think the reason for this is that every time a team attempts a two-point conversion they become more comfortable with it.  Because the two-point conversion rule is so new to the NFL, coaches and players alike are unaccustomed to the situation.  The coach may have difficulty deciding what kind of play to run, and the players may not be able to execute the play the way they would under normal circumstances.  

The second test with significance was for fourth downs attempted.  Comparing with 2-point conversion percentage gives an R-square of .9955.  The idea here is that the more times a team is in a pressure situation, the more likely they will be able to convert in a pressure situation down the road, much like the case of the 2-point conversion attempts.

The third and final variable that I found to be somewhat significant was the yards per rush for all the teams over three years.  This gives an R-square value of .8988:


2000 2001 2002
Yards per rush, all downs 4.050 4.052 4.229
2-point conversion percentage 44.0% 46.9% 52.7%

This is not as statistically significant as the previous two variables, and the standard error for this factor is not as low as the previous two examples.  However, there does seem to be some extremely strong evidence that there is a direct correlation between average yards per rush and two-point conversion percentage.  The threat of the run makes a defense stack their front with eight players.  Merely from watching the games with no specific data, I have seen most of the conversions completed via the pass.  Teams with a good running game can use a play-action fake to set up a short pass for their two-point conversion.

From all of the data and tests that I have run, the conclusion that I have reached is that it would be best for teams to go for two every time they score a touchdown.  In every year from 2000-2002, fourth down attempts, 2-point conversion attempts, and yards per rush have increased.  With this, the 2-point conversion percentage has also increased.  This leads me to believe that if teams continue to go for two more and more, their conversion percentage will continue to grow.  

It would take a conversion percentage of only 50% to make going for two every time worthwhile.  Every team will miss at least one or two extra point kicks every single year, so a 2-point conversion percentage of 50% would give a team more points on the year than if they kicked the extra point every time.  

This is not to say that I would go for two if I had just scored a touchdown to tie the game with 30 seconds left.  In this case, obviously, I would definitely kick the extra point.  In all situations other than this type of circumstance, I think it is beneficial for a team to go for two. 

A team would not only be able to score more points and win more games by going for two every time.  There is also the strong probability that kicker salaries would drop.  Kicker point production would be way down as a result of this action, thus creating less necessity for the kicker.  In a world of the salary cap that we currently reside in, this saved salary is invaluable.  There is obviously a possibility that decreasing kicker salaries would not occur as a result of the player's union or other circumstances, yet going for a two-point conversion every time leaves this as a possibility.

Comments?  Make them in our discussion thread, or email Christopher at mosesce @ sbu.edu.  If you are interested in writing a guest column, something that takes a new angle on the NFL, please email us your idea at info @ footballoutsiders.com.

Posted by: admin on 17 Nov 2003

comments