Applying Ballpark Factors To MMDA Team Statistics

Customizing ballpark factors is one of the most difficult and agonizing decisions managers face each spring. Finding the most harmonious rate for single and home run frequency can add wins to a team's bottom line, but finding that perfect balance of offense and defense requires a deep understanding of player personnel—and also a lot of educated guesswork.

For years, I have endeavored to create ballpark factors for MMDA teams based on the statistical record, but until recently I had been stymied in my effort to find home/road splits. That changed when a fellow Strat obsessive pointed me to the line scores found within the League Stats menu.

With access to all teams' line scores, I summed all the runs a team scored and allowed at home, and then summed the same totals for each team on the road. Once I had those totals I could figure per-game rates by dividing by 81, and then compare those two rates to quantify the extent to which each MMDA park amplified or suppressed run production in a given season.

While I fully intend to apply this park-factor epiphany to the 2015 statistical record, I will do so only at the conclusion of the season, when teams have visited every park on their schedule and playing time has been fully distributed. Also, it takes some time to build the framework, so I only want to perform the calculations once per season.

However, this doesn't preclude me from revisiting the recent past and applying park factors to the 2014 data. From highest to lowest, here are the park factors I calculated for the 2014 season:

No	Team	Home R/G	Road R/G	Park Factor	Multiplier
1	Miami	8.86	7.99	111	1.055
2	Wilmington	9.33	8.68	108	1.038
3	Goldenrod	9.28	8.64	107	1.037
4	Death Valley	9.10	8.49	107	1.036
5	West Side	9.01	8.43	107	1.034
6	Elysian Fields	8.52	8.04	106	1.030
	Northeast	9.46	8.93	106	1.030
8	St. Louis	8.09	7.73	105	1.023
9	Brooklyn	10.17	9.79	104	1.020
10	Columbus	7.86	7.60	103	1.017
11	Rochester	8.42	8.22	102	1.012
12	Fighting Irish	9.05	8.95	101	1.006
13	Lake County	7.99	7.91	101	1.005
14	Jersey	8.51	8.51	100	1.000
15	Cape Cod	9.22	9.59	96	0.981
16	Rosehill	8.01	8.44	95	0.974
17	Mohawk Valley	7.83	8.47	92	0.962
18	California	8.40	9.65	87	0.935
19	Spokane	6.53	7.93	82	0.912
20	Northwest	6.75	8.40	80	0.902

The first two columns are self-explanatory. Home R/G accounts for all runs scored and allowed at each team's home park, while Road R/G is the same calculation for each team in road games.

The Park Factor column divides the home and road R/G rates and multiplies the quotient by 100 to arrive at an index score, where 100 is average. Notice how Jersey games in 2014 featured 8.51 R/G at home and 8.51 R/G on the road, thus the team's park factor of exactly 100. At the extremes, Miami's home park amplified run production the most (about 11 percent), while Northwest's suppressed output the most (about 20 percent).

Important note: MMDA teams play a wide-ranging distribution of road opponents based of the league's unbalanced schedule. Therefore, park factors are subject to distortion based on the fact that 44 percent of each team's road schedule is played in just four parks, those of the four divisional opponents.

The Multiplier column reduces by half the impact of the park factor—in order to reflect the fact that each team plays only half its schedule at home—and expresses it in a manner that can be applied to individual statistics. This last part is key. By taking any team's rate of runs scored per game, for example, and applying the park-factor multiplier, we can calculate an adjusted rate of runs scored that estimates how each team would perform if it played its home games in a neutral environment.

 Batting Statistics

Baseball-Reference applies park adjustments to its signature metrics, adjusted OPS+ and ERA+, and I will replicate that process below, but in a dumbed-down fashion befitting of someone without Sean Forman's aptitude for advanced mathematics.

Here are the park-adjusted OPS+ leaders for the 2014 season, where OPS+ is figured in the same fashion as the old Total Baseball model. That figure is then divided by the park-factor multiplier to produce the Adjusted OPS+. 

No	Team	Raw OPS+ (Rank)	Multiplier	Adjusted OPS+
1	Brooklyn	130 (1)	1.020	128
2	Fighting Irish	112 (4)	1.006	112
	Northeast	115 (3)	1.030	112
4	Goldenrod	115 (2)	1.037	111
5	Rochester	108 (5)	1.012	107
6	California	99 (10)	0.935	106
7	West Side	108 (6)	1.034	105
8	Cape Cod	102 (9)	0.981	104
9	Death Valley	106 (7)	1.036	102
10	Rosehill	98 (12)	0.974	101
	Northwest	91 (15)	0.902	101
12	Wilmington	103 (8)	1.038	99
13	Mohawk Valley	95 (14)	0.962	98
14	Jersey	97 (13)	1.000	97
15	Spokane	85 (18)	0.912	93
	Miami	98 (11)	1.055	93
17	St. Louis	87 (16)	1.023	85
18	Lake County	84 (19)	1.005	83
	Elysian Fields	86 (17)	1.030	83
20	Columbus	79 (20)	1.017	78

Because they play in the most extreme pitcher's parks in the league, California (+4), Northwest (+5) and Spokane (+3) receive the biggest boosts via adjusted OPS+, while Miami tumbles from 11th place to 15th after adjusting for park context.

One other fact becomes clear: Brooklyn (5.67 R/G), Northeast (5.02) and the Fighting Irish (4.89) truly did have the top offenses in the league last year. Even after adjusting for home park, they rank 1-2-3 on the OPS+ board.

Pitching Statistics

Now, let's apply the same principle to the pitching side of the ledger and adjust ERA+ by park-factor multipliers. Here's the Adjusted ERA+ leaderboard.

No	Team	Raw ERA+ (Rank)	Multiplier	Adjusted ERA+
1	Rochester	111 (4)	1.012	112
2	Mohawk Valley	114 (3)	0.962	110
3	Spokane	119 (1)	0.912	109
4	Fighting Irish	106 (5)	1.006	107
	West Side	103 (6)	1.034	107
6	Northwest	118 (2)	0.902	106
7	Death Valley	102 (7)	1.036	105
8	Northeast	101 (8)	1.030	104
9	Miami	97 (13)	1.055	103
10	St. Louis	100 (9)	1.023	102
11	Brooklyn	98 (11)	1.020	100
	Goldenrod	97 (14)	1.037	100
13	Elysian Fields	96 (16)	1.030	98
	Columbus	97 (15)	1.017	98
	Jersey	98 (12)	1.000	98
16	Rosehill	99 (10)	0.974	97
	Wilmington	93 (17)	1.038	97
18	Lake County	92 (18)	1.005	92
19	California	91 (19)	0.935	85
20	Cape Cod	83 (20)	0.981	81

Here, we see that while Spokane and Northwest had the top pitching staffs in the league based on the raw data, they fall back to the pack once adjusting for ballpark context. The Crows fall two places to No. 3, while the Stars drop four spots to No. 6.

In fact, the top pitching staff in the league last season belonged to Rochester, despite the fact that the Pilots ranked just fourth in unadjusted ERA (3.56) and ERA+. Mohawk Valley, in its pitcher's park, finished No. 2, but moved up one spot after the adjustment.