16 Jan 2004
by Michael David Smith
DVOA is, without a doubt, a great method for measuring the success of football teams. It's clearly significantly better than the simplistic traditional statistics like total yards or completion percentage.
And yet, if we're going to be honest with ourselves, we need to address what I would call the Panther Problem.
The Panther Problem is, of course, the fact that the Carolina Panthers are clearly a better team than DVOA indicates. According to DVOA, the Panthers finished the regular season as the 18th-best team in the league. Does anyone agree with that?
One problem with DVOA is that it doesn't take clock management into account. If you've got the lead in the fourth quarter, you'd much rather run for no gain and take 40 seconds off the clock than throw an incompletion and take five seconds off the clock. But DVOA counts a run for no gain and an incompletion as equal plays. Could it be that the Panthers win so many close games not because they're lucky, but because they manage the clock well? And, conversely, could it be that the team with the highest-rated DVOA among teams with losing records, Tampa Bay, manages the clock poorly? Perhaps DVOA should give some sort of bonus to the offense on any play that runs more than 30 seconds off the clock when the offense is winning in the fourth quarter. And in turn, some sort of DVOA deduction should be given to an offensive team that runs less than 10 seconds off the clock when the offense is winning in the fourth quarter.
So far, no statistic exists that measures clock management effectively. The only one that comes close, time of possession, is all but meaningless because it is usually an effect -- not a cause -- of winning. The question is, what would be a meaningful measurement of clock management?
Fortunately, the Panthers and Bucs played each other in two close games this year, so we can look to those games for some answers. I decided to examine the play-by-play logs of those two games and see what I could find. Now, I'll grant you that this is a preliminary investigation. With only two games, we're dealing with a very small sample size. But if the Bucs had won these two games, they would have won the NFC South, and the Panthers would be sitting at home watching the Bucs in the playoffs.
Let's introduce two new statistics, Seconds Per Play When Winning In The Fourth Quarter, and Seconds Per Play When Losing In The Fourth Quarter. (And if someone has a better name, I'm all ears.) You want a high number in seconds per play when winning, and you want a low number when losing.
When the Panthers had the lead in the fourth quarter of those two games, they averaged 28.9 seconds per offensive play. That's pretty good. When you have the lead, you're trying to take time off the clock. When the Bucs had the ball and the Panthers had the lead, the Bucs averaged 19.3 seconds per offensive play. That's not great. When the other team is winning, you want to move the ball
When the Bucs had the lead in the fourth quarter of those games, they averaged 24.5 seconds per play. Not quite as good as the Panthers. And when the Panthers were on offense and trying to come from behind, they took only 12 seconds off the clock per play. That's very good.
One easy way to express this would be the difference. If we subtract seconds per play while winning from seconds per play while losing, we get: Carolina +16.9, Tampa Bay +5.2.
Clearly, Carolina showed more urgency when coming from behind and more patience when leading.
Now let's look at some of the specifics from the two games:
In the Nov. 9 game at Carolina, the Panthers led 20-7 at the beginning of the fourth quarter but fell behind to 21-20 when the Bucs scored with 4:44 remaining. A big part of clock management is wise use of timeouts. The Panthers' defense had two timeouts at its disposal, so when the Bucs got an interception and had the ball deep in Panthers territory, the Panthers were able to call two timeouts and leave their offense time to score after a Tampa Bay field goal made
the score 24-20.
At this point, the Panthers understood what they needed to do, and they called nothing but passes: Six consecutive passes until they got the game-winning touchdown to Steve Smith with 1:06 left.
Now it was the Bucs' turn to mount a quick drive, and what did they do? A facemask penalty on Kenyatta Walker on the first play. Now it's first-and-21 from the 11-yard line with 56 seconds left, and the Bucs complete a pass to Michael Pittman for 26 yards, which sounds good until you remember that the Bucs had already used all their timeouts and ran 21 seconds off the clock after Will Witherspoon tackled Pittman inbounds. (The Bucs needed about 55 yards in 55 seconds to get into field goal range, so 26 yards in 21 seconds is OK, but it's not as good a play as DVOA would indicate.) After that it was a four-yard gain and three incompletions, and the Panthers took the field for the kneel-down.
The other contest, the Panthers' 12-9 overtime win, had some similar clock-management issues. The Panthers started the fourth quarter with a 9-3 lead and held that lead until the last play of regulation. Although the Panthers weren't particularly effective running the ball, they stuck with it for clock management purposes. When you're trying to run time off the clock, a running back who can consistently gain a couple of yards without fumbling is an asset. Remember, even the most accurate passers throw clock-stopping incompletions on a third of their throws, and good quarterbacks throw interceptions more often than butter-fingered running backs lose fumbles.
So when you see that on one fourth-quarter drive against the Bucs, Stephen Davis had runs of 3, -1, 3, 1, 6, 4, and 2 yards, that doesn't sound like much. But when you realize that Davis was tackled inbounds on every play, the drive consumed nine plays and took four and a half minutes off the clock, it sounds pretty good.
Now, let's be honest for a moment: The Bucs did score a touchdown on the last play of the fourth quarter, and if Kris Jenkins hadn't blocked the extra point, leading to the Panthers' overtime win, none of us would be talking about the Panthers' clock management. But somewhere on that Panthers squad is success that DVOA isn't measuring. Part of it is that blocked extra point. Maybe clock management is another part of it.
Aaron says: Football Outsiders, where our motto is "all formulas subject to improvement." Darn good idea here. Unfortunately, I built our database from the play-by-play logs at ESPN.com and FOXSports.com, neither of which gives the official clock time for each play. You will find that statistic on the (much more complex and less grammatically standard) logs at NFL.com. If we can achieve our goal this offseason, writing software to parse the NFL.com game logs, we should be able to grab this time data and do a much better study of clock management at the end of games. If all goes well, that adjustment will be in the statistics we present during the 2004 season. In the meantime, if you are Andy Reid, pay attention to this.