2006 Kicker Predictions vs. Actuals (1 Viewer)

[GoBears84]

I've been trying to optimize the experts predictions for the Projections Dominator. Not the simplest of tasks, so I decided to stick with kickers first, because there are fewer of them and fewer predictions.

I have the 2006 data from this years PD and I'm using standard scoring (3 pts/FGM, 1pt/EPM, -1 for missed FG or EP).

I found last years PD projections (projforxx.php where xx is the intials of the expert) dated 9/4/2006 and was able to open them in Excel.

I took Mike Herman's and David Dodd's Kicker Projections and subtracted the predictions from the actuals to get a residual (a measure of how far off the prediction was). Any kicker without a prediction was removed from the data set. Initially, I left in data for kickers that were injured or lost their jobs - with the assumption being that all projections should be offset appropriately. For example, both predictably over predicted Vanderjagt. Interestingly, they were closer for Vandy than they were for Vinateri.

I then loaded the data into my statistical program (JMP 6.0.3) and did a oneway analysis of the data, assuming unequal variances of the predictions. According to the analysis, the variances were approximately equal, though Dodds was slightly better, 30.2 to 32.1.

When I pulled out the outliers (2x+ variances) of Gould, Koen and Hall, then Dodds was statistically much better, with the difference in predictions at 20.1 to 23.3. Looking at graphs of the residuals, Herman had three distinct groups where he was either off very high or very low or close to target. Dodd's predictions were evenly dispersed.

What does all of this mean? In terms of predictions, Dodds is as good as Herman. Keep in mind that nobody was able to predict the contributions of Andersen, Cundiff, Gramatica, Novak and Suisham nor accurately predict Gould, Koen and Hall - and that's 25% of the kickers.

I'll work on QB's tonight, but if anybody has any input to my methods, I'd be happy to hear it.

Joel

 

[GoBears84]

I've been trying to optimize the experts predictions for the Projections Dominator. Not the simplest of tasks, so I decided to stick with kickers first, because there are fewer of them and fewer predictions.

I have the 2006 data from this years PD and I'm using standard scoring (3 pts/FGM, 1pt/EPM, -1 for missed FG or EP).

I found last years PD projections (projforxx.php where xx is the intials of the expert) dated 9/4/2006 and was able to open them in Excel.

I took Mike Herman's and David Dodd's Kicker Projections and subtracted the predictions from the actuals to get a residual (a measure of how far off the prediction was). Any kicker without a prediction was removed from the data set. Initially, I left in data for kickers that were injured or lost their jobs - with the assumption being that all projections should be offset appropriately. For example, both predictably over predicted Vanderjagt. Interestingly, they were closer for Vandy than they were for Vinateri.

I then loaded the data into my statistical program (JMP 6.0.3) and did a oneway analysis of the data, assuming unequal variances of the predictions. According to the analysis, the variances were approximately equal, though Dodds was slightly better, 30.2 to 32.1.

When I pulled out the outliers (2x+ variances) of Gould, Koen and Hall, then Dodds was statistically much better, with the difference in predictions at 20.1 to 23.3. Looking at graphs of the residuals, Herman had three distinct groups where he was either off very high or very low or close to target. Dodd's predictions were evenly dispersed.

What does all of this mean? In terms of predictions, Dodds is as good as Herman. Keep in mind that nobody was able to predict the contributions of Andersen, Cundiff, Gramatica, Novak and Suisham nor accurately predict Gould, Koen and Hall - and that's 25% of the kickers.

I'll work on QB's tonight, but if anybody has any input to my methods, I'd be happy to hear it.

Joel

This is additional analysis of the PK data. Has I’ve gone through the other positions I’ve refined my approach and wanted to update the kicker data. As indicated above, I have the 2006 data from this years PD and I'm using standard scoring (3 pts/FGM, 1pt/XP, -1 for missed FG or XP). However, in reviewing the data to determine PPG values I discovered that the 2006 kicker data in PD also includes playoff games. I went to the FBG website and copied over the 2006 regular season data.

I found last years PD projections (projforxx.php where xx is the initials of the expert) dated 9/4/2006 and was able to open them in Excel. I neglected to include Chris Smith’s data the first go around, but it’s included here.

Please note that the analysis method below is that recommended to me by the statistician at my company. We discussed it again today and we still feel it is appropriate. It is different than that recommended by others, but I’m happy to share the data with anybody else who would like to see it. Just PM me.

I calculated the Points Per Game (PG) for each PK based on the number of games claimed to have been played in the 2006.

From there I took PK Projections from Dodds, Herman and Smith and divided them by 16 and subtracted the predictions from the actuals to get a residual (a measure of how far off the prediction was). Then based on the suggestion of ookook, I squared the residuals.

I then sorted the data by PPG scored. It was immediately apparent that Robbie Gould was a statistical outlier. He significantly out performed all of the expert predictions. He was not included in the analysis.

Because PK’s are limited on the roster, I limited my analysis to the top 24, the top 12, and kickers 12-24

The means of the Points per Game residual from the predictions were (minus data for Gould):

…………..…………Top 24………Top 12………...12 to 24

Dodds PG…………..0.89…………..1.07…………..0.64

Herman PG…………1.35…………..1.39…………..1.26

Smith PG…………..1.16…………..1.34…………..0.92

The variances were:

…………..…………Top 24………Top 12………...12 to 24

Dodds PG…………..0.94…………..1.05…………..0.64

Herman PG…………1.33…………..1.37…………..1.30

Smith PG…………..1.28…………..1.45…………..0.92

The analysis indicates that the results for Dodds and Herman are statistically different, and Smith fell somewhere in between. The data for PK’s 12-24 had more of in impact than that of PK’s 1-12. Looking at those kickers that were off the most, I see, Nugent, Lindell, Rayner, Kasay and Reed.

What’s also interesting is that although the analysis indicates a statistical difference in the data set, the mean of the differences is no larger than that which was observed with any of the other positions.

Joel