'Tis Better to Have Rushed and Lost Than Never to Have Rushed at All
Midway through the first quarter yesterday, and Philadelphia had the ball with 2nd-and-1 on their own 44-yard line. Instead of rushing for the first down, Philadelphia passed incomplete, leaving 3rd-and-1. Again, instead of rushing for the first down, Philadelphia called a pass play, which led to Donovan McNabb being sacked, fumbling, and losing the ball. Next play, Favre passes to Ferguson, touchdown, 7-0 Green Bay.
The Eagles eventually pulled out the win -- oddly enough, in part because Green Bay punted instead of ran on 4th-and-1 -- but if they had not, this is the kind of play that Gregg Easterbrook's Tuesday Morning Quarterback would mention in a comment entitled "'Tis Better to Have Rushed and Lost than Never to Have Rushed at All." It is one of TMQ's great pet peeves, teams that get pass-wacky on third-and-short instead of just pounding the ball for the first down.
Of course, TMQ is only right to complain about this if teams really are better off running the ball on 3rd-and-short situations. Therefore, to test this important axiom of TMQ, I consulted the comprehensive Football Outsiders database of every single play from the 2003 season. The results are pretty clear. In almost every single short yardage situation, running the ball is more likely to lead to a first down or touchdown than passing the ball. (As you will see later, Philadelphia is even better than average rushing in these situations, which makes that play call yesterday even worse.)
I started with every single play during the 2003 regular season where there was only one yard needed for a first down. That's the ultimate short yardage situation, of course; you should be able to just fall forward and get the first down. With one yard to go, teams ran 1212 times, and passed 364 times. The rushing plays were far more effective: 71% of rushing plays resulted in first down or touchdown, compared to only 53% of passing plays.
OK, what if you move the ball back a bit. How about every single play with two yards needed for a first down. That's 542 pass plays and 637 run plays. Once again, the rushing plays were far more effective: 67% of rushing plays resulted in first down or touchdown, compared to only 50% of passing plays.
OK, let's move back a bit more. How about every single play with three yards needed for a first down. At this point, teams are passing more often than rushing, with 592 pass plays and 511 run plays. The running plays are slightly less effective, with only 62% of rushing plays gaining a first down or touchdown, but they are still better than passing plays, which still get a first down or touchdown only 50% of the time.
OK, you say, well, three yards to go isn't really a "must run" play. The defense is probably still playing for both the pass and run. What about those 3rd-and-short plays when the defense is likely to be crowding the line, expecting a run to get a first down?
Well, there were 156 pass plays on 3rd-and-1, and 53% of them made a first down or touchdown. But there were 555 run plays, and 72% of them made a first down or touchdown. 4th-and-1 plays were much more likely to be runs than 3rd-and-1 plays, but the percentage for conversions are the same. Runs converted more often than passes.
Get closer to the goal line, where that one yard you need to get is the only yard remaining, and the advantage of running the ball becomes even stronger. 61% of the 84 run plays on 3rd or 4th-and-goal on the 1-yard line were converted. Only 33% of the 15 pass plays on 3rd or 4th-and-goal on the one were converted. That's a grand total of five plays.
What about when you are on the goal line, but it is first or second down? Well, that's the only short-yardage situation where a pass is more likely to have success than a run, possibly because play action is more likely to confuse a defense on first or second down. There were 57 pass plays and 168 run plays on 1st or 2nd-and-goal from the 1-yard line. The pass plays scored a touchdown 58% of the time, while the runs scored a touchdown only 52% of the time.
Interestingly, move the play back a yard, and passing loses its small advantage over running. On 1st or 2nd-and-goal from 1 or 2-yard line, both runs and passes are successful 51% of the time.
The combination of all these situations is what we call "power success" in our offensive line and defensive line ratings. That consists of all third and fourth downs with two yards or less to go, plus first and second downs on the 1 or 2-yard line. Overall, teams are successful 65% of the time when running in these situations, and only 48% of the time when passing. (As an aside, the offensive line and defensive line ratings are now completely updated through the end of the 2003 season, including directional stats for the first time since Week 12.)
Here are all the results summarized in a fun little table:
|1 to go||53.0%||70.8%||364||1212|
|2 to go||49.7%||67.1%||542||637|
|3 to go||49.9%||62.2%||592||511|
|1-2 to go||51.0%||69.5%||906||1849|
|1-3 to go||50.6%||67.9%||1498||2360|
|1st/2nd-and-goal on 1||57.9%||51.2%||57||168|
|1st/2nd-and-goal on 1-2||50.5%||50.8%||99||262|
|3rd/4th-and-goal on 1||33.3%||60.7%||15||84|
|3rd/4th-and-goal on 1-2||36.5%||57.5%||52||113|
|Total "Power Situations"||47.7%||65.1%||661||1225|
These numbers, of course, are for the entire NFL as a whole. Perhaps there are some teams where the running games are so poor that it makes more sense to pass in power situations than to run the ball. It turns out that there are teams that have more success passing than running in power situations, but only four of them. For three of the teams, the conversion rates on passes aren't much different than rushes: Cincinnati (59% passing, 58% rushing), Cleveland (45% passing, 44% rushing), and the New York Jets (67% passing, 63% rushing).
And then there are the Tennessee Titans. Tennessee ranks next-to-last in power rushing success, converting only 47% of the time when they run in power situations. That's 8 first downs and 8 touchdowns in 34 attempts. But they converted 71% of the time when they passed in power situations. That's 8 touchdowns and 9 first downs in 24 attempts. Yes, this was the best passing conversion rate in the NFL for power situations. This included scoring touchdowns on 6 out of 8 passes from either the 1 or 2-yard line (the other two touchdowns came on other 3rd-and-short passes that went for first downs and then some).
What's ironic here is that Tennessee was the subject of TMQ's first "Better to Have Rushed and Lost" comment in last week's article. Gregg wrote:
Facing third-and-1 in the first quarter, leading 7-0, Tennessee called pass; Steve McNair sprinted backward 10 yards, incompletion, punt. On the Flaming Thumbtacks' next possession, facing third-and-1 again, McNair ran the sneak, first down. Three snaps later, the Thumbtacks faced third-and-2; pass, intercepted. Then, trailing 10-7 in the third quarter, Tennessee faced third-and-1. Eddie George runs for the first down, touchdown on the possession. Then, game tied at 17 with 1:22 remaining, Tennessee faced third-and-1. Eddie George runs for the first down, game-winning field goal on the possession.
So Tennessee faces third-and-short on five occasions. The Titans pass twice, for an interception and an incompletion; they run three times, for three first downs. Does anyone other than TMQ draw a conclusion from this?
The strange thing is that, prior to their game against Baltimore, the Thumbtacks were the only team in the NFL who clearly had established that they should pass rather than run in these situations. They only had two "power situations" in their game against the Patriots this weekend. They got called for a False Start on one, and on the other, they passed -- and completed a 10-yard pass to Derrick Mason on 3rd-and-2. (For good measure, they also passed on their only 2nd-and-2, and completed that pass as well.)
Despite the fact that they were so much better passing than running in power situations, the Titans still ran more often than they passed. The average NFL team ran on 65% of power situations. The Titans ran on only 58% of power situations. The six teams that passed more often than the Titans did in power situations is not exactly a parade of winners. The only NFL team that passed more often than ran in power situations was Cleveland. Second was Detroit, which passed 30 times and ran 30 times, an odd decision by Mariucci considering that the Lions converted 22 of these runs -- and only 12 of these passes. You also have Arizona, Oakland, and both New York (Jersey) teams.
Which team was most likely to run in power situations? Would you believe that it was a team with a terrible running game -- and the best record in football? Yes, the New England Patriots ran on 40 "power" plays, and passed on only 12 "power" plays. Despite taking TMQ's advice to heart, the Patriots actually weren't much better running (63% conversions) than passing (58% conversions). The only other team that ran in three-quarters of power situations was San Diego. I hear they have a running back there.
Here is the table of how all 32 offenses did in 2003, with how often they rushed in power situations, how often they converted in the air, and how often they converted on the ground:
|Team||% Runs on
|Team||% Runs on
You may remember that TMQ often writes, "At this point Tuesday Morning Quarterback has written so many items about the Bills bringing defeat upon themselves by going pass-wacky on shortage-yardage downs that I might as well just enter a generic Buffalo short-yardage-fiasco item into my AutoText." While the Bills, surprisingly, were middle-of-the-pack when it comes to run-pass split in power situations, they were almost last in the league in conversion rate when passing in these situations. Maybe the new coach up there will understand this and Gregg can get rid of his AutoText item.
Note: numbers may differ slightly from official NFL numbers due to play-by-play log errors on a handful of plays.
Postscript: Here is another table that includes some of the situations that were requested in the first few comments after this article went up on Monday morning. Unfortunately, I do not yet have plays split out for the final two minutes of a half, so you'll have to live with just getting the second and fourth quarters as a whole:
On the subject of yards per play, yes, pass plays on 3rd-and-short do gain more yards, on average, than rushing plays. Removing goal-to-go situations, where a short field keeps down the length of plays, a rush play on 3rd/4th-and-1-2 averages 3.8 yards, while a pass play averages 5.6 yards. Is the sharply increased chance of being forced to punt (or worse, turning the ball over on downs) worth the extra 1.8 yards per play? I think that the answer is likely "no," but hopefully I can look at that with further research in the offseason. (Just giving you another reason to keep reading Football Outsiders.)
1 comment, Last at 05 Sep 2010, 8:01pm
#1 by Poly (not verified) // Sep 05, 2010 - 8:01pm
1. I think there is a slight benefit of passing plays over runs in that more yards are gained on average. IOW, if you just ran so often that the rates were equalized (obviously the defense would adjust somewhat, so 100% runs is not best), then downs conversion would be a wash but passing slightly more desirable. Given that, what differential of passing versus running completion percentage makes sense? However, how to dial in the other variables inherent in the different types of plays (interception and sack risk, versus fumble or tackle behind line) to determine the optimal mix?
2. Comments about quality of the teams are very hard to read in the articles, based on expecting us to remember 2003 team records. Better to always make this explicit in articles.
3. I wonder how a given team should decide how much to pass or run? Obviously, quality of the power run game as well as pass game come into it. So each team would be different. For that matter on different games, the opponent's capabilities would differ. How would you advise a coach to determine the appropriate percentage?
4. What do you think of rollout plays by the quarterback (like what Theisman ran in the 80s) that are explicitly designed for the player to look at both pass and run options (and run from a power set, but seemingly more of a pass play in not using the power line).