Features of Kaufman Stadium that effect Runs Scored

by Jeff Zimmerman (TucsonRoyal) on Dec 17, 2008 1:56 PM EST Comment 30 comments

I did some research on park features that effect games scores for my blog and BeyondTheBoxscore and here are the stats for Kaufman Stadium:

Park	Original Runs scored (league adjusted)	Fair Area	Foul Area	Humidity %	Elevation (ft)	Average temperature	Projected Runs scored (league adjusted)
Kauffman Stadium	9.82	109689	23528	61.5	877	76.8	9.98
Regression Equation	8.87+0.4266[AL Team adjustment]	-0.0000098	-0.000054	-0.0176	0.00021	0.0513	-----
Equation for Kauffman	9.30	-1.08	-1.28	-1.08	0.19	3.94	9.98

The equation created from the analysis predicts Kaufman pretty good.

Here are the individual factors in an individual table:

Factor	Change in Total Runs Scored per Game
10 degree F increase	+0.51
Increase in RH by 10%	-0.18
10,000 sq ft increase in foul area	-0.54
10,000 sq ft increase in playing area	-0.10
1000 ft increase in elevation	+0.21

Feel free to discuss or tear apart.

2 recs | Comment 30 comments

Story-email Email Printer Print

Comments

you seemed to have made a fairly radical assumption that all teams’ players have equal ability. I would think that’s a fairly big omitted variable.

Runs Scored needs to be neutralized by talent level somehow.

plus what were your p-values of your variables? were they all significant?

by ZeppelinDZ on Dec 17, 2008 3:49 PM EST reply actions 0 recs

Ok, i took you data and replicated your regression

you have really high p-values on fair area and humidity, which probably says those 2 variables are irrelevant and shouldn’t be included. rerun without those 2 and you get a fairly reasonable fit given the sample size.

But again, you have a huge omitted variable issue imo because you are assuming that all teams are equal in talent. This is a great idea to pursue, but your model as stated is very likely invalid.

by ZeppelinDZ on Dec 17, 2008 3:59 PM EST up reply actions 0 recs

Thanks for the help

I am not exactly sure how to correct for talent, but I will look into it. I am currently looking into creating my own decimal park factors.

BTW – What program do you use for p-values?

by Jeff Zimmerman (TucsonRoyal) on Dec 17, 2008 4:17 PM EST up reply actions 0 recs

excel will do it

by ZeppelinDZ on Dec 17, 2008 4:18 PM EST up reply actions 0 recs

dont use linest(), use the regression function in the data analysis menu

it will give you a more complete output.

if you really want to get serious with it, get a full function stat program. (R is a good free-ware one)

by ZeppelinDZ on Dec 17, 2008 4:20 PM EST up reply actions 0 recs

basically what you want to aim for in this form probably will look like:

RS = a + b*[park variables] + c*[player/team variables]

its a more complete version on how are runs scored… which is one of the fundamental questions of sabermetrics, and thus incredibly complex. I would recommend getting off that line of inquiry because it is a much bigger question than you are trying to answer.

what I would do instead, regress your variables (and others you can find) versus park factors. PFs will neutralize talent to a degree.

I see part of your analysis was to see how well your regression would predict vs. PFs, but really that can be done easier by regressing to PF and checking residuals and adj. r-squared.

by ZeppelinDZ on Dec 17, 2008 4:27 PM EST up reply actions 0 recs

you have peaked my interest

I would also include a variety of dummy variables for leagues, dome, convertible dome, regions (The NE US might have more variable weather than the SW), etc.

Moreover, if you broke down your analysis by month, you might gain more knowledge on weather as well.

Obviously this could get very in-depth depending on how much time you want to put into it.

by ZeppelinDZ on Dec 17, 2008 4:40 PM EST up reply actions 0 recs

It all started with monthly data ..

… when I kept hearing from announcers that pitchers dominate the post season. No the don’t, the reason is that is freaking cold out. If you take the last game of the WS, there might have been 3-4 more HRs if it was 80 vice 40. I plan to look into warm weather vs cold games in the post season. I really want a good Table 2 so that when it warms up in the summer, you can say that it should be about XX runs more scored per game.

I never did post the KC affiliate data on my blog. I will here on Royal Review when I get a chance.

by Jeff Zimmerman (TucsonRoyal) on Dec 17, 2008 4:49 PM EST up reply actions 0 recs

Thanks again for the information

I did find that if I used the five park dimensions instead of fair area the SD increased to .44 and the r-squared went to .57 with petco being the only one with a difference over 1

by Jeff Zimmerman (TucsonRoyal) on Dec 17, 2008 4:41 PM EST up reply actions 0 recs

interesting

I wanna know what love is, I want you to show me

by LeoBloom on Dec 17, 2008 3:53 PM EST reply actions 0 recs

It's been a while since I had

Econometrics and Statistics, but to get around the talent level issue, could you use VORP or WAR and neutralize them out so you basically had 9 replacement level players fielding, pitching, and hitting? Because what I really hate about these things is that the Royals can go from being a hitter’s park one year, to a pitcher’s park the next, mainly because of how bad the Royals pitching staffs have been in the past. I think that exposes a critical flaw in park adjustments. Mainly because the overriding factor in runs scored is talent level.

by AxDxMx on Dec 17, 2008 9:56 PM EST reply actions 0 recs

I thought the overriding factor was GRIT

OMG Banny. FWIW I am only crdtng u w/3 runs allwd bc of DDJ OMFG

by devil_fingers on Dec 17, 2008 10:29 PM EST up reply actions 0 recs

As always

Overlooked!

by true_blue on Dec 17, 2008 10:39 PM EST up reply actions 0 recs

Grit?

Is that what that awful taste is in my mouth after watching Ross Gload bat?

by AxDxMx on Dec 18, 2008 1:11 AM EST up reply actions 0 recs

How do we stat grit?

someone come up with a formula.

It should measure like batting average. 1.000 is perfect, .000 is the worst. Can someone do that in this crazy world of sabermetrics?

Don't forget to send your broken maples to the US Forest Service.

by 306008 on Dec 18, 2008 7:49 AM EST up reply actions 0 recs

Grit formula

Grit

by Jeff Zimmerman (TucsonRoyal) on Dec 18, 2008 11:08 AM EST up reply actions 0 recs

amazing.

Don't forget to send your broken maples to the US Forest Service.

by 306008 on Dec 18, 2008 1:26 PM EST up reply actions 0 recs

Funny you should ask

earlier this year, RR’s own NYRoyal published a post on the GRIT average.

Given that others have done similar studies with similar results, I proposed a name-change:

Earned Respect as Statisically Tangible And Defensible: ERSTAD

OMG Banny. FWIW I am only crdtng u w/3 runs allwd bc of DDJ OMFG

by devil_fingers on Dec 18, 2008 11:57 AM EST up reply actions 0 recs

From what I can see that other likely errors not mentioned yet could include...

I would doubt that the temperature and elevational changes are true linear equations, most likely they would have some sort of curvature, which could lead to some error, most notiably with the Rockies elevation. Also the shape of the outfield wall could be a factor, the wind, crowd noise, domes, day game temperature variation, and jet lag. What it basically come down to is that taking actual runs scored avarged in a park over a year ajusted by the teams run differential vs the rest of the major leauges on the road, and leauge ajustment, is most likely the best way to get close to what the parks affects are instead of trying to figure it out from park factors.

Go Royals!

by BabyBlues on Dec 18, 2008 1:47 AM EST reply actions 0 recs

I have taken some new features into account ...

and am getting better results … mainly using the distance to the fences and away team error rates (. I really want to do wind, but as I stated in the main article, I haven’t yet figured a way to incorporate it. I just got a list of decimal park factors for wOBA. I want to run them and then will re-publish.

The data seem to be pretty linear (I have been doing some work on minor league parks). Temp vs run data might actually curve up from a small sample I have run.

by Jeff Zimmerman (TucsonRoyal) on Dec 18, 2008 11:05 AM EST up reply actions 0 recs

oh thought of a couple more

1. Height of the walls (fenway)
2. Astroturf even maybe type of grass/sod used
3. Surface area of dirt on the infield.

Go Royals!

by BabyBlues on Dec 18, 2008 4:07 PM EST up reply actions 0 recs

Comments

Looked into turf, 4 or 5 teams have, and it is field turf. With the parks with the turf, they were below where they should be -less runs and lower park factors (just got some good park factors). Height of wall caused no change (used in initial analysis – will use again when re-compute).

Infield dirt. The areas for this is tough to do, but will probably look into eventually.

by Jeff Zimmerman (TucsonRoyal) on Dec 18, 2008 4:37 PM EST up reply actions 0 recs

Let's just think like a coach

I dont care why you gave up the home run Joel, you didn’t get the job done.

But coach, if you factor in our park and the wind and the jet lag on the hitter, it should have been a fly ball to the warning track.

Don't forget to send your broken maples to the US Forest Service.

by 306008 on Dec 18, 2008 7:50 AM EST reply actions 0 recs

Hopefully this will change how the coach's think ...

… knowing that on that hot Saturday afternoon game there will be some fly balls that used to go to the warning track will go out. The pitcher is just pitching the same, but the stadium environment has changed. As in Colorado, pitchers need to have longer leash.

by Jeff Zimmerman (TucsonRoyal) on Dec 18, 2008 10:37 AM EST up reply actions 0 recs

This is getting as bad as JoGiu

with a tape measure at RFK stadium saying all those balls I hit should have been homeruns but you have the numbers on the OF wall wrong. I should have 30 HRs this season!

Don't forget to send your broken maples to the US Forest Service.

by 306008 on Dec 18, 2008 7:51 AM EST reply actions 0 recs

did that seriously happen?

it would be cooler if it did

OMG Banny. FWIW I am only crdtng u w/3 runs allwd bc of DDJ OMFG

by devil_fingers on Dec 18, 2008 11:57 AM EST up reply actions 0 recs

It seriously happened.

I remember it on baseball tonight. I’m sure there is an article SOMEWHERE on it, but I don’ t have it.

Don't forget to send your broken maples to the US Forest Service.

by 306008 on Dec 18, 2008 1:27 PM EST up reply actions 0 recs

Bout half way down the page

http://www.ballparktour.com/RFK_Stadium.html

by true_blue on Dec 18, 2008 6:43 PM EST up reply actions 0 recs

here is some of the write up

“That’s not right,” Guillen said. “It’s farther down the lines, and it’s definitely farther in center field.”

Though Guillen has said on several occasions that he would like to finish his career in Washington, he admits to being frustrated by his home park. The numbers show why: He has 18 homers on the road, one at RFK Stadium.

“I know myself,” Guillen said. “I know my power. You guys know if I’m in a different stadium, I have 28, maybe 30 home runs.”

by true_blue on Dec 18, 2008 6:44 PM EST up reply actions 0 recs

I will be interested to see how the renovations change the run-scoring environment

I have no idea if they will help or hinder offensive production, but I do know that renovations have had MAJOR impacts at other venues over the years. For example, Boston built a 2nd deck behind home plate, and turned that ballpark from hitter friendly to neutral. I believe something similar happened in the old Arlington park as well.

I can tell you that the prevailing wind pattern at the K is out to left or out to left-center, and this is especially enhanced once the warm part of summer begins. Surely that will be changed one way or another, considering all the new structure going up both behind the OF and behind the entrance as well.

Mr Glass, this is a pro sports team, not a retail store - run it like one!

by loyal2sdad on Dec 18, 2008 1:43 PM EST reply actions 0 recs