Batter's Box Interactive Magazine - Updated Park Factors (All-Star Break)

About 2 weeks ago, I posted mid-season park factors, adjusted for road park. Here I will present current factors calculated using the same method, as well as Offensive and Defensice rankings based on park-adjusted runs per inning.

I recommend that a park factor composed of 70% 2003 and 30% 2002 be used to evaluate a player's offensive or a pitcher's defensive contributions. Nevertheless, only the 2003 park factors were used to derive the team offence and defence rankings.

American League (2002 factors in parentheses)
Team ..............  HPF ... PPF   (2002)
Kansas City ....... 1.122  1.130 (1.113, 1.122)
Toronto ........... 1.089  1.094 (0.990, 0.989)
Boston ............ 1.072  1.075 (0.968, 0.968)
Texas ............. 1.065  1.071 (1.100, 1.107)
Minnesota ......... 1.031  1.031 (0.974, 0.972)
White Sox ......... 1.024  1.026 (1.030, 1.032)
Tampa Bay ......... 1.008  1.010 (0.983, 0.982)
Anaheim ........... 0.967  0.966 (0.970, 0.968)
Detroit ........... 0.954  0.950 (0.944, 0.939)
Cleveland ......... 0.950  0.945 (1.022, 1.022)
Yankees ........... 0.939  0.935 (0.967, 0.966)
Seattle ..........  0.933  0.931 (0.945, 0.941)
Oakland ..........  0.923  0.918 (1.034, 1.036)
Baltimore ........  0.923  0.917 (0.959, 0.958)

rankings..........Offence (rnk) Defence (rnk)
Yankees ........... 1.191 (1)  0.969 (7)
Boston ............ 1.163 (2)  0.957 (6)
Seattle ........... 1.132 (3)  0.830 (1)
Baltimore ......... 1.126 (4)  1.147 (13)
Toronto ........... 1.120 (5)  1.032 (9)
Anaheim ..........  1.058 (6)  0.904 (4)
Oakland ..........  1.045 (7)  0.894 (3)
Kansas City ....... 0.974 (8)  0.950 (5)
Texas ............  0.967 (9)  1.226 (14)
Minnesota ......... 0.918 (10) 0.998 (8)
Cleveland ......... 0.916 (11) 1.047 (10)
Tampa Bay ......... 0.855 (12) 1.133 (12)
White Sox ......... 0.852 (13) 0.885 (2)
Detroit ........... 0.683 (14) 1.089 (11)

National League (2002 factors in parentheses)
Team ...............HPF .... PPF   (2002)
Arizona ........... 1.178, 1.187 (1.063, 1.068)
Colorado .......... 1.118, 1.127 (1.175, 1.192)
Montreal .......... 1.067, 1.073 (0.980, 0.981)
Houston ........... 1.033, 1.036 (1.060, 1.063)
Cubs .............  1.028, 1.030 (0.994, 0.992)
Cincinnati .......  1.009, 1.009 (1.083, 1.087)
Milwaukee ......... 0.995, 0.996 (0.985, 0.982)
Atlanta ........... 0.973, 0.973 (0.999, 1.000)
San Francisco ..... 0.972, 0.969 (0.938, 0.933)
San Diego ......... 0.970, 0.965 (0.950, 0.945)
St. Louis ......... 0.963, 0.960 (0.975, 0.971)
Los Angeles ....... 0.958, 0.956 (0.939, 0.935)
Philadelphia ...... 0.955, 0.953 (0.907, 0.903)
Pittsburgh ........ 0.936, 0.931 (1.058, 1.061)
Mets .............. 0.926, 0.924 (0.924, 0.921)
Florida ........... 0.919, 0.912 (0.969, 0.967)

rankings...........Offence (rnk) Defence (rnk)
St. Louis ......... 1.231 (1)  1.092 (11)  
Atlanta ........... 1.193 (2)  0.996 (8)
Florida ........... 1.098 (3)  1.062 (10)
Colorado .......... 1.069 (4)  1.038 (9)
Philadelphia ...... 1.059 (5)  0.839 (3)
San Francisco ..... 1.041 (6)  0.931 (7)
Houston ..........  1.032 (7)  0.871 (4)
Pittsburgh ........ 0.993 (8)  1.107 (12)
Cincinnati ........ 0.973 (9)  1.239 (16)
Mets .............. 0.965 (10) 1.178 (15)
Milwaukee ......... 0.960 (11) 1.139 (13)
San Diego ......... 0.926 (12) 1.174 (14)
Montreal .......... 0.912 (13) 0.882 (5)
Cubs .............. 0.908 (14) 0.909 (6)
Arizona ........... 0.864 (15) 0.774 (2)
Los Angeles ......  0.777 (16) 0.720 (1)

Posted by robertdudek on Monday, July 14 2003 @ 09:25 AM EDT.

Updated Park Factors (All-Star Break) | 7 comments | Create New Account

The following comments are owned by whomever posted them. This site is not responsible for what they say.

_Jabonoso - Tuesday, July 15 2003 @ 12:34 PM EDT (#97612) #

Again and again, it furthers proves that pitching goes farther than slugs...

Mike D - Tuesday, July 15 2003 @ 01:40 PM EDT (#97613) #

Robert,

I understand that the Mariners play in a big pitchers' park. Accordingly, they have 50 road dingers and only 40 at home.

Intuitively, I would think that Safeco should have a significant negative impact on SLG, and a moderate negative impact on BA/OBP (because home runs turn into fly outs). Accordingly, Bret Boone takes a 110-point slugging hit at home, and a 35-point OBP hit. That makes sense to me.

But Mike Cameron is at 219/310/417 at home, and Edgar Martinez is a shocking 248/363/355 at home. Does Edgar repeatedly hit 384-foot fly balls to left that get caught one foot before the Safeco fence? I don't mean to be specious here -- obviously, the air is likely very humid with the roof open, killing the carrying power of fly balls.

The big reason for the club's OPS discrepancy (100 points) between home and road is, oddly, doubles. Do you have any theory as to why this is so? The M's have 57 doubles at home and 111 (!) doubles on the road, in approximately 100 more plate appearances. Seattle doesn't play enough road games on turf or at Fenway, where the potential for more doubles is obvious, for this to be the deciding factor. Shouldn't line drives in the gap and down the line, and hard grounders down the line, be roughly equivalent between ballparks -- and as for gappers, with a spacious outfield, maybe even slightly advantageous to hit at Safeco? Yet Mike Cameron has 14 road doubles and five at home. Does the humid air turn would-be doubles into outs -- and if so, how? Does Seattle have thick outfield grass?

For the record, I should note that I'm skeptical of the argument that it's the sheer ballparkiness of the Seattle ballpark that is responsible for Cameron whiffing 48 times (against 18 walks) at home.

robertdudek - Tuesday, July 15 2003 @ 02:13 PM EDT (#97614) #

Cameron is a whiffer, there's no doubt about that. Most parks do not affect walk and homerun rates very much.

Here are some event park factors for Safeco:

Year Team singles  2b/3b    HR     OBP     SLG
2000 SEA   0.940   0.819   0.873  0.940  0.849
2001 SEA   0.963   0.863   0.838  0.935  0.898  
2002 SEA   1.009   0.885   0.824  0.981  0.908
2003 SEA   0.981   0.687   0.994  0.955  0.919

Each of these is a ratio of home park to road park performance for pitchers and hitters, figured separately and then averaged. Singles are figured as singles per ball in the field of play. So are 2B/3B (figured together because the sample size of triples doesn't make separating them out worthwhile). Homeruns are figured as homeruns per ball put into play (i.e. balls in the field of play plus homeruns).

What seems to be happening this year is that more of those long flies are leaving the park (hence a higher HR factor than in the past and a lower 2b/3b factor). Singles have been the least affected over the course of Safeco's existence. What may also be happening is that outfielders are playing a little deeper than in other parks, which might boost singles a bit and reduce 2B/3B.

I think the doubles/triples rate will regress towards the .82-.88 range, and so will homeruns. Safeco, like most pitchers' parks, is tough on extra base hits. Ichiro seems perfectly suited for his park: as an outfielder with great range, he has some extra opportunities to catch balls that might have been homeruns in other parks, and his ability to beat out infield hits is inhibited less by the park than any other type of offensive event.

Gitz - Tuesday, July 15 2003 @ 03:47 PM EDT (#97615) #

Mike, even when the roof is "closed" it's not really closed. Air still gets in at a healthy clip, possibly enough so that the roof is, for all practical purposes, "open." It's just that nobody gets wet if it's raining.

And, to that end, it has been unusually warm and dry here in the Pacific Northwest this year. I don't know what effect this has had, but the humidity -- people in Florida would keel over in spasms of laughter to hear someone in Seattle moan about humidity -- has been less this year than usual, or so I've been told.

Personally, I think the whole idea of Seattle being too rainy has been a well-hidden consipiracy to keep outsiders -- such as yours truly -- away from the area. Of course, I wasn't here the year there was measureable rainfall for 100 straight days. I love the rain and all, but yeesh!

Pepper Moffatt - Tuesday, July 15 2003 @ 07:59 PM EDT (#97616) #

http://economics.about.com
I recommend that a park factor composed of 70% 2003 and 30% 2002 be used to evaluate a player's offensive or a pitcher's defensive contributions.

There's been a few times on this board where someone has mentioned that we use x% of some stat and y% of some other stat and 1-x-y% of the rest to make superhappyuseful stat. The All-Star discussion is a good example of this.

My question: How on earth are you guys calculating these percentages? What's the methodology behind the selection? How can we test that x & y are the proper linear combination instead of a & b. Furthermore, how do we know that the combination should be linear at all?

Mike

robertdudek - Tuesday, July 15 2003 @ 08:35 PM EDT (#97617) #

Mike,

My answers in order:

1) Currenlty, about 63% of 2002-2003 data is 2002 and the other 37% is 2003. I double the weight of 2002 on a game basis and half the weight of 2002 (63%/2 = 31.5; 37%*2 = 74; prorated down to 100% gives you 29.9% and 70.1%).
2) Only that more recent data should be given a greater weight.
3) I have no idea.
4) We don't.

That's what I often do with mid-season park factors - I give the current season double the weight of the previous, after taking sample size into account. This seems logical to me and helps to reign in the outliers a bit.

It's nothing more than a rule-of-thumb. Feel free to discard or replace it as you see fit.

Pepper Moffatt - Tuesday, July 15 2003 @ 11:45 PM EDT (#97618) #

http://economics.about.com
It's nothing more than a rule-of-thumb. Feel free to discard or replace it as you see fit.

I'm definately not saying it's incorrect.. don't get me wrong. I have absolutely no idea what the right answer is and your giving 4 times the weight to 2003 games RE: to 2002 games makes inuitive sense. It makes a lot more sense to me than Nelson Lu's method of *only* considering what has happened in 2003, even when the season is a couple years old.

To be honest, I'm not sure *how* you'd figure out what the correct functional form would be, let alone the parameters. I guess if you wanted to estimate the correct 2003 park estimates you could estimate (using past data), something like:

Y = b1 + b2 * X2003.1 + b3*X2002 + b4*X2001, etc.

Where X2003.1 is the PF for the first half of the 2003 season, X2002 is the PF for the first half of the 2002 season, etc and Y is the PF for the whole season. But since Y should be weighted as Y = aX2003.1 + (81 - a)X2003.2, where a is the number of games already played at the park, you'd have:

(81 - a)X2003.2 = aX2003.1 + b1 + etc.

The major problem with that is that you're going to run into a really nasty multicollinearity problem. Plus the relationship need not be non-linear, and either the functional form / parameters may be different over time.

I guess I'm just of the opinion that since stat-type methodology isn't at all in the baseball mainstream, that us SABR types should be *really* careful of what we present as facts, lest we discredit that way of thinking. That, and because I've done enough statistical work to know the limitations of methodology, I think I'm a lot more skeptical than most when I see numbers presented with explanation.

Mike

Updated Park Factors (All-Star Break) | 7 comments | Create New Account

The following comments are owned by whomever posted them. This site is not responsible for what they say.