This Is Why Coaches Get Annoyed

A few weeks ago, the Football Outsiders blog linked to a post on the personal blog of Paul DePodesta, "formerly the assistant GM of the Oakland A’s and general manager of the Los Angeles Dodgers, and now special assistant to the general manager in San Diego." 

DePodesta wrote about the difference between process and outcomes, using the example of a blackjack player who hit on 17 (a terrible process) but received a 4 (a fantastic outcome).  I was going to write something about this post at the time, but then realized I didn't have much to add. 

But now I wish I had, if only because then there would at least have been a chance that SI contributor Michael Lombardi would have been 0.001 percent more likely to have seen it, and maybe saved himself the embarrassment of what he wrote yesterday.

Lombardi writes about game management, which he defines as: "a term used to describe the affect [sic] each play call -- both offensively and defensively -- has on the eventual outcome of the game."  Helpfully, he provides two examples each of good and bad game management so that we can better understand the term.

And that's where he gets into the kind of process/outcome problems noted above. 

First of all, you're never going to guess which team he first cites for good game management.  Yeah. Them.

I'll give it you in the full unedited glory:

Let's go back to the Patriots-Chargers divisional playoff game in 2006, after the Chargers had taken a 14-3 lead at the two-minute warning of the first half: Before their final drive of the half, the Patriots offense accumulated only 67 yards and turned the ball over once. Each New England play call was critical, as the Chargers still had two timeouts remaining and a chance to get the ball back and possibly put the game out of reach. At this point, the clock is as big an opponent for New England as the Chargers.

The Patriots offense took the field in a spread formation with three receivers and ran a draw from the shotgun formation. Kevin Faulk gained seven yards and more important the clock kept moving. Next call: another shotgun spread draw, and this time New England picked up the first down and the clock continued to run. With those two draw calls executed to perfection, the Patriots placed themselves in the perfect position -- they changed field position, turned the clock from foe to friend and placed the Chargers defense on the defensive. The Patriots went on to score a touchdown on the drive to cut the lead to 14-10 and went on to win the game. All of this happened because of how they started their last drive.

I'll grant right off the bat that those ended up being two pretty good play calls.  They may even have been the best of all possible play calls in that situation.  The New England spread draw has always been tough to stop, just ask the Eagles.

But let's look at the official play-by-play to see how the rest of that drive actually went down:

So yes, they started with the two good running plays to pick up a first down and run some clock.  But then Brady threw an incomplete pass and on the next play Mankins was called for holding, which pushed the Patriots all the way back to their own 35 yard line and stuck them in a poor down and distance situation. 

And here's my problem.  If the two draw calls "changed field position, turned the clock from foe to friend and placed the Chargers defense on the defensive," wouldn't it then have to be true that the holding penalty "changed field position back, turned the clock from friend to foe and got the Chargers defense back on the offensive"?  

But wait, you say, that's simply the players' fault.  You can't blame that on game management.  So why do the coaches get all the credit when things work -- the spread draws -- and none of the blame when they don't -- the next two plays? 

Of course, Brady being Brady, all he did was go 5-for-6 the rest of the drive to quickly march his team down the field for the game-tightening touchdown.  But that's not "game management," that's "having the best quarterback in the league."

- - - - - -

Here's his next example of great game management:

Another example of wonderful game management occurred in 1995, when the Browns played the Patriots in an AFC wild-card game .Trailing by 10 points with less than two minutes remaining, the Patriots offense was moving the ball. Suddenly they faced a fourth-and-two at the Browns 30. Bill Parcells had two choices: either allow Matt Bahr to attempt a 47-yard field goal and then hope to gain control of the onside kick or just try and gain the two yards and keep the drive alive. Parcells walked over to Bahr and asked him, "Can you make this?" When Bahr hesitated, Parcells ordered the offense to stay on the field. Parcells would rather take his chances on fourth down than send an unsure kicker on the field. The Patriots ended up losing the game, but every decision Parcells made down the stretch gave New England a chance to at least tie the game.

Now this playoff game I don't remember, so I went back to the wikipedia write-up for as much detail as I could find.  Evidently the Patriots ended up converting the fourth down, picking up 14 more yards, and giving Stover a chance at a 33-yarder, which he made.  They then recovered the ensuring onsides kick, but lost when Bledsoe threw four straight incompletions.

My problem with this one isn't the judgment -- I'm all for coaches who make decisions based on people and not just numbers -- but the superficiality of the analysis is striking.  Let's see how this could instead have been written:

Parcells walked over to Bahr and asked him, "Can you make this?"  When Bahr hesitated, Parcells ordered the offense to stay on the field. 

We can get some sense of the probabilities at play here by examining this paper by David Romer, a professor at Cal Berkeley.  Romer argues that football coaches punt way too often on fourth down, and he includes this helpful statistic:

"One case where the departure from win-maximizing choices is particularly striking and relatively easy to see arises when a team faces fourth down and goal to go on its opponent’s 2-yard line early in the game.  In this situation, attempting a field goal is virtually certain to produce three points, while trying for a touchdown has about a three-sevenths chance of producing seven."

(He says "early in the game" because in late game situations teams have to be more aware of the score.  If you only need two points to win, taking a sure field goal is clearly the better choice.) 

It's at least an arguable notion that fourth-and-goal from the two-yard line is a very similar situation to fourth-and-two on the 30-yard-line at the end of a close playoff game.  We could expect the team with a little more room to work with to be slightly more successful because they have more options, but as a ballpark number, the 43 percent chance of conversion seems like the best thing we have.

Now compare that to Bahr's chances of making a 47-yard field goal.  For the season, Bahr was 5-for-7 between 40 and 49 yards and 2-for-5 on kicks beyond 50.  Given the weather conditions (very cold) and Bahr's age (very old) it's probably safe to assume that his probability of making it was a lot closer to 2-for-5 than 5-for-7... 

And then go into comparing those percentages, as well as the chances of scoring touchdowns from the 30 vs. the assumed point of recovery for the onsides kick, etc., etc.  Because if you haven't already thought through all of that -- and I guarantee you the modern era coaches have -- then you're not really doing a great job with game management.

Or with writing this article.

- - - - - -

Lombardi also offers some examples of poor game management.  Here's his first one:

In Week 7 last season, the Chiefs were beating Oakland 6-0 in the second quarter. The Raiders had just one first down prior to the drive. Faced with a third-and-two from the Kansas City 18, Oakland rookie head coach Lane Kiffin called for a run resulting in a 1-yard gain. Instead of kicking the field goal, the Raiders went for it on fourth down and failed to convert. The Raiders ended up losing 12-10. Many naysayers pointed to the fourth-down call as the mistake, but in reality, the third down call was the real miscue. Had the game been managed correctly, Kiffin would have known a field goal was not an option before making his third-down call. With that knowledge, the play book is wide open on third down and the potential is there for a big play.

Ok, let's work backwards on this one:

With that knowledge, the play book is wide open on third down and the potential is there for a big play.

For an offense that Lombardi helpfully points out had only one first down all day. 

Had the game been managed correctly, Kiffin would have known a field goal was not an option before making his third-down call.

Why wasn't it an option?  The Raiders were losing 6-0.  They ended up losing only 12-10.  Seems to me like taking the three points there could have, um, won the game.

Faced with a third-and-two from the Kansas City 18, Oakland rookie head coach Lane Kiffin called for a run resulting in a 1-yard gain. Instead of kicking the field goal, the Raiders went for it on fourth down and failed to convert. The Raiders ended up losing 12-10. Many naysayers pointed to the fourth-down call as the mistake, but in reality, the third down call was the real miscue.

And here's the meat of the issue.  The Raiders had the second-worst passing game in the league last year (in terms of yards gained).  They had the sixth-best rushing attach (same measure).  Is it really so crazy for Kiffin to think he might have a better chance at the first down by running the ball?  Do the words "Daunte Culpepper of 2007" mean anything to you?

Furthermore, coaches have to think beyond just "game management" to "franchise management."  Oakland clearly wasn't making the playoffs last year, maybe Kiffin thought this could be one of those character-building moments when a team relies on superior will to stomp its way to a first down.  Didn't work, but given the conversion percentages referred to in the above paper, it wasn't a bad risk.

- - - - - -

And here's his last example (thankfully):

Another example of poor game management occurred during Week 1 of last season when Denver played Buffalo. Denver had just missed a fourth-quarter field goal that would have put it ahead. Leading 14-12, the Bills got the ball back on their own 33 with 3:32 remaining. The Bills' main objective is to get at least one first down, which would have pretty much assured them of the win. If they couldn't get a first down, their next objective should be to reduce the amount of timeouts Denver had in its arsenal. The Bills ran the ball on first down, and the Broncos choose to not stop the clock. On the next play, Bills rookie running back Marshawn Lynch mistakenly ran out of bounds and the clock stopped. Conventional wisdom says the next play call must be a running play in order to get the clock moving again or at least force the Broncos to call one of their two remaining timeouts. But the Bills called a pass, which fell incomplete -- stopping the clock with 2:36 remaining.

The Bills ended up losing when Denver's Jason Elam kicked a field goal as time expired. The Bills gave this game away with very poor game management . Even if the Bills had just taken a knee the final two offensive plays, they would have won the game.

Okaaaay, this is really saving the best for last.  Yes, the Lynch play was stupid.  Maybe he was drunk at the time. 

If you want to call having a rookie running back in the game at that point poor game management, I guess you could make that case -- although he doesn't.  But let's again go back to the play-by-play and see how this really played out.

On first down, the Bills ran up the middle and gained three yards.  So far, so good.  Now let's assume for the moment that on second down Lynch hadn't run out of bounds, but still gained two yards.  The Broncos would have called a timeout at that point, stopping the clock at about the 2:51 mark.  The situation at that point would have been third and five, with the Broncos still having one timeout left and the two minute warning.  Lombardi seems to think the Bills could have won the game by just kneeling the ball on third down, but let's assume they're not quite that dumb and instead run the ball up the middle, gaining three yards and taking 10 seconds off the clock.  Denver would have called it's last timeout then with about 2:41 remaining.  Following the punt, call it maybe 2:30.

Given that Denver actually a) got the ball back at 2:13, 2) used its second timeout on a play where Javon Walker got hurt but walked off (he could have scrambled off in a no timeout situation) and 3) burned its third timeout after a -3 yard run by Travis Henry, would having used those timeouts on the last drive really have made that much of a difference?  Maybe, but I doubt it.

All of which suggests to me that passing the ball on third and five in an effort to gain a first down and really make things tough on the Broncos wasn't such a bad call.

The most egregious error here, though, is that Lombardi leaves out what happened immediately after the third-down incompletion.  Buffalo's Brian Moorman crushed a 52-yard punt to the Broncos' 10 yard line, minus a five-yard return, which would have put Denver on its own 15 yard line.  But, one of the Bills' blockers was called for holding, which wiped out the punt, moved Buffalo back 10 yards, and ended up with Denver taking over on its own 34, rather than the 15. 

That was the play that killed the Bills, not the aggressive call of a pass on third-and-five which -- given how poorly Buffalo's defense played on that last drive -- was almost certainly the best choice.  (The best defense is having your offense on the field.)

- - - - - -

So what would actually be an example of poor game management, Mr. Smart Guy, you ask?  How about calling any routes that aren't in the end zone when you're down to your last 20 seconds, have no timeouts and Donovan McNabb is your quarterback. 

If he'd just used that one any of those three instead...