The SoapBoxers

Why ESPN Park Factors Are Wrong

April 15, 2010

kosmo - See all 763 of my articles

An interesting thing about baseball is that every field is different.  The dimensions are different, the amount of foul territory is different, and the altitude is different (see Coors Field for a good example of this).

For this reason, baseball statistics must be taken in their proper context.  The typical way to do this is to calculate a “park factor” for a park and adjust a player’s statistics according.  ESPN has a page devoted to this.  However, their park factors are, for lack of a better word, wrong.  Let’s investigate.

The formula ESPN uses (using runs as an example) is:

((homeRS + homeRA)/(homeG)) / ((roadRS + roadRA)/(roadG))

 (RS = Runs Scored.  RA = Runs Allowed.  HomeG = Home Games.)

Here’s a very simple example:

Rockies 6, Padres 3 @ Coors Field

Rockies  4, Padres 2 @ Petco Park
 

The park factors for runs would be:

Coors: 

((6+3)/1)/((4+2)/1))

9/6 = 1.500

Petco

((4+2)/1)/((6+3)/1))

6/9 = .667

 
On the face, this looks pretty solid.  However, it relies on one flawed assumption: all teams are on a level playing field for road games.  However, this is not true.  Baseball has an unbalanced schedule in which you play many more games against teams in your own division that you do against teams in other division.  Your road stats will be affected by the fact that you play a disproportionate number of games in those parks.

To examine this flaw in greater detail, let’s construct a league as such.

National League – 16 teams

• 8 teams play in parks that have a league average of homers allowed
• 4 teams play in parks that allow 80% as many homers as the league average
• 4 teams play in parks that allow 120% as many homers as the league average

The National League, as a whole, is thus neutral.

American League: 14 teams, whose parks average out to neutral.

My team plays in a park that allows a league average number of home runs.  For the sake of simplicity, let’s set this at 1 HR/game (total for both teams).

The four other teams in my division play in pitcher’s park that allow 80% as many homers as a neutral park (0.8/game).

Of the other eleven teams in the league, there are seven that play in neutral parks and four that player in hitter’s parks that allow 1.2 homers/game.

A typical unbalanced National League schedule has 72 games against divisional opponents, 75 against the rest of the league (6.82 games per team), and 15 against American League teams.  The last few numbers vary a slight bit due to the fact that there are more teams in the National league that the American League.

My unbalanced schedule will be constructed similar to this:

• 81 games at home:  (81 games X 1.0 HR/game = 81 HR)
• 36 games (half of the 72) at divisional parks:  (36 X 0.8 = 28.8 HR)
• 3.41 road games (half the 6.82) against each of the 7 neutral park teams (3.41 X 7 teams X 1.0 HR/game = 23.87 HR)
• 3.41 road games (half the 6.82) against each of the 4 hitter’s park teams (3.41 X 4 teams X 1.2 HR/game = 16.368 HR)
• 7.5 road games against AL teams (half of the 15), assuming a neutral sampling of parks (7 games X 1.0 HR/game = 7.5 HR)

Home: 81 HR

Road:  76.538 HR

Calculated park factor: 81/76.538 = 1.058

The unbalanced schedule has made it appear that the neutral park is 5.8% easier to homer in – but it’s really not!  If you plop this exact same park into a division where the other teams play in hitter friendly environments, you’ll see the opposite effect – the ESPN park factor will suggest that the park is pitcher-friendly.

The ESPN park factors are OK for quick and dirty analysis, but take them with a grain of salt.  This is particular true when the parks within a division lean heavily one way – a park that goes against the grain in that division will have its own effect exaggerated.

Sports
Baseball, ESPN, Park Effects, Park Factors