/cdn.vox-cdn.com/photo_images/4223154/127519334.jpg)
(To read the previous article, click here )
Digging deeper into the numbers based on some great comments from readers I discovered to errors in the data. To be specific, because of a few miscodings of stadiums (e.g. New Yankee Stadium, etc.) some years were counted multiple times. The new data set included 36 parks and stadiums and there were 6.7 seasons on average per venue.
I re-ran the analysis and here are the results:
Re-run of the Analysis | Original Analysis | |
Overall Correlation (Park Factors_Runs to Wins) | .18 | .11 |
Hitter-friendly Parks (>= 1.05) | -.46 | -.42 |
Pitcher-friendly Parks (<= .94) | .11 | .10 |
Neutral Parks (between 1.04 and .95) | .04 | -.25 |
As you can see, even with the mis-codings the results are very similar, aside from the neutral parks. Since 2004, there has essentially been no relationship between a team's home park run environment and their chances of winning. However, we still see the sizeable negative relationship when looking at extreme hitter parks.
(Now, I also have to note that we are dealing with a small N here as we have averaged performance across the 36 parks. The individual correlations certainly are not significant at the .10 level or higher, so take into that this is directional at best when thinking through the results.)
So the question still remains what explains the pattern. Why are extreme hitter parks seemingly such a structural disadvantage for teams?
In the comments section, Nivra suggested that because pitching staffs operating in a hitter-friendly park 81 games out of the year will be worn down more (e.g. relief pitchers must throw more innings), they were at a disadvantage relative to their pitching-friendly park colleagues.
To test this I first looked at the percentage of innings thrown by relievers and it's relationship to park factors. The hypothesis here is that the higher the park factor, the higher the percentage of innings thrown by bullpens as starters will be chased early due to more runs scoring and higher pitch counts.
Overall Correlation (% of IP_Relievers and Park Factors_Runs): -.17
Hitter-friendly Parks (% of IP_Relievers and Park Factors_Runs): .42
Pitcher-friendly Parks (% of IP_Relievers and Park Factors_Runs): -.31
Neutral Parks (% of IP_Relievers and Park Factors_Runs): .13
The relationship isn't very strong across all parks and the relationship is in the opposite direction. However, we see a wide divergence at the extremes and the directionality is what we would expect.
Relievers throw a higher percentage of innings through the course of a season if they pitch in parks with an extreme run scoring environment. There is essentially no relationship in pitcher-friendly parks and a very small one in neutral parks.
One way to think about the pressure this can put on a staff over the course of a year, let's take the following hypothetical based on the average Park Factors over the past eight years:
Team A plays in the most hitter-friendly park in the league (Park Factor = 1.26). Even assuming they play their 81 road games at the most extreme pitcher-friendly parks in the league their average Park Factor for the year will be somewhere in the 1.08 neighborhood. Compare that to Team B, which has most pitcher-friendly park in the league (.81). Even if that team played all it's road games at the most hitter-friendly parks, the average park factors for it's games would only be .97.
That's a big difference in terms of the pressure on a pitching staff, particularly the likelihood that relievers will throw a large percentage of innings, throughout the course of a year.
If we hold various factors constant and look at the effect on team Wins, the percentage of innings thrown by relievers turns out to have almost 5 times the impact of Park Factors and more than double Starting Pitcher FIP.
In other words, the more you rely on your bullpen the lower your chances of accumulating wins. And since hitter-friendly parks tend to increase the percentage of innings your pen works in a season, wins will be harder to come by for these teams.
That being said, I think there is more to the story. The type of park you play in appears to bias your chances of winning, but the extent to which you construct a roster that either takes the greatest advantage of your home park or translates across park types should also play a role.
I will try to attack this question soon in another column.