home vs. away performance

Media types have heavily scrutinized and occasionally criticized the NFL for allowing its players to behave irresponsibly (e.g., to get arrested for DUI, or to “make it rain” in a strip club).  Even during the season, it seems rare that a week goes by in which one or more NFL players wasn’t arrested (see the “PFT police blotter”  on profootballtalk.com for a list of arrests).  How a team acts is at least partially a reflection of its coaches and partially a reflection of the strategies used to put the team together (drafting and trading strategies – usually of the general manager). 

A popular hypothesis floating around  these days is that the Cincinatti Bengals were willing to overlook character flaws on draft day if the players they drafted had enough football ability.  It’s easy to show that the Bengals had a lot of high-profile arrests, but it’s much harder to say that those arrests indicate that the team is intrinsically different from any other NFL team.  The Bengals might have just had bad luck, or the Cincinatti police department may have a particular grudge against the team. 

A well-coached team will generally not have many arrests or character issues (which is obvious).  Nevertheless, any cross-section of 60 or so (including practice squad) rich males in their 20s and early 30s seems likely to encounter at least one DUI, if not more, in the time that it takes for a football season to elapse.  So as far as a metric of how well coached a team is, I don’t think arrests by itself is very useful.  It includes too many random elements.

A possibly useful metric, though, is how a team performs at home vs. how it performs away (controlling for confounders such as domes, unique wind patterns, particularly loud stadiums, etc.).  When a team goes on the road, its players are isolated from their friends, families, and favorite bars/clubs.  The team goes to a hotel, and typically has an enforceable curfew.  At home, on the other hand, there is a lot more freedom for players to get into trouble, or to at least have a few drinks/stay up late, thereby affecting performance on Sunday. 

It might be informative to look at home vs. away performances differentials of different coaching regimes for the same team.  Compare, for example, the Jim Fassel home vs. away differential to the Tom Coughlin home vs. away differential.  That would control for confounding, stadium specific factors (because they’d be relatively constant across regimes).


Peter King on homegrown (drafted) teams vs. acquired (by trade or via free agency) teams

Except for his uncanny ability to get candid interviews with NFL players and coaches, Peter King usually fails to impress me.  He’s not bad; he’s just not that good either.  His predictions are (even by his own admission) sometimes just wild speculation, such as the prediction that Vince Young will play this week in Houston regardless of his injury (I own Vince Young in a fantasy league, and luckily I went with my backup, Culpepper, over Young this week; note to fantasy players: do not listen to Peter King’s fantasy advice).

But today, I liked something he wrote in his weekly SI column.  King actually put together a table with the number of players from each draft class of the Miami Dolphins from 1998 to 2003 and how many of those draftees are still on the 53-man roster.  The answer: none.  That’s right, not a single player drafted between 1998 to 2003 is still on the Miami Dolphins, according to King.

This would be informative if he put together some draft stats about other teams, too, and controlled for changes of head coaches.  The only numbers about other teams that King provides is that New England fielded 9 draftees from the 1998 – 2003 drafts.  Perhaps good anecdotal evidence, but not statistical proof by any means.

I’m sure an important determinant of whether players stick around for a long time at one place is consistency at the head coach position.  Miami is right up there with Washington in terms of changing head coaches often.  Control for that, and compare draftees still on the roster for the same team that drafted them: that could offer an informative test of how much building through the draft matters.

Maybe I can do this myself later this week.

Griese’s 2-minute drill vs. Philadelphia

I have to admit that I had given up on the Bears vs. Philly yesterday.  I watched the game up until the Bears punted the ball away in the 4th quarter with about 4 minutes left, down 16 – 12.  As a Bears fan, I was happily surprised to later learn that the Bears got the ball back, drove 97 yards, and scored.

And then I read that Griese called his own plays during that last drive because radio communications had gone out.  Hmmm… Bears had a stagnant offense all game, failed to score a touchdown, and then they suddenly can drive 97 yards and score without Turner calling the plays?  And Griese actually managed to use Devin Hester on that drive, too, on a play that wasn’t simply a go-route (he had a 21-yard reception on that drive)!

Why doesn’t Turner call simple slant passes to Hester?  I haven’t seen one yet.  Look at the Patriots offense: most of their passes are mid-range passes where the receiver crosses the middle of the field.  The receivers then turn them into big gains with their speed.  Randy Moss, Wes Welker, and Stallworth are all fast guys, but none of them are as fast as Hester or even Bernard Berrian.  Can you imagine what Hester or Berrian might do in the Patriots scheme?

I’m not suggesting, by the way, making Hester go airborne in traffic over the middle; that’s Moose’s job.  I’m simply pointing out that doing something other than WR screen and go-route plays to Hester might be a good idea, and Turner is not doing it.  Hell, I think putting Hester in shotgun and hiking to him is a good idea, too. 

By the way, did you notice that Griese didn’t have the arm to get it to Hester in the first half, when Hester was about three steps ahead of the corner on a go-route?  Underthrew it.

fantasy football’s effect on salary distribution

I’ve often wondered whether the growing popularity of fantasy football has resulted in NFL teams spending progressively more on “skill” players.  Because of the salary cap constraint, this would obviously mean relatively less spent on defense and special teams players. 

Why would teams’ choices of salary distributions amongst players be driven by fantasy football?  Quite simply, fantasy football fans create a large source of revenue, and teams are first and foremost interested in maximizing revenue.  Thus, Chad Johnson, a fantasy football darling, as well as a circus clown, might get a relatively higher salary than he would without fantasy football, because Chad Johnson owners watch Bengals games and perhaps even buy Chad Johnson jerseys, etc.  I’m certainly guilty of watching games that I don’t care that much about just to see if my players do well.

The hopefully testable hypothesis, then, is: fantasy football’s growing popularity, and consequential increase in revenue for teams from skill players, has caused teams to spend relatively more of their salary cap on skill positions.

If this is true, the implications might include that teams could take advantage of this trend by purchasing underappreciated non-skill players.  There are two margins here: marginal revenue product and marginal winning percentage.  A skill player might increase the first dramatically while his impact on the second is ambiguous.  Conversely, if skill players’ salaries are all bid up as teams compete for their higher marginal revenue products, then a team interesting in maximizing wins (rather than revenue) could take advantage by buying those non-skill players who will have the largest positive impact on marginal winning percentage.

 Is it testable?

featuring the TE against a Tampa 2 defense: little to no evidence for it

Another part of Easterbrook’s “feature the TE” hypothesis states that good teams use their TE on seam routes against the Tampa 2 defense. This seems plausible, and it also seems testable. While looking over the TE targeting stats, I noticed some teams, such as the NY Giants, seems to always face opponents that target their TE a lot. Is this because those teams run the Tampa 2 often, and their opponents are throwing to their TE more in response?

The relevant statistic to examine to answer this question is the ratio of TE targets to total passes thrown.  So, in order from the highest ratio to lowest, here are the stats from 2006:

Rank Defense TE-target to pass ratio
1 Philadelphia 0.211429
2 Buffalo 0.208577
3 Carolina 0.208
4 New Orleans 0.204641
5 Kansas City 0.197629
6 Denver 0.195167
7 Tennessee 0.192453
8 NY Giants 0.188713
9 Tampa Bay 0.188641
10 Washington 0.187243
11 Miami 0.187123
12 Dallas 0.18591
13 Houston 0.180198
14 Cincinnatti 0.18018
15 NY Jets 0.178571
16 San Diego 0.17658
17 Baltimore 0.166994
18 Cleveland 0.166333
19 Indianapolis 0.166265
20 Jacksonville 0.164436
21 Green Bay 0.163107
22 Arizona 0.159004
23 Atlanta 0.157282
24 New England 0.15251
25 Oakland 0.148781
26 St. Louis 0.148559
27 Pittsburgh 0.147448
28 Detroit 0.1409
29 San Francisco 0.137066
30 Minnesota 0.128548
31 Chicago 0.118761
32 Seattle  0.1170635

It is perhaps suprising, given Easterbrook’s hypothesis, to see that of the ten defenses that inspire TE-targeting the most, only one is a prominent Tampa 2 defense (Tampa Bay, #9).  Chicago’s defense, ranked #31, which is nominally a Tampa 2 team but in reality plays Tampa 2 only about 1/3 of the time, saw opponents target their TE only 11.9% of the time.  Other prominent Tampa 2 teams are Detroit (#28) and Indianapolis (#19).

These numbers do not offer much support to the part of Easterbrook’s hypothesis that says featuring the TE is a good way to counter the Tampa 2.  It is always possible that head coaches are unaware of the hypothesized benefits of featuring the TE against a Tampa 2, but I think it’s more likely that the determining factors of whether a team features the TE in a game are: 1.  the studliness of the TE, and 2. the speed of the linebacking corps.  It doesn’t matter what sort of defense San Diego is playing, for example.  They will feature Antonio Gates regardless.  Ditto Winslow, Tony Gonzalez, and Jason Witten.  But they might consider featuring them a little less if they are being covered by good coverage linebackers such as Brian Urlacher or Julian Peterson. 

I will attempt to follow up on that thought in a future post.

the Easterbrook “feature the TE” hypothesis

As I mentioned in an earlier post, Gregg Easterbrook hypothesized last week in his TMQ column on ESPN.com that featuring the tight end (i.e. throwing to the TE a lot) is a common feature of winning football teams.  Easterbrook states this is because tight ends can take advantage of the weaknesses of the Tampa 2 defense, which is supposedly all the rage in the NFL these days. 

Let’s take a closer look at Easterbrook’s hypothesis that featuring the tight end helps define a winning team in the era of the Tampa 2 defense.

Here are the total number of times tight ends were targeted by each team in 2006 (taken from stats kept by fftoday.com):
Team: TE Targets
Arizona: 47
Atlanta: 118
Baltimore: 142
Buffalo: 39
Carolina: 62
Chicago: 92
Cincinatti: 33
Cleveland: 165
Dallas: 116
Denver: 71
Detroit: 42
Green Bay: 110
Houston: 51
Indianapolis: 136
Jacksonville: 85
Kansas City: 103
Miami: 114
Minnesota: 67
New England: 125
New Orleans: 68
NY Giants: 129
NY Jets: 45
Oakland: 62
Philadelphia: 102
Pittsburgh: 55
San Diego: 137
Seattle: 66
San Francisco: 96
St. Louis: 34
Tampa Bay: 78
Tennessee: 100
Washington: 100

The team targeting its tight ends most in 2006 was Cleveland, with 165 targets. The top 12 teams were (* means that team made the playoffs):

1. cle
2. bal*
3. sd*
4. ind*
6. ne*
7. atl
8. dal*
9. mia
10. gb
11. kc*
12. phi*

8 out of 12, or 66%, of the top TE-targeting teams made the playoffs. But this statistic could simply indicate that teams that pass more overall tend to make the playoffs; teams that are generally pass-oriented would probably pass to the TE often as a byproduct of being a pass-happy squad.

Let’s examine a more telling statistic: the ratio of TE targets to passes thrown. By this measure, the top 12 teams are:

1. cle
2. sd*
3. atl
4. bal*
5. nyg*
6. ind*
7. ne*
8. dal*
9. kc*
10. ten
11. sf
12. was

Now only 7 out of the top 12 TE-featuring offenses (when passing) made the playoffs. 

 These stats seem to support Easterbrook’s hypothesis that featuring the TE is correlated with winning, but they do not say anything about whether that correlation arises from the defensive schemes offenses face.  Does Cleveland throw to its TE more than any other team in the league because they face more opponents running the Tampa 2 defense?  Or is it because Kellen Winslow, Jr. was the best weapon on their offense in 2006?

I will attempt to test that part of the hypothesis, that featuring the TE is the best response to an opponent running the Tampa 2, in my next post.

why Griese succeeds where Grossman failed

I don’t believe in Brian Griese as a “playmaker”. I do, however, believe in Griese as a smart quarterback. I’m also sure that Grossman has better physical quarterbacking abilities, such as arm strength and a quick release, than Griese.

This is not to say that I don’t think Griese will win many games this year at the helm of the Chicago Bears. He will. But so would any journeyman QB who is smart enough to use his tight ends and check down receivers rather than go for the wideouts on every pass. It might be a mere coincidence, but recall the games that Grossman won last year and who his best receivers were. Almost inevitably, Desmond Clark pops into mind.

Gregg Easterbrook, in his weekly TMQ column, makes the case that winning teams seem to be those teams that feature the tight end (although he ignores some losing teams that also feature the tight end, such as the Cleveland Browns and Kellen Winslow, Jr.). His reasoning is sound: against the Tampa 2 defense, the tight end is often covered by a linebacker rather than a safety. Furthermore, the safeties usually stay deep to prevent big plays in the Tampa 2, so the tight end can take advantage of the seam or area between the zones of the corners and the safeties – if the quarterback is smart enough to take 5 to 10 yard passes instead of going for the 20+ yard options that wideout routes often offer.

Grossman is blessed and cursed with a strong arm and quick release. This led him to try to go for the big play more often than he should have; in those games where he utilized his tight end(s), the results were much better.

Griese, so far, has had similar results. In Griese’s two starts, he has targeted the tight end trio (Desmond Clark, Greg Olsen, and John Gilmore) a total of 23 times (14 for Clark, 7 for Olsen, and 2 for Gilmore). 14 of those passes were caught.

In Griese’s two starts, he has already thrown 14 passes Desmond Clark’s way, whereas Grossman targeted Clark a total of 12 times over the first 3 games of the season. Clark’s stats with Griese at QB have been: 7 receptions for 44 yards and 1 TD vs. the Lions and 3 receptions for 62 yards and 1 TD vs. the Packers.

On the other hand, Grossman targeted the tight end trio a total of 16 times over the first THREE games. 12 were caught.