Assessing Football Strategy: Is Running the Football Often Necessary?
We’ve had so many insightful comments recently, so I wanted to take some time to address some of the issues raised by frequent DCTers in previous articles. All of these thoughts are in relation to the in-game strategy of the Cowboys and other NFL squads. . .
I don’t think David Buehler is being asked to purposely avoid touchbacks. On the two “shanks” from Sunday afternoon, he appeared to simply mis-kick the football. Plus, I don’t see any way Joe D would risk a big return simply for the ability to more consistently pin the opposition at the 15-yard line instead of the 20. Here’s why:
With a 1st and 10 at the 15-yard line, offenses have expected points of 0 for the drive (meaning the average points they score and the average points scored by the opposition following punts/turnovers is about equal). At the 20-yard line, the offense’s expected points are about 0.3. Over the course of, say, 1,000 kickoffs (far more than the Cowboys would conduct in one season, but a fine sample size for demonstration purposes), the Cowboys would “gain” 300 points if they performed the “kick-it-high-and-not-too-deep-and-then-have-perfect-coverage-strategy.”
The problem is that Buehler is incapable of consistently kicking the ball high and not too deep, and the coverage unit is incapable of never missing tackles. Note that, with a first down at the 30-yard line, offenses have 1.0 expected points for the drive. For the short kickoff strategy to make sense, the Cowboys would have to let the return team reach the 30-yard line (and no farther) on 30% of kickoffs or less.
The story doesn’t end there, of course, since the infinitesimal odds of the ‘Boys stopping the return team at exactly the 30-yard line undoubtedly offset the expected points gained from a failure of the return team to reach the 15-yard line. For now, let’s forget kick returns which surpass the 30-yard line but do not reach the end zone and simply factor in kick returns for touchdowns.
If we assume just a 1% touchdown rate, the ‘Boys would yield approximately 70 points on said touchdowns over a 1,000 kickoff sample size. All of a sudden, the kickoff team’s coverage has to get a whole lot better, since they would need to stop the opponent at the 15-yard line on 76.3% of kickoffs (If 763 of the 1,000 kickoffs ended there, the Cowboys would “gain” 0.3 expected points per return, or 228.9 points. The 10 kickoff returns for touchdowns equate to -70 points, while the 227 returns to the 30-yard line add up to -158.9 expected points).
If we would factor in all of the returns which exceed the 30-yard line but fail to reach the end zone, that “success rate” of 76.3% would have to jump significantly, probably to well over 90%. You think Buehler and the coverage unit can stop the return team inside their 15-yard line nine times out of 10? Me neither.
The debate between running the ball versus running effectively continues. You all know I find myself in the latter group, and the numbers seem to support the idea that rushing the football just isn’t as important as it once was. According to Advanced NFL Stats, passing yards-per-attempt is the most important statistic as it relates to winning–or at least the one most correlated to winning–with a strength of correlation of 0.61. Rushing attempts comes in at second with a 0.58 strength of correlation. So rushing the ball frequently leads to wins, right?
Not quite. Remember, these numbers represent the correlation between a specific statistic and winning football games, not necessarily causation. Teams do not win football games because they run the football, but rather run the football because they are already winning. The high strength of correlation between rushing attempts and winning seems to be limited to being just a correlation, not representative of causation. This idea is supported by the negative correlation (-0.17) between passing attempts and winning–losing teams throw the football.
On the other hand, a team’s passing efficiency probably will not increase too much if they are losing. Sure, a defense might play a little softer near the end of games so as to not yield big plays, but the net yards-per-attempt is highly unlikely to be affected as much as the rushing attempts from the team which is winning.
The strength of correlation between rushing yards-per-attempt and winning is 0.18–over three times less than that of passing efficiency. So why run the football at all? The reason I still think rushing efficiency is important is because the majority of the positive effects of a strong rushing game (in terms of efficiency, not total yards) are actually represented in a team’s passing efficiency. We’ve all heard the truism that “you need to run the ball to set up the pass.” While this is far from a necessity, rushing the ball well certainly aids an offense’s ability to throw the football effectively.
So with your permission, I’d like to alter “you need to run the ball to set up the pass” to “you may run the ball, if you would like to do so, and if you can do it with relative success, it should help you perform what really wins football games–throwing the football efficiently.” I don’t think that one’s going to get adopted, but whatever.
So when you hear me say things like “Rushing the ball is only important insofar as it helps to garner big plays via the passing game,” these numbers are the reason why.
Lastly, in my Cowboys-Niners post-game review, I discussed why Jim Harbaugh’s decision to decline a 15-yard penalty on Dallas that would have given his offense a 1st and 10 at the Cowboys’ 22-yard line in favor of a made field goal was a toss up call–meaning neither strategy was significantly superior to the other. A couple readers then asked me how I could claim it was a mistake for Harbaugh to kick a field goal on 4th and 1 at Dallas’ 38-yard line (which I argued was his real mistake). How can a 4th and 1 at the opponent’s 38-yard line be better than a 1st and 10 at their 22?
The reason is the uncertainty built into the field goal. San Fran didn’t know the Cowboys would commit a personal foul to give them an opportunity for a first down, but they also were unaware if David Akers’ would connect on the field goal attempt. The odds of Akers missing the field goal (probably somewhere between 40% and 50%, based on historical kicking data and Akers’ own career success rate in that range) surely trumps the small chance of the Cowboys committing a penalty. The 1st and 10 at the 22-yard line is “worse” in a way than the 4th and 1 at the 38 because the former scenario has three points built into it. To take the 1st and 10, Harbaugh had to take three points off the board. To take the latter, Harbaugh simply would have had to forgo a field goal attempt with expected points in the range of about 1.8. In the end, Harbaugh should have never had to take any points off the board because he should have not been so risk-averse on the fourth down.
Having said all that, I have had a slight shift in my thoughts regarding the decision to accept or decline the penalty on Dallas (once the Niners had already decided to perform a sub-optimal strategy in attempting a 55-yard field goal). When I gave you win percentages in my last article in relation to each strategy (91% for accepting the penalty, 90% for declining it), the time remaining on the clock was not a factor in those numbers. Many people might argue that the Niners were smart to decline the penalty for just that reason, as a 10-point lead with seven minutes to play is almost insurmountable (almost, of course).
However, San Francisco would have had a first down near the Cowboys’ red zone, allowing them to run even more time off of the clock. At worst, they could have drained the clock down to five minutes or less and be left with a kick that would probably be no more than a 35-yard attempt. I’ll take a 95% chance of a 10-point lead with five minutes left on the clock (and a very solid chance–probably around 50%–of making another first down and putting the game almost certainly out of reach) versus a 100% chance of the same lead with 7:30 to play.