Home Run Park Factors

[TooLiveBrew]

Glennron's post on Sheets/Coors Field in the "Silver Lining" thread (#23) got me wondering about park factors, which led me to an interesting THT article from Greg Rybarczyk:

"Home Run Park Factor -- A New Approach". Some excerpts:

 

 

What's wrong with the current HRPF calculation?

 

Perhaps this can be answered with a single example: in 2002, the HRPF for Chase Field (then Bank One Ballpark) was 48, and then the very next year it rose to 116. Need I say more?

 

...

 

Unfortunately there are LOTS of uncontrolled noise factors that influence the current park factor calculations. Here are a few:

• Atmospherics: conditions will vary from park to park due to differences in climate, and the weather varies from month to month and day to day at the same park.
• Roster makeup: a team's lineup might be heavily skewed towards either right- or left-handed power hitters, magnifying the impact of asymmetry in their home park.
• In-season roster changes, injuries or discretionary days off: these might lead to a slugger playing more games at home than away, or vice versa.
• Ball characteristics: baseballs stored in the dry heat of Phoenix will go farther than balls stored in higher humidity environments (naturally as at Dolphins Stadium, or artificially as at Coors Field)

...

 

Multi-year Park Factors: Still Flawed

 

A much better way of judging a park's propensity for home runs is to use a multi-year average. A multi-year park factor washes away many of the problems listed above, but not all of them.

 

The extreme individual performances (e.g. Cody Ross' three-homer game on Sept. 11, 2006 at Dolphins Stadium) will largely balance out, but the roster makeup factor (left-handed sluggers at Yankee Stadium, right-handed sluggers at Fenway Park) will typically still be there, as will the ball characteristics factor. And of course, interleague play and the unbalanced schedule are still significant factors.

 

Weather effects will typically be reduced, but not entirely removed; day to day variation will be smoothed out, but some weather patterns run on cycles longer than a season. Finally, the building of numerous new ballparks, and recent fence changes on existing parks (Citizens Bank Park, PETCO Park, Miller Park, Comerica Park, Kauffman Stadium, Minute Maid Park) render some of the multi-year park factors problematic for the near-term future.

 

Overall, we should expect and demand more of a park factor. So, if we're thinking about improving the HRPF metric, what would we change? Some thoughts:

• HRPF should not change if the park itself hasn't changed.
• HRPF should describe the impact of the park itself, not the performances of the players who reside in it or visit it.
• HRPF should incorporate atmospheric factors such as wind and temperature that change from game to game.
• HRPF should include individual sub-factors for different hit directions (LF, CF, RF) rather than different hitter types (LH, RH).

 

If this should be in the stats forum, that's fine (& my apologies). I just felt it'd get more discussion in the ML forum, and I've wondered about the problems inherent in the current PF measurements for some time now -- and also wondered how the BF.net community feels, too. I put the second bullet point just above in bold (the other bold parts are author's emphasis), because that's been my long-standing question about park factors in general.

 

 

[logan82]

Sheets curve ball doesn't curve in Colorado.

[TooLiveBrew]

I know... just got me thinking.

[logan82]

Your thread got me thinking how Sheets curveball doesn't seem to break and he has to rely on chnageups, which he almost never used before this year. Also how we seem to have at least one pitcher get shelled every time we go to Colorado.