Sep 23, 2007

Improved FG Kicker Ranking

I previously looked at FG kicker performance. My evaluation of each kicker was based on how much better the kicker did than would be expected given how distant his attempts were. However, there were several flaws in the original analyisis, which I've hopefully corrected here.

1. Vandergajt's perfect 2003 season was an outlier that distorts the data greatly. Technically, the regression treated his average miss distance as zero, which made his expected accuracy very low. I decided to exclude this one season in the regression.

2. I realized that including average miss and made distances in calculating expected FG% severely overfits the data. Average miss and made distances are affected directly by missed and made field goals, which is essentially what FG% is. Although those variables help describe the distribution of FG attempt distances, they hurt more than help the model. I excluded them.

3. It was pointed out that many of the top kicker seasons belonged to warm weather and indoor kickers, so the model needs to account for home field environment. I included dummy variables for indoor stadia and warm weather teams.

4. It was also suggested that I should use a non-linear relationship between average attempt distance and FG%. For example, a 50-yd FG attempt is more than twice as difficult as a 25-yd attempt. The best fit turned out to be logarithmic, so I used ln(FG%) as the dependent variable in the regression. In case anyone is curious, the graph below illustrates the average FG% for kicks from various distances.


Below are the results of the regression (n=122). About 8% of the variance in FG kicker seasonal performance can be accounted for by situational variables: average attempt distance, warm weather, and a domed home stadium.










VARIABLECOEFFICIENTP-VALUE
Indoor0.0080.726
Warm Wx0.0170.442
Avg Att Dist-0.0140.004
r-squared0.077




Indoor home stadiums and warm weather kickers do not appear to enjoy a significant advantage throughout a season. Neither variable was close to significance. I got the same results for Indoor and Warm Wx regardless of various combinations of independent and dependent variables. But I left both variables in the final regression model. It's a general rule to keep insignificant variables in a model if their coefficients have the expected sign and they would intuitively help account for factors that are not the primary focus of the model. Although the coefficients may be less than perfectly accurate, they help get you closer to the truth than farther from it.

The table below lists each FG kicker's performance from the 2003-06 seasons. The list is sorted from best to worst in terms of exceeding the expected FG accuracy given each kicker's average attempt distance and home stadium environment.

Click on the table headers to sort.































































































































YearPlayerTeamAvg Yds AttAct FG%Exp FG%Act-Exp
2003HansonDet.38.10.960.780.18
2005RackersAriz.38.10.950.780.17
2004HansonDet.38.10.960.790.16
2003VanderjagtInd.33.81.000.840.16
2005NedneyS.F.38.50.930.770.15
2003WilkinsSt.L.35.90.930.800.13
2004WilkinsSt.L.35.90.930.800.13
2004VinatieriN.E.34.90.940.810.13
2006KasayCar.38.90.890.770.12
2006LindellBuff.35.90.920.800.12
2003JanikowskiOak.39.00.880.780.10
2005WilkinsSt.L.39.10.870.770.10
2006StoverBalt.33.20.930.830.10
2006GouldChi.36.80.890.790.10
2004JanikowskiOak.36.30.890.800.09
2004StoverBalt.34.80.910.810.09
2006ElamDen.33.30.930.850.09
2003GrahamCin.37.20.880.790.09
2006VinatieriInd.36.60.890.810.08
2006HansonDet.36.30.880.800.08
2006KaedingS.D.35.80.900.820.08
2003LongwellG.B.36.20.890.810.08
2004LongwellG.B.36.20.890.810.08
2005KaedingS.D.36.40.880.800.08
2005VanderjagtInd.32.80.920.840.08
2003AndersonTen.36.50.870.800.08
2005DawsonClev.31.60.930.860.07
2005StoverBalt.35.90.880.810.07
2003ElamDen.36.20.870.800.07
2005PetersonAtl.31.60.920.850.07
2004GrahamCin.37.20.870.800.07
2006WilkinsSt.L.36.80.870.800.07
2006CarneyN.O.32.40.920.860.06
2006NugentN.Y.J.33.80.890.830.06
2005GrahamCin.34.00.880.820.05
2005M. BryantT.B.38.40.840.790.05
2003KasayCar.36.90.840.790.05
2006AndersenAtl.34.30.870.820.05
2004ElamDen.36.10.850.810.05
2004KasayCar.36.90.840.800.04
2003BrienN.Y.J.37.30.840.800.04
2003StoverBalt.34.70.870.830.04
2003K. BrownHou.38.20.820.780.03
2006FeelyN.Y.G.34.40.850.820.03
2003AndersenK.C.39.10.800.770.03
2003AkersPhil.36.40.830.800.03
2004AkersPhil.36.40.830.800.03
2006ScobeeJax.39.00.810.780.03
2006GrahamCin.36.90.830.810.03
2004ReedPitt.34.00.850.820.03
2005FeelyN.Y.G.36.40.830.810.02
2005LindellBuff.35.30.830.810.02
2004BrienN.Y.J.35.90.830.810.02
2005MareMia.34.80.830.810.02
2003P. DawsonClev.33.60.860.840.01
2005TynesK.C.35.40.820.810.01
2006NedneyS.F.34.30.830.820.01
2006BrownSea.36.20.810.800.01
2005BironasTen.37.40.790.790.01
2005ReedPitt.35.40.830.820.01
2004DawsonClev.34.70.830.820.01
2006LongwellMinn.33.60.840.830.01
2005VinatieriN.E.36.40.800.800.00
2005KasayCar.38.90.770.77-0.01
2003CundiffDall.36.00.790.80-0.01
2004CundiffDall.36.00.790.80-0.01
2003HallWash.38.80.760.77-0.01
2006TynesK.C.36.90.770.79-0.02
2006K. BrownHou.39.40.760.78-0.02
2005NugentN.Y.J.35.70.790.80-0.02
2005HansonDet.35.70.790.81-0.02
2004MareMia.39.00.750.77-0.02
2004HallWash.38.80.760.78-0.02
2004VanderjagtInd.35.50.800.82-0.02
2004KaedingS.D.35.30.800.82-0.02
2004AndersonTen.37.60.770.80-0.02
2004LindellBuff.29.50.860.88-0.03
2005J. BrownSea.41.30.720.75-0.03
2004J. BrownSea.39.70.730.76-0.03
2006BironasTen.34.40.790.82-0.03
2006RackersAriz.37.60.760.79-0.03
2005GouldChi.35.00.780.81-0.03
2006BryantT.B.36.70.770.81-0.03
2004ScobeeJax.35.20.770.81-0.04
2005AkersPhil.39.60.730.76-0.04
2006AkersPhil.34.30.780.82-0.04
2003MareMia.37.40.760.80-0.04
2004TynesK.C.37.90.740.78-0.04
2003J. BrownSea.39.70.730.78-0.04
2003CarneyN.O.38.40.730.78-0.04
2004CarneyN.O.38.40.730.78-0.04
2005ElamDen.38.00.750.79-0.04
2005ScobeeJax.36.40.770.81-0.04
2003ChristieS.D.35.90.750.80-0.05
2003EdingerChi.38.40.720.78-0.05
2004EdingerChi.38.40.720.78-0.05
2005CarneyN.O.33.60.780.83-0.05
2005K. BrownHou.35.60.770.82-0.05
2005LongwellG.B.37.70.740.80-0.06
2006JanikowskiOak.38.90.720.78-0.06
2005EdingerMinn.36.50.740.80-0.06
2006RaynerG.B.35.70.740.80-0.06
2003ConwayClev.36.80.740.80-0.06
2006ReedPitt.35.60.740.81-0.06
2006MareMia.37.10.720.79-0.07
2006DawsonClev.36.70.720.79-0.07
2003EllingMinn.36.80.720.79-0.07
2004EllingMinn.36.80.720.79-0.07
2003FeelyAtl.38.10.700.78-0.07
2004FeelyAtl.38.10.700.78-0.07
2004K. BrownHou.37.50.710.79-0.08
2006GostkowskiN.E.32.70.770.85-0.08
2005CortezInd.35.30.710.81-0.10
2003VinatieriN.E.33.00.740.84-0.11
2003ReedPitt.34.40.720.83-0.11
2006VanderjagtDall.34.00.720.84-0.12
2005JanikowskiOak.37.60.670.78-0.12
2003LindellBuff.33.90.710.83-0.12
2003GramaticaT.B.37.20.620.79-0.17
2004GramaticaT.B.37.20.620.79-0.17
2003MarlerJax.35.90.610.81-0.20
2004GramaticaInd.34.60.580.82-0.24


One thing that immediately stands out to me is how consistent some kickers are. Notice how many times some kickers' seasons are stacked together in order--Wilkins, Longwell, Akers (twice), Cundiff, Carney, J. Brown, and Edinger. Some kickers are incredibly inconsistent--Janikowski and Vanderjagt in particular.

The bottom line in this analysis is that, accounting for attempt distance and home stadium environment, the standard deviation of adjusted accuracy (Actual - Expected FG%) is 7.7%. A FG kicker one standard deviation above the mean kicks 7.7% more accurately than expected given his attempt distance. The average number of FG attempts for a team is 29, so an extra SD of accuracy would yield an additional 2.33 field goals worth 6.7 points in a season.

A rough estimate of the spread between best and worst FG kicker each year is about 34% of accuracy. That would equate to a difference of 9.9 field goals worth 29.6 points. This makes sense because it represents roughly a 4-standard deviation spread between best and worst.

The kickers who had enough qualifying attempts over the 2003-2006 seasons are ranked below in terms of how far they exceeded their expected FG% given their average attempt distance. Vanderjagt was manually assigned the top score for 2003 for his cumulative ranking. The number of years each kicker qualified is also listed. Take the guys with 1 or 2 years of kicks with a grain of salt.



















































RankPlayerYearsAvg AttExp FG%Act FG%Act-Exp
1Wilkins436.90.790.900.105
2Hanson437.10.800.900.101
3Nedney236.40.800.880.082
4Stover434.70.820.900.076
5Rackers237.90.780.850.070
6Peterson131.60.850.920.069
7Graham436.30.810.860.058
8Kasay437.90.780.830.052
9M. Bryant138.40.790.840.051
10Kaeding335.80.810.860.045
11Andersen236.70.790.840.041
12Elam435.90.810.850.040
13Gould235.90.800.830.031
14Brien236.60.810.840.031
15Vanderjagt*434.10.830.810.030
16Longwell435.90.810.840.027
17Anderson237.10.800.820.025
18Vinatieri435.20.820.840.025
19Nugent234.80.820.840.022
20P. Dawson133.60.840.860.015
21Brown136.20.800.810.008
22Janikowski438.00.780.790.006
23Dawson334.30.820.830.003
24Akers436.70.790.79-0.003
25Lindell433.70.830.83-0.003
26Cundiff236.00.800.79-0.008
27Bironas235.90.800.79-0.013
28Tynes336.70.790.78-0.016
29Scobee336.90.800.78-0.017
30Hall238.80.770.76-0.017
31Carney435.70.810.79-0.019
32Feely436.80.800.77-0.024
33Mare437.10.790.77-0.027
34K. Brown437.70.790.76-0.029
35J. Brown340.20.760.73-0.032
36Bryant136.70.810.77-0.035
37Reed434.90.820.78-0.037
38Christie135.90.800.75-0.052
39Edinger337.80.780.73-0.056
40Rayner135.70.800.74-0.062
41Conway136.80.800.74-0.062
42Elling236.80.790.72-0.073
43Gostkowski132.70.850.77-0.084
44Cortez135.30.810.71-0.103
45Gramatica336.30.800.60-0.195
46Marler135.90.810.61-0.203

6 comments:

pitchfork2k1 said...

Nice work. One thing I noticed is that the Steeler's Jeff Reed is pretty low. Heinz field is notoriously tough for kickers, and I don't think Reed is any worse than opposing kickers there. I'm not sure if you have the data to account for it.

Another thing that might be cool is if an analogous analysis for kickoff ability. The charts could be combined by converting to an estimated point value.

Jon M. said...

So, in the revised numbers, #28 Tynes (-.016) is better than #32 Feely (-.024), even if just marginally. Maybe Jerry Reese does know what he's doing. Maybe.

Brian Burke said...

pf-I've heard that about Heinz Field too. It would be interesting to compare Steelers kickers performance home vs. away using this method.

Brian Burke said...

Jon-The difference between -0.016 and -0.024 is miniscule over the course of one season. It would only be about 0.08 of a field goal (about 0.24 points over 16 games). You could consider them both average. Additionally, they had nearly identical average attempt distances, so it would be easy to compare their accuracy rates directly.

Tarr said...

Perhaps the indoor/warm weather factors would be more significant if you considered where each kick took place, as oppose to just applying one factor to a kicker based on his home field. Ideal would be to model every stadium seperately, although that's a ton of work and the data might be too sparse.

I believe FO has three kicking locations: Denver, Florida, and everywhere else. Maybe indoor as well; I don't remember.

Brian Burke said...

Tarr-I was considering adding an addtional dummy variable for Denver, but there's only been one kicker home-based there during the data period. So it would capture Elam's individual performance within the stadium variable.

And it gets even more complicated than just individual stadiums. You'd really have to model the wind, temp, etc. for each kick. Just saying "Giants Stadium" averages out the conditions for every game there. Some days it could be storming, on others it could be a sunny day with a calm breeze.

This is a tough nut to crack, but what I have so far is (hopefully) light years better than simple FG accuracy percentage. It's also why I decided to leave the variables in the final model despite their non-significance.