Comments on Advanced NFL Stats: Does FG Accuracy Decline In Clutch Situations?

Angle shouldn't matter too much. At 44 yards,...

2010-10-11T17:46:41.093-04:00

Angle shouldn't matter too much. At 44 yards, the goalposts "shrink" by about 4 inches when kicking from the hash. Would matter most on very short kicks (20-25 yards), but still has little effect given how close the kicks are.

I also think angle wouldn't matter since teams don't go out of the way to get the ball in a certain spot to kick during normal play.

Do you have any stats comparing accuracy of soccer...

2010-02-02T11:51:37.138-05:00

Do you have any stats comparing accuracy of soccer style v. straight-on kicking in the NFL?

To add one more to ric's list of factors to co...

2009-11-05T16:04:01.843-05:00

To add one more to ric's list of factors to consider: bad snaps.

it seems like where the kick is on the field left-...

2009-11-01T22:42:34.295-05:00

it seems like where the kick is on the field left-middle-right should also matter. i dont think you can discount the angle. id be willing to bet that, that dip in the graph would be due to the angle it is kicked at.

also: what about wind stats? or stats on the amount of times the defense pushes through the line but doesnt block? or times when a player comes around the edge of the o line? it seems like these may be contributing factors.

If the average success rate at some distance is p,...

2009-10-28T23:18:04.710-04:00

If the average success rate at some distance is p, then the RMS uncertainty in the success rate is Sqrt(p*(1-p)/N), where N is the number of attempts. It might be helpful to plot those uncertainties (excel will do it for you) and see whether the black hole at 44 is statistically meaningful or not. I suspect you'll find the answer is: not.

Or, if I might venture a guess: were there about 20 kicks from 44 yards in the clutch/losing sample?

Perhaps the hole at 44 is because the kickers k...

2009-10-26T18:05:18.255-04:00
Perhaps the hole at 44 is because the kickers know it's not an easy kick, but they also know that they are definitely expected to make it. Farther out they don't feel that expectations are so high so the pct goes back up to the non-clutch value.

That would affect 43 and 45 yarders as well. Small sample size random variation. I would be surprised if there wasn't at least one unexplained bump on the chart.

Now I really wish that announcers would stop saying:

"He missed one earlier in the game...so how much confidence can the coach have..."

There is all sorts of bs regarding kickers. For projections, you can basically assume that all NFL kickers are the same and you won't be far off.

'Clutch players' also tend to acquire thei...

2009-10-26T13:59:26.484-04:00
'Clutch players' also tend to acquire their moniker based a lot on situations totally unrelated to the game. For all intents and purposes, a 40 yard field goal attempt in week 1 of the season should be no different to a 40 yard FG attempt in the Super Bowl.

Similarly with QB late game TD drives. Perform one in the Super Bowl and you're "Mr Clutch". Perform one in Week 1, and you're just some random QB. The pitch is the same size, the ball is the same, but because of media interest you are judged to be clutch if you play well in big games.

From a statistical view, how a player performs in a given game is really just a random variable, so the title of clutch is more a case of what dice rolls came up on a big day. A field goal kicker can be no more 'clutch' on game winning kicks than a captain can be clutch at calling the coin toss for overtime.

Perhaps the hole at 44 is because the kickers know...

2009-10-26T11:53:58.400-04:00
Perhaps the hole at 44 is because the kickers know it's not an easy kick, but they also know that they are definitely expected to make it. Farther out they don't feel that expectations are so high so the pct goes back up to the non-clutch value.

This is inline with my similar reading of "cl...

2009-10-26T09:27:54.212-04:00
This is inline with my similar reading of "clutch" players in other sports. I'm sorry that I don't have the time to search for and provide reference links, but I know that the sabremetricians have largely debunked the theory of a "clutch hitter". Similar studies in basketball have debunked the theory of a "clutch shooter".

I think the definition of "clutch" is "all the special times in my mind that I conveniently remember, disregarding all the contrary evidence that I have conveniently forgotten".

Interesting. A study in Chance magazine ten years...

2009-10-25T22:08:55.827-04:00
Interesting. A study in Chance magazine ten years ago found that the success rate for lead-changing kicks was significantly lower than non-lead-changing kicks.

http://saframetrx.com/2008/08/01/probability-field-goal-pat-good/

As for icing the kicker, the research is mixed.

http://en.wikipedia.org/wiki/Icing_the_kicker
http://www.maa.org/mathland/mathtrek_11_15_04.html
http://sportsillustrated.cnn.com/2005/writers/dr_z/01/21/mailbag.z/index.html

Now, if you could only disprove that "icing&q...

2009-10-25T21:40:15.306-04:00
Now, if you could only disprove that "icing" the kicker was a useful tactic.

if you had the data couldn't you look at total...

2009-10-25T19:26:18.031-04:00
if you had the data couldn't you look at total career clutch kicks and determine retrospectively who the best/worst clutch kickers were? If the best kickers are in the league for say, 15 years (selection bias, i know), your numbers would give you a sample of ~30-45 clutch attempts each for at least the most durable of kickers. Not a huge sample but enough to do a hypothesis test and look for significant differences, no? Something tells me you'll find at least a positive bias in the success rates for these "survivors" against the league avg. Maybe even a few that are significantly below average, which would be curious.

Right. Thanks. That's the line of scrimmage fo...

2009-10-25T18:08:17.318-04:00
Right. Thanks. That's the line of scrimmage for the kicks, not the kick distances. That's how my database lists them. It should be 35, 37, and 38 yards for the median kick distances. I'll correct it above.

"The median attempt distance for non-clutch k...

2009-10-25T17:59:28.619-04:00
"The median attempt distance for non-clutch kids was 18 yards, but it was 20 and 21 yards for clutch and clutch(losing/tied) respectively."

I don't think that sentence is accurate. Could it possibly be 38, 40 and 41 yards?