Sabermetric Overview Series Part I: Park factors

One of my favorite aspects about baseball is that every ballpark is unique. Imagine if the NBA's Gund Arena had a huge wall running up one sideline, or if three-pointers were 30 feet in one corner and 19 feet in another. Or if the Staples Center had an earthen mound at halfcourt. What if every sport had a Polo Grounds?

It's an amazing nuance (mostly) exclusive to baseball. But because of that wonderful little nuance, it can be hard to compare stats from players of one team who play in one unique park to those of players on another team in a completely different park. Because of all the variables when considering differences in parks (altitude -- which affects a lot more than just the distance of fly balls -- fence distances, wall heights, foul-ground area, batter's eyes, etc.) it is difficult to accurately detail each park effect. More on that later, but there are simple ways to approximate how each park plays, called park factors.

Simple park factors are a set of stats that describe how hitter- or pitcher-friendly each park is. It can be calculated with any stat you like, and it will show you how a park plays triples, for instance, or SLG, or batting average, or ninth inning doubles in the month of august or anything. Commonly, the overall park effect is determined for runs, like this:

(home RA + home RS)/home games
------------------------------
(away RA + away RS)/away games

Park factors are usually notated as a number around 1 (or 100): a park factor of 1.00 (or 100) means the park plays perfectly average, a 1.33 plays 33 percent better for hitters, a .89 plays 11 percent better for pitchers, etc. (One thing to remember, as is true with all statistics, is that you must have a decent sample size or the information could be inaccurate. You can never get a stat down to a one constant and inarguable number, but the bigger the sample size the closer you will be. For park factors, 3-5 years is typically accepted as a reasonable sample size.) According to Baseball Prospectus, Great American Ball Park has played as a 1.038 over the past five years, making it a moderate hitter's park overall.

This is a simple way to do it, and there are more complicated methods of obtaining a more precise measurement. Greg Rybarczyk, who maintains the advanced home run distance calculator hittrackeronline.com, recently put his spreadsheets to work to create a more accurate home run park factor (HRPF) that tries to isolate the effect of the ballpark on home runs (eliminating things like roster makeup, seasonal weather and situational managing.)

HRPF is simply calculated like any other park factor:

(home HR allowed + home HR hit)/home games
------------------------------------------
(away HR allowed + away HR hit)/away games

But Rybarczyk throws it all out the window and actually uses an Excel program that plugs in most quantifiable variables and simulates thousands of balls hit with many trajectories and spins at all 30 ballparks to determine a more accurate effect created just by the ballpark. It's very advanced, and as such it's easy to be skeptical of something that's difficult for most to understand. But hittracker's track record, as it were, is solid, and its methods are explained in such great detail here that I'm willing to trust that his spreadsheets are calibrated correctly.

After running his simulation, Rybarczyk came up with a table of adjusted HRPFs you can download here. Great American Ball Park is No. 2 (111), behind only Coors Field (118). Also of note, right field at Great American is one of the easiest fields in the majors to launch. AT&T Park, where the All-Star Game/Home Run Derby was held, rings up an 83, second-lowest in the majors, which might help to describe why the Derby was so dull.

Park factors are a good way to help teams build. If you're the GM of a team that plays in a park with a low factor for home runs, for instance, you might be OK signing the pitcher whose only flaw is his sky-high HR rate in a crackerbox ballpark. Or if your team plays in a homer-friendly park, you probably want to avoid pitchers with suspect flyball and homer rates. Park factors have also helped spawn other stats -- OPS+ and ERA+ adjust OPS and ERA to a player's park and league. For instance, with ERA+, a pitcher with a low ERA in a pitcher-friendly park won't rate as well as a pitcher with the same ERA in a hitter's park.

That's what I've got. I hope this was not too much at once, and if anyone would like to make some amendments or adjustments, that could only be helpful.

Additional reading:
Home Run Park Factor--A New Approach by Greg Rybarczyk
Baseball Reference's explanation of park factors
Baseball Between the Numbers, Chapter 8.2: How much does Coors Field really matter?

Additional art: