"Ballpark Factor" Revisited

by BobbyMac on Dec 22, 2005 3:39 AM EST Comment 8 comments

A lot has been made recently about the "Ballpark Factor" of U.S. Cellular Field in Chicago. Specifically, it is now regarded by most as the best HR park in the AL. After considering it, I challenge this assumption, and in fact the whole way "Ballpark Factor" is computed.

For "Ballpark Factor", both The Bill James Handbook and baseballhq.com use the same concept - take the homers hit in a park (home and opposition) and divide it by the homers hit on the road (home and opp). Actually, the home run rates are divided, so taking the 2004 A's as an example:

Home: 154/4992 (A's 84, opp 70).
Road: 153/5082 (A's 82, opp 71).
==> ballpark HR factor of 102, or +2%.

If I go to baseballhq.com, I can see that for the period 2003-2005, US Cellular is a whopping +31%+37% (lefty/righty) for HR. So, in theory games held at The Cell should produce about 1/3 more HR than games held at a "neutral park". Therein lies the rub. The Chicago White Sox play a disproportianately high percenage of their games in VERY POOR home run parks.

Cleveland: -21-28 (L/R HR% 03-05)
Kansas City: -27/26
Detroit: 0/-15
Minnesota: 0/0 (was negative both sides 02-04)
<information taken from baseballhq.com>

In fact... if you look at White Sox games in these ballparks, there were approximately 1.5 HR hit in each game. Here are the top HR/G for various parks the Sox played in (again, both home and opp HRs are counted):

Fenway: 4.7 HR/9 ip
Ameriquest: 3.9
Rogers Centre: 3.7
Wrigley Field: 2.8
U.S. Cellular: 2.6

In fact, if you give each road ballpark a weight of "1" (instead of giving - in essence - each ballpark a weight of the number of innings played there), the Sox road games averaged 2.1 HR/9 ip.

So, if you don't give extra weight to the parks in the Sox division, The Cell comes up with a BF of 121 in 2005 (2.6 HR/9ip home divided by 2.1 HR/9ip road). This is still high - perhaps higher than any other park using this method, I haven't applied the same calculations to Texas or Toronto. But it's a lot different than the 131/137 BF as published.

Take it for what it's worth. Javier Vazquez will allow a lot of homers in US Cellular. I'm just saying it's likely not as bad as it may appear to be at first. I think the above method is more legitimate for computing BF, and I think that if done for San Diego, a large change will be noted also.

0 recs | Comment 8 comments |

Story-email Email Printer Print

Comments

Kaufman, Context Chnages

Kaufman moves back the fences in 2004. Before then, that was a HR-friendly place. And before then, the U.S. Cellular was pretty close to the 121 rate you estimate as it's "true" park factor.

by FI on Dec 22, 2005 7:24 AM EST reply actions

im a bit confuzzled

what are the 4992 and 5082 numbers? at bats? hits? fly balls?

by cephyn on Dec 22, 2005 2:34 PM EST reply actions

at bats

The Oakland example is an example of how both Bill James Handbook and baseballhq.com compute BF now. HR/AB ratio between home games and away games. I maintain that this incorporates large biases that can be removed with some work on the metric.

by BobbyMac on Dec 23, 2005 11:32 AM EST up reply actions

ive never really thought about this before

but should we be using AB? Is there a good reason for this? I mean, if you hit a ball on the ground, there's 0% chance of a HR (ok, well a miniscule chance of an inside the park) is it possible we should maybe look at HR/FlyBalls? Because if we're looking at HR effect of the park, its ridiculous to look at ground balls, isn't it? The park has no effect on whether the batter hits a grounder or a fly ball right? The park just has an effect on whether a fly ball gets over the fence or not.

Am I being dumb and missing something, or do I have a point here?

by cephyn on Dec 23, 2005 12:10 PM EST up reply actions

I agree

... yet another way the standard metric can be improved. It might be true that players try for more loft when playing at "HR parks". I'm not sure I have that data available to me in a convenient electronic format (i.e. G/F data across ballpark splits), but it would be cool to see a study on that.

My point is only regarding the weighting of every AB the same, when teams play an unequal number of games (hence AB) at various road parks. So, the average Sox (or Padres) road game will be a lot different than the average Rangers (or Rockies) road game. It may be accurate to say that The Cell increased power output by 35% over their road games. But to say that Javier Vazquez will neccessarily allow more HR playing for the White Sox - just because the CHA metric is higher than the ARI metric - isn't accurate. Part of the "package" of being a White Sock is getting to play road games at Kauffman, Metrodome, Comerica, and Jacobs, and those games will knock some HR's off your total, while at the same time increasing the BF metric for the Sox.

by BobbyMac on Dec 23, 2005 1:13 PM EST up reply actions

Context

How do you counter the argument that the index is one of context....finite number of parks etc. take your point further and include HR proficiency with your opponents ...Division and League.....that would seem to be a transient number over time as teams change as well....

by Buzzards Bay on Dec 25, 2005 7:29 PM EST reply actions

I agree again...

That's exactly one of my points, albeit stated more generally. BF is an index of context. SO... you cannot accurately say that because a player played in Arizona (+15% HR in 2003-2005) and is moving to Chicago (+35% HR), they will therefore allow +17% more home runs in their home games (1.35/1.15).

The other point I'm making is that to neutralize the contextual influences somewhat, you can take the HR% in home games and compare this not to the composite HR% in road games, but rather to the average of each road stadium's games. So, if HR(x) is the Home Run percentage at venue x, I'm saying that

HR(home)*13/(HR(1)+HR(2)+ ... + HR(13))

is a more-accurate metric than taking

HR(home)/HR(road),

where HR(road) is the rate computed by using the sum of all homers hit and allowed in road games.

Think about an analogy...

Assume: 75% of league plate appearances are against RHP. Analyzing the BA differences between two full-time batters...

Over 200 AB, Batter A has 100 AB against LHP and 100 against RHP. He hits .500 against LHP (50 hits), and .200 against RHP (20 hits). That's 70 H/200 AB => .350

Batter B has 50 AB against LHP and 150 AB against RHP. He hits .300 against everyone (60 hits) => .300

If these are their true levels of ability, who will hit better over the next 200 AB? If Batter A sees 75% RHP, he will pick up 25 (.500 x 50) plus 30 (.200 x 150) hits, for 55 total (.275). Batter B will again log 60 hits (.300).

by BobbyMac on Dec 26, 2005 12:07 AM EST reply actions

Ballpark

I think your view needs to be entered into the mainstream....BF has become part of the grammar of analysis....in my view your insight creates a real degree question.....

by Buzzards Bay on Jan 6, 2006 8:20 PM EST reply actions