Park Factors are one of the more important concepts that the statistical revolution in baseball has given us. Understanding the impact that a given environment has on the game is essential to understanding performance. The numbers are by no means perfect but by using multiple years of data we have a statistical framework to handle the impact of quirks like the Green Monster at Fenway or the bay breezes at AT&T Park. When the Padres target fly ball pitchers or the Yankees lineup skews heavily left-handed, we understand the logic at play.
Often the relationship between a certain player’s performance and their home environment can be extreme. Larry Walker is a notable example since he was part of this last Hall of Fame Ballot and because he hit in possibly the most extreme offensive environment in baseball history. Walker played in the pre-humidor days of Coors field, spending nine and a half of his seventeen seasons with the Rockies at a time when their home park was extremely favorable to hitters As a result, Walker hit .348/.431/.637 at home in his career against 278/.370/.495 on the road. These numbers are often cited when his candidacy is being discussed, sometimes as a point against his enshrinement. More recently, park factors helped free agent Cody Ross cash in. The Arizona Diamondbacks gave Ross a three year/$26M deal following a season where he hit .298/.356/.565 and .232/.294/.390 away from Fenway. Prior to that Green Monster-induced frenzy, Ross had gotten just one year and $3M out of Boston the year before.
In both of the examples above, there is a subtle assumption that the road splits half of the equation is neutral. However, the baseball schedule is not balanced. Teams play their division rivals nearly twice as often as any other team and therefore spend twice the time in their rivals’ home stadiums. Larry Walker did not just play those 81 away games in some neutral stadium every year; he played 44% of those games in the other NL West parks, several of which are very tough places to hit. Cody Ross played 9 games each in the other AL East parks. Is this something we should include in our analysis and projection models or do these differences in road numbers even out, making the assumption of neutrality basically correct?
To answer this question and to better understand the impact of divisional play, I created a number of different "factors" for each team using park factors (care of Fangraphs) by handedness for team’s non-home games as they occur on the current unbalanced schedule. For each team, I created a "road factor" which is the total park factor impact of the team’s road schedule, a "divisional factor" which shows the team’s in-division park factors on the road and combined each team’s divisional factor and home park factors
Methodology and Caveats
To calculate these metrics I used as simple a system as possible. I first took the average of each division’s park factors to make a general "divisional factor" for each division. Next, for each team I took the average of the other parks in that team’s division to make team specific "divisional factors," in other words, the divisional factor without the team’s home park. Unless there are new stadiums or structural changes to stadiums, these numbers would remain consistent and exhibit the greatest influence on road numbers.
The next step presents a problem. The schedule teams play in not only unbalanced due to divisional play, but it is also unbalanced with respect to the league play and inter-league play. To calculate the "road factors" here, I use the divisional factors I created in the first step, thus giving equal weight to each non-divisional league team and then, for inter-league play, I simply assume a neutral environment. This means that the numbers here are estimates of a theoretical road factor, and not the road factor for any team in a given season. For example, next season, the Red Sox will play at least three road games against every AL team, but they will play an extra road game in Detroit and in Seattle. It is a small difference and over multiple seasons it will certainly regress toward the road factors presented here, but it is an important distinction. The same caveat goes for inter-league play. In 2013 the Red Sox will play games on the road in Colorado, San Francisco, Los Angeles and Philadelphia, which is a mix of both high and low offensive parks that actually ends up favoring hitters slightly.
The relative importance of those two differences is going to vary depending on specifics and it is certainly worth calculating a road factor based on a specific schedule, but the degree to which these small differences of schedule impact the numbers would only be statistically significant in an extreme case where the few variations all favor one extreme.
The numbers available on Fangraphs are for 2011 and therefore they do not reflect either Marlins Park or the change to the fences at CitiField, making the entire National League a bit murky. It would be a minor footnote if these two parks effects seemed to be neutral in the small sample that we have, but that is absolutely not the case. Marlins Park was extremely unfavorable to home runs and also suppressed doubles. The change at CitiField did boost home runs, but it may have reduced other extra base hits. In both cases, it is too soon to be certain what the long-term factors will be. Given the methodology I used here, this effects the entire National League, so I have calculated both the pre-2012 National League and an alternate based on the less-than-reliable one-year sample for these two parks, which is not specific to handedness.
The data is available here for review and further manipulation, but here are a few of the major takeaways.
Road Splits are not exactly neutral, but the variations in them are small.
The variations in singles, doubles, and home runs from both sides of the plate do tend to even out in the vast majority of cases and the most extreme difference in any area was just 7%. In the case of the Walker’s road splits, it is fairly reasonable to assume the park factors are basically neutral, given that his career spanned nineteen seasons and whatever small variations there were almost certainly evened out over such a large sample. The case of Cody Ross is not quite so clear cut as Boston’s typical road schedule does favor right-handed home runs by 2%, but that is still a very small effect in the grand scheme of things.
Park Factors for Triples are problematic
The low number of triples overall and the high level of variation in triple totals makes these park factors problematic. Even with the regressed, five year sample that Fangraphs uses to create the factors I used here, I still feel that triples are an issue. The case of CitiField makes this point clear. Prior to the fences being moved in, CitiField had a park factor of 107 for RH triples and 112 for LH triples by Fangraphs system and the one-year park factor for 2011 as calculated by ESPN was 121 for triples. The Mets also had Jose Reyes, who led the league with 16 triples that season. That was as many as the entire Atlanta Braves team. With the fences moved in and Reyes signing with Miami, the one-year park factor for triples at CitiField dropped to 63. Miami was fourth in the league in triples with Reyes on the team and his 12 triples were three times as many as his closest teammates. The new park in Miami subsequently had a park factor of 126 for triples for 2012. It is not unreasonable to suggest that triples cannot be accurately included in park factors thanks to their rarity and the ability of one player to swing the numbers so radically.
The AL East deserves its reputation as a hitter’s paradise
The parks of the AL East increase home runs and doubles on the road for teams in this division almost across the board. Home runs are particularly easy to come by in the AL East thanks to the launching pads of Yankee stadium and Camden Yards and the slightly less dramatic effect of the Rodgers Centre. The Red Sox and the Rays both play in parks that cut down on home runs, but the unbalanced schedule makes up for some of that with those extra games in New York, Baltimore and Toronto.The Green Monster turns such a large number of outs and singles into doubles, that it almost single-handedly ups the road factor for doubles for AL East opponents.
The NL East maybe the new NL West
The NL West has often been considered the AL East’s polar opposite, a division where pitchers can flourish in the large wide open spaces of Petco, At&T Park and Dodgers Stadium. That reputation may actually be exaggerated, as the average of NL West parks is basically neutral with the notable expectation of suppressing LH HR by 3%. A pitcher tossing home games at one of the three aforementioned parks is going to do well at home and face a pretty neutral road schedule which is nice, but NL pitchers might start loving trips East in the near future.
The early returns from the changes at CitiField and the new park in Miami point to a division that would be far more oppressive to offense than the NL West. In the new NL East environment, every team but the Mets can expect a 2% suppression in RH HR, a 3% suppression of RH doubles and a 1-3% supression in LH doubles. It is likely that this regress towards neutrality, but the effect may still be on par or even greater than the NL West as Marlins Park becomes Petco East and CitiField remains an offensively challenged place.
Singles are essentially neutral on the road
A 7% different in LH singles and a 10% gap in RH singles separate Petco from Coors Field, but once you begin fitting the effects to road schedules any difference in park effects for singles comes out in the wash, in large part because these two extreme are in the same division. Colorado has things the worst, with a road factor of 99 for singles from both sides, while the White Sox, Padres, Reds and Brewers all have a road factor of 101 from the right side. Everyone else is neutral.
Here are the most extreme variations between road schedules:
Doubles: The Rockies, Red Sox and Rangers RHH ‘Road Factor’ is 99 for Doubles while the Padres ‘Road Factor’ is 102. Seven teams have a ‘Road Factor’ of 101. From the left side, the AL East teams that get to play nine games a year in Fenway get a 2% boost in doubles, while six teams have a road factor of 101%
Home Runs: Home Runs are the stat most associated with a ballpark’s effect on the game and while the "Road Factor" effect is still small, there is notable variations. At the extremes, the Red Sox and the Rays gain some extra dingers on the road with Sox lefties getting a 3% boost and righties a 2% boost and Rays hitters gaining 2% from both sides. The most extreme negative effects are in the AL Central, where the Indians encounter a road factor of 96 for LH home runs and the White Sox face a 98 road factor for LH home runs and a 99 for RH home runs. The Reds have a road factor of 98 for righties while the Red Sox, Rays and Indians all have a road factor of 102 for RH homers. The NL West is a stingy division for home runs, but thanks to Arizona and Colorado the effect is not as dramatic as one might expect.
I am not confident that the effects of these differences are statistically relevant, but it is interesting to see that the road splits are not entirely park-neutral. If we begin to look at ‘road factor’ regularly calculated for specific schedules there is a good chance that some more impactful factors will emerge on occasion. At the individual player level, these factors are small enough to be ignored most of the time, but on team level, road factor is worth considering as a one or two percentage change can be relevant in that large of a sample.