Writers of Pro Football Prospectus 2008

14 Nov 2007

Darren Sproles, Meet Devin Hester

I can't believe I didn't think of looking for this immediately after Sunday night's special teams debacle, but I noticed a reader (simply signed "M") mention this in the DVOA discussion thread. How bad were the Colts' special teams Sunday night, and was it one of the worst special teams games of the last dozen years?

The estimated value of Indianapolis' special teams on Sunday night was -17.7 points. You had two missed field goals, the two touchdown returns, and Adam Vinatieri could only reach the 10-yard line on one of his four kickoffs. Mike Scifres had five punts; the Chargers downed three of them and Craphonso Thorpe actually lost yardage on the two he tried to return.

Folks might remember the essay in the Giants chapter of PFP 2006 that went over the ten best and worst special teams games. The worst game was Cincinnati at Carolina, Week 14 of 2002. That was the week Steve Smith returned two punts for touchdowns, while Bengals punter Travis Dorsch (in his only NFL regular-season game) had a 40-yard free kick and a 10-yard punt. The estimated value of Cincinnati's special teams in that game: -18.6 points. That would make it the only game worse than the Colts this week... except that last year, while nobody was looking, there was actually a game that was even worse.

In their Week 14 Monday Night 42-27 loss to Chicago, St. Louis' special teams cost the Rams a mind-blowing -19.3 points. That's the game where Devin Hester returned two different kickoffs for touchdowns. Jeff Wilkins missed a 48-yard field goal, and the Rams fumbled an extra point attempt. Willie Ponder couldn't get past the 20 on three of his kick returns and fumbled another one (though the Rams recovered). The Rams had one punt return for a grand total of one yard. Punter Matt Turk was actually above average; if he had been just average, this game would have broken the -20-point barrier.

These are the only three games of the last dozen years where the special teams, when translated into DVOA, were below -50% for the game.

Posted by: Aaron Schatz on 14 Nov 2007

23 comments, Last at 15 Nov 2007, 9:44pm by Colts ST Coach


by Brian (not verified) :: Wed, 11/14/2007 - 10:24pm

I guess the obvious question is--how accurate/meaningful is ST DVOA if a single game can drop a team from 16th to 34th?

There are about 30 pass attempts per team per game, but only about 3 or 4 kick returns. ST plays are fairly random events to begin with, and it seems like there just aren't enough ST plays to reliably indicate a central tendency of performance, even through an entire season.

by Brian (not verified) :: Wed, 11/14/2007 - 10:25pm

I meant 31st.

by Catfish (not verified) :: Wed, 11/14/2007 - 10:48pm

34th may have been a more appropriate ranking

by Matt Weiner (not verified) :: Wed, 11/14/2007 - 10:57pm

I'm curious how bad the Steelers' ST was this week -- one kick return for a touchdown, one nearly so, but they made all their FGs and XPs. Is this just an ordinary bad ST game?

by Formersd (not verified) :: Wed, 11/14/2007 - 10:59pm

Significant enough that it can cost you a game. No way the Chargers should have won that game...

by PatsFan (not verified) :: Wed, 11/14/2007 - 11:43pm

So, for those of you who are fans of teams with good special teams play, how does your team do it?

The Pats do it by using some top-line starters on ST and by having a number of experienced ST vets (and paying for it -- while they are cheap in absolute terms, they're rather more expensive than someone from the 7th round or off the street making the minimum). The approach seems to work, but there's always that risk to the starters.

So I'm curious to see what other teams with high-ranked special teams do, to see what alternatives there are to using starters on ST.

by PatsFan (not verified) :: Wed, 11/14/2007 - 11:47pm

So, how does ST DVOA account for a made (or missed) FG after a free kick fair catch? :)

by admin :: Thu, 11/15/2007 - 12:10am

Remember, there have been 2,888 regular season games since 1996, with two teams per game, and this is the second-worst special teams performance out of those 5,776. So your general run-of-the-mill special teams game won't have this effect.

by dbt (not verified) :: Thu, 11/15/2007 - 1:20am

#6: The Bears have all of their starters pick at least one ST unit to play on. They also have spent a reasonable amount on ST starters (they've had the NFC's non-return special teams pro-bowler the last two years in a row, I believe). Guys who have made a huge impact include Brendan Ayenbadejo (pro bowler), FO favorite The Original Adrian Peterson, and former teammate now in Miami Cameron Worrell.

Plus there's the whole Hester thing.

by azibuck (not verified) :: Thu, 11/15/2007 - 1:58am

There's no logical place to put this, so I'll ask it here. I was just watching NFL Replay of the Colts-Chargers, and before Vinatieri misses the FG, the Colts were penalized for a false start on 4th and inches. Before that, replay respotted the ball from a first down gained, to 4th and inches.
They show a clip of Dungy saying this:
"The problem I had with the whole thing is when we had the ball 1st- and 10 at the 16, the ball was outside the 16-yard line. Then, the stake was inside the six, which is hard for me to figure out how that happened.

....I’m trying to find out how the chains got set there. It was clearly 1st-and-10 outside the 16-yard line.�

Sure enough, immediately following the game, they have the segment with Pereira, and this situation is roundly ignored. I remember this happening in a game last year. I almost wrote a letter to the NFL, but the notion is so crackpot. But it's right there on the film. Does anyone know if this was addressed?

by Bobman (not verified) :: Thu, 11/15/2007 - 3:08am

#3 Catfish, you beat me to it.

by Richard (not verified) :: Thu, 11/15/2007 - 3:17am

5: To argue that the Chargers shouldn't have won is to say that the Colts should have and that's just ridiculous.

by Thinker (not verified) :: Thu, 11/15/2007 - 10:16am

In the NFL, everything more fine than a yard is ignored unless there is an official measurement.

There have been several games this year where the spotting of the ball (from the network broadcast view) appears to routinely be off by a foot or two. It is naturally more noticable when it happens near the sticks or on an important 3rd/4th down, but it seems to happen more often.

On much rarer occasions, it appears to me that the sticks are placed incorectly.

Before a 1st down or TD play, those few inches turn out to be virtually meaningless. Same with after most long gainers or a 1st down. There are times, though when it impacts game outcomes.

There have been games when this has seemed to be routine random error - wrong, but equal for both sides. Unfortunately, other times there appears to be a favored team.

by Dr. Fan of the Winner of the Linked Game (not verified) :: Thu, 11/15/2007 - 11:10am

I guess the obvious question is–how accurate/meaningful is ST DVOA if a single game can drop a team from 16th to [32nd]?

Aaron already addressed this partly, but it's also important to realize that special teams are the component of DVOA with the lowest weight. That's partly because they are inherently less important than offense and defense, but also because they have less relative predictive value.

This can show up in games, such as the one linked by my alias, where one team has a better defensive DVOA, a much better offensive DVOA, but a much worse special-teams DVOA. That can lead to the losing team having a much higher DVOA for the game than the winning team, but part of that much higher DVOA depends on special teams receiving a somewhat artifically low weight because of their relatively poor predictive value. Four touchbacks on punts and one-out-of-three field-goal kicking can really hurt a team in a particular game, but don't affect the prediction of future results for that team as much.

Of course, there can be other plays whose importance in a game outweighs their effect on DVOA. (I should add, especially with the game I linked, that it's important to remember that DVOA is based solely on the official play-by-play, and doesn't take into account any calls by the referees that anyone believes, rightly or wrongly, to have been wrong...)

by gerry (not verified) :: Thu, 11/15/2007 - 1:22pm

# 9 and don't forget Isreal Idonije. The guy is 6-6 275 lbs, runs a 4.7/4.8 in the 40, is versitle (plays all four positions on the D-line fairly well) and is absolulely crazy. He's blocked two field goals and one extra point in three consecutive games this year. He's the wedge breaker on kickoffs and usually has a couple of great sticks a few times a game (just watch him on kickoffs - He and Devin Hester are the only guys worth watching). Oh, and he was an art major in college.

by Wanker79 (not verified) :: Thu, 11/15/2007 - 2:23pm

Re: 1 & 14

The data range is also a factor (I doubt I'm using the correct terminology here, but oh well). There's 124.8% between the top TOTAL DVOA and the bottom TOTAL DVOA. There a range of 86.5% in OFF. DVOA and 52.5% in DEF. DVOA. But S.T. DVOA only has a range of 15.5%. So the ST rankings are subject to much more movement than the other three catagories.

Click my name to see the TOTAL, OFF., DEF., and ST DVOA all graphed on the say y-axis scale.

by ernie cohen (not verified) :: Thu, 11/15/2007 - 2:36pm

re 14:

Actually, the regression shows that ST DVOA is undervalued wrt O/D DVOA, not overvalued, in predicting future results. The problem with ST DVOA is that it is very noisy, so the ST DVOA of a team computed over part of a season is not a reliable measure of ST ability.

In other words, if the season was much longer (so that the ST DVOA numbers were more statistically meaningful), you would want top multiply all ST DVOA numbers (by about a factor of 2) in computing total DVOA.

by Vern (not verified) :: Thu, 11/15/2007 - 5:10pm

Anyone know what the ST DVOA was for the Steelers - Patriots AFC Championship game back in 2001?

Blocked FG returned for a TD.
Punt returned for a TD.

Gameday play-by-play linked on name.

by Vern (not verified) :: Thu, 11/15/2007 - 5:11pm

Anyone know what the ST DVOA was for the Steelers - Patriots AFC Championship game back in 2001? Blocked FG returned for a TD. Punt returned for a TD. Game linked on name.

by Vern (not verified) :: Thu, 11/15/2007 - 5:13pm

F'in spam filter on this site sucks.

I'll try one last time. Anyone know what the DVOA was for the Pats Steelers 01 playoff game was? Linked on my name.

by Dr. Following the Citations (not verified) :: Thu, 11/15/2007 - 5:43pm

ernie cohen,
I'm having trouble reconciling the first and second sentences in your post. Can you refer me to a link where Aaron or someone else explains that?

by ptfe (not verified) :: Thu, 11/15/2007 - 7:30pm

Re 13: [Ball/stick placement] It is, of course, bizarre that the least precise thing in the NFL is the spot of the ball, but we still see that dog-and-pony show of measuring the inches in almost every game. The best way to minimize the problems with these inherent inaccuracies, of course, is to be sure that the sticks and ball line up on 1st down: regardless of whether the ball is placed in the correct spot on subsequent downs, this at least guarantees that a team need only go 10 standard yards to keep a drive alive, even if those yards are ticked off in as-yet-unstandardized ref-yards.

On that note, I'd be interested to see even preliminary data showing the number of times measurements are done in each game, what the outcome of those measurements is, and the relevance of them to the game's outcome. Related to that, I'd be interested to know how often the ball is spotted incorrectly by a ball length or more during a typical game, and how often the sticks fall that far from the actual spot on 1st down.

Sounds like a lot of work for -- well, someone else, frankly. But that's why we post things on this board and don't just do the damn studies on our own.

by Colts ST Coach (not verified) :: Thu, 11/15/2007 - 9:44pm

Hey! I asked this at #35 in Audibles at the Line this week. You stole all of my glory 'M'!

Anyway, thanks for answering it.