Park Factors & Park Factors & Park Factors, Oh My

February 12, 2018

We all know a park’s dimensions, foul territory, hitter’s backdrop, atmospheric effects, etc. play a significant role in shaping our projections and on a player’s performance. Collectively, we know these effects as park factors. We are probably most aware of a park’s home run park factor. I’m sure that for many parks, you have a perception in your mind as to its home run friendliness. The data might say otherwise, but at least you think you know, unlike, say, triples, which I’m sure most haven’t a clue which parks are best for boosting the three-bagger. Unfortunately, while the idea of park factors is sound, they are extremely problematic to rely on.

Why is this the case? Because the park factors you find from various sources don’t even agree with each other! Every source uses a different methodology to calculate factors and since the sources don’t always divulge all the fun math and logic behind their methods, you have to just put blind faith in its accuracy.

Park factors aren’t opinions though. Whether we have developed the most perfect way to calculate park factors or not, a park factor is a fact, like how many homers Aaron Judge hit last season. There should be a correct way to calculate them and many incorrect ways to calculate them.

So let’s get right to this problem. Here, presented in front of you, are the home run park factors, separated by batter-handedness, from three different sources — FanGraphs, Baseball Prospectus, and StatCorner. The FG factors are halved on the Guts! page, so I unhalved them to compare to the other two sets. If you want to dive into the methodology, here are the two sources that provide them:

FanGraphs
StatCorner

2017 HR Park Factor Comparison

Team	RHB			LHB			Diff, Large to Small
	FG	BP	StatCorner	FG	BP	StatCorner	RHB	LHB
ARI	108	109	120	108	110	106	12	4
ATL	90	91	88	108	108	109	3	1
BAL	110	111	117	122	106	108	7	16
BOS	102	94	101	80	86	77	8	9
CHC	110	104	112	88	115	85	8	30
CHW	110	102	113	116	97	120	11	23
CIN	112	102	105	120	108	120	10	12
CLE	98	94	99	110	106	110	5	4
COL	126	106	115	116	112	113	20	4
DET	100	108	101	104	98	101	8	6
HOU	102	102	105	106	103	108	3	5
KC	84	86	74	86	99	86	12	13
LAA	98	103	105	90	93	94	7	4
LAD	92	97	91	110	106	114	6	8
MIA	80	92	80	78	94	90	12	16
MIL	110	102	107	122	103	127	8	24
MIN	108	108	116	94	104	95	8	10
NYM	96	90	98	98	92	98	8	6
NYY	124	111	124	128	124	141	13	17
OAK	94	102	89	82	100	87	13	18
PHI	132	119	133	116	116	131	14	15
PIT	78	90	62	96	96	94	28	2
SD	90	93	97	92	88	86	7	6
SEA	96	94	98	100	101	108	4	8
SF	76	86	75	64	75	54	11	21
STL	88	90	82	94	98	98	8	4
TB	88	101	96	98	89	97	13	9
TEX	100	103	100	104	105	105	3	1
TOR	104	100	97	108	95	99	7	13
WAS	100	107	111	90	96	87	11	9

The three columns after the team abbreviation are the right-handed home run park factors from each of the three sources, followed by a trio of columns representing the left-handed home run park factors. The last two columns indicate the difference between the highest and lowest park factors for each team by handedness. Those are the focus of today’s discussion.

Ideally, there would be a correct way to calculate park factors and everyone would know it. Every site would be familiar with this correct method and publish the same park factors, just like every site publishes the same numbers of home runs Aaron Judge hit in 2017. But this clearly isn’t the case. This is far from the case.

Check out some of the differences between the highest and lowest factors!

We’ll start with the right-handed batter factors. PNC Park in Pittsburgh is either a standard home run suppressing park (according to BP) or possibly one of the most difficult places to hit a home run in over the entire history of baseball (according to StatCorner). How could the park factors in that one park vary so drastically?!

Coors Field in Colorado is the other park with a gap of at least 20 between top and bottom. That’s biazarre, didn’t we all agree that Coors was one of the game’s great home run parks? Well apparently BP disagrees. FanGraphs’ Coors park factor ranks it second in baseball behind Citizens Bank Park in Philadelphia, while the other two suggest it’s just another strong home run park, but doesn’t stand out.

Moving onto the left-handed home run factors, we find the biggest gap in all the land in Chicago, the North side to be exact. Wrigley Field is either a terrible place for lefties to mash one out of the park, or one of the best places. How could this park be so off? It’s not even like one factor is just over 100 and another is just below. The lowest is at 85 and the highest at 115! Something is wrong here.

It must be the Chicago wind, as Guaranteed Rate Field on the South side is another with a gap above 20. Milwaukee also joins that list either as fairly neutral or one of the best home run parks in the league, along with AT&T Park in San Francisco, where the factors can’t agree on exactly how torturous the park is for left-handed home runs.

Another interesting observation about these park factors is the highest and lowest from each source, and the gap between those two marks. Let’s compare:

2017 HR Park Factor High-Low Gap Comparison

	RHB			LHB
	FG	BP	StatCorner	FG	BP	StatCorner
Max	132	119	133	128	124	141
Min	76	86	62	64	75	54
Gap	56	33	71	64	49	87

So for both sides of the plate, Baseball Prospectus’ home run park factors have the narrowest range, StatCorner’s the widest, and FanGraphs is in the middle. I have no idea what this means about each source’s numbers.

Between my Pod Projections and my xMetrics like xHR/FB rate which drive my HR/FB forecasts, I use park factors extensively. But if every source calculates different factors (sometimes wildly different) for the same statistic, they become completely useless.

Let’s brainstorm, pick apart the methodology, and come up with the best, most accurate park factors we can use for all our future projecting activities and backwards-looking analysis.