The Nuts And Bolts Of BIP-Based Park Factors

Yesterday, we rolled out the latest update of my batted ball-based park factors, through the All Star break. Today, we’ll delve a bit more into some park-specific details.

How are such park factors calculated? In a nutshell, I’ve taken every batted ball hit in every park, applied major league average production for its exit speed/launch angle bucket, incorporated run values, and scaled the resulting projected production to an average of 100. Individual parks impact fly balls, line drives and even ground balls differently; the net of those discrete impact is the overall park factor. First, let’s look at the 2017 factors — for all BIP, and split out by BIP type — for the 30 MLB parks:

2017 Park Factors By BIP Type (thru 7/9)
FLY PF LD PF GB PF ALL PF
ATL 117.8 96.2 102.6 104.1
AZ 99.5 108.1 92.8 100.6
BAL 86.6 94.8 106.4 93.3
BOS 102.8 83.2 97.5 94.2
CIN 145.7 116.9 72.1 115.4
CLE 104.8 103.9 92.6 101.5
COL 172.3 117.9 104.1 131.1
CUB 108.5 97.1 84.3 98.6
CWS 100.2 103.1 108.8 104.6
DET 64.6 98.4 122.7 86.8
HOU 132.8 88.6 107.0 108.5
KC 72.6 103.9 103.6 91.7
LAA 92.5 76.1 94.2 86.8
LAD 98.8 100.2 89.2 97.0
MIA 83.2 108.3 99.6 96.8
MIL 129.3 103.8 116.3 115.3
MIN 99.7 96.5 115.0 101.7
NYM 93.0 109.4 99.0 101.1
NYY 94.8 102.9 93.7 97.7
OAK 96.2 90.7 75.4 90.2
PHL 97.7 100.2 97.9 98.4
PIT 108.3 108.0 69.5 97.6
SD 100.7 111.2 136.7 114.1
SEA 87.1 98.8 107.1 95.6
SF 68.2 92.3 85.1 82.0
STL 82.4 108.3 76.7 91.6
TB 103.1 91.8 149.9 107.7
TEX 109.1 91.7 105.3 100.4
TOR 109.4 102.4 99.2 103.9
WAS 95.9 103.2 120.0 103.8
MLB 100.0 100.0 100.0 100.0
STDEV 21.9 9.3 17.8 10.0

Color-coding is used above to note significant divergence from league average. Red cells indicate values that are over two full standard deviations above league average. Orange cells are over one STD above, yellow cells over one-half-STD above, blue cells over one-half STD below, and black cells over one STD below league average. Ran out of colors at that point. Variation of over two full STD below league average will be addressed as necessary in the text below.

It should be noted that the overall and fly ball park factors in particular correlate very well from year to year, minimizing the need for a multiple year approach. The midseason 2017 and full season 2016 overall park factors had a 0.62 correlation coefficient, 0.65 if you exclude the Braves, who changed stadiums. The average annual overall park factor correlation coefficient between 2013 and 2017 is a solid 0.59.

Fly ball park factors correlate even more closely from year. The midseason 2017 and full season 2016 fly ball park factors had a 0.70 correlation coefficient, 0.77 if you exclude Atlanta. While the correlation coefficients for grounders (0.42 from 2017 to 2016) and especially liners (0.33) are significantly lower, there is still a measurable degree of correlation. There isn’t as much randomness at play in these park factors as you might think.

Individual ballparks affect the various BIP types differently; they also impact the frequency of each type of base hit. The table below indicates the 2017 All Star break park factors for singles, doubles, triples and homers in each major league stadium:

2017 Park Factors By Hit Type (thru 7/9)
1B PK FCT 2B PK FCT 3B PK FCT HR PK FCT
ATL 101 110 65 99
AZ 94 115 152 96
BAL 106 82 52 96
BOS 107 102 64 74
CIN 89 112 155 143
CLE 96 116 59 102
COL 105 107 275 131
CUB 93 104 131 104
CWS 104 90 123 108
DET 106 94 136 70
HOU 96 104 69 127
KC 104 100 97 74
LAA 100 76 66 97
LAD 95 105 61 104
MIA 101 92 117 96
MIL 95 126 115 117
MIN 106 95 86 97
NYM 101 98 90 103
NYY 100 88 60 109
OAK 92 104 114 91
PHL 99 94 116 102
PIT 93 114 115 96
SD 104 101 141 115
SEA 102 85 53 105
SF 101 89 143 61
STL 99 98 107 86
TB 107 95 113 104
TEX 100 95 44 111
TOR 102 106 50 103
WAS 105 99 53 104
MLB 100 100 101 101
STDEV 5 11 48 17

The year-to-year correlation coefficients for home runs are the strongest among the individual hit types. The midseason 2017 and full season 2016 home run park factors correlated at 0.74; the average annual correlation coefficient from 2017 back to 2013 is also strong at 0.70.

Triples correlate almost as well, at 0.66 for 2017 vs. 2016; the average going back to 2013 is identical at 0.66. Singles (0.46 from 2017 to 2016, average of 0.37 back to 2013), and doubles (0.47, 0.42) also have correlated to a reasonable extent.

Some key numbers to refer to as we examine some of the individual parks:

MLB FLY BALL HR%
105+ mph = 84.0%
100-105 = 53.2%
95-100 = 16.3%
90-95 = 3.0%

MLB FLY BALL 2B%
95-100 mph = 12.9%
90-95 = 7.2%
85-90 = 4.0%

MLB LD 2B%
100-105 mph = 25.4%
95-100 = 19.1%

Next, let’s dig into some of the detail regarding some of the more notable ballparks:

– COLORADO – It’s no surprise that the most hitter-friendly park in the game is Coors Field. It’s the only MLB park that materially inflates singles, doubles, triples and home runs. In the air, the percentage of fly balls clearing the fence in the 100-105 (70.7%), 95-100 (24.6%) and 90-95 mph (6.2%) buckets are all notably high. Interestingly, seven of the 27 ground ball triples in all of MLB through the break were hit in Coors; all were hit by Rockies, five of them by Charlie Blackmon.

The following just about sums up the magnitude of the Coors Effect: overall, the batted balls hit there were the fifth weakest of those hit in the 30 MLB parks. The fly balls hit there were the second weakest. Those flies “should have” generated .302 AVG-.790 SLG worth of production; instead, hitters batted .388 AVG-1.050 SLG in the air, the most damage in the majors.

– CINCINNATI – Well, lookee here…..Coors is not the most homer-friendly park in the majors; Great American Ballpark is. It led in that department in 2016 as well, with a 134 home run park factor. Going back to 2013, it has never had a homer park factor lower than 121, and has never ranked lower than 7th among the 30 clubs in that department. Fly balls “should have” generated a .321 AVG-.837 SLG through the break, the 6th weakest flies in the majors. They actually went for .363 AVG-1.052 SLG, the third most production.

This homer haven outdid Coors in homer percentage in each of the most variable fly ball buckets; 75.6% on 100-105, 26.3% on 95-100, and 8.0% on 90-95 mph fly balls.

– MILWAUKEE – Miller Park is the stealth, go-to hitters’ park that isn’t necessarily seen as such. Only three parks — the three we’ve discussed so far — are materially inflating both doubles and homers thus far in 2017. You’re not going to hit many truly cheap homers here (only 1.8% of 90-95 mph fly balls have cleared the wall), but if you hit it well, it’s gone. 66.7% of 100-105 and 23.6% of 95-100 mph flies have gone for homers thus far this season.

Very quietly, the Brewers’ home has also shown a tendency to inflate ground ball offense. From 2015-17, 100+ park factors have been recorded on both fly balls (146.2, 115.1, 129.3) and grounders (126.4, 101.8, 116.3).

– SAN DIEGO – Petco is the most misunderstood MLB park. Most still consider it to be a pitchers’ park, but the record disagrees. While it’s graded out as very hitter-friendly thus far in 2017 and in 2015 (116.9 overall park factor), it would be overly simplistic to deem it as strictly a hitters’ park. The marine layer is a huge factor in San Diego, and effects swing wildly from year to year.

In the air, you get pretty average results from 100-105 mph (56.9% homer rate), and there aren’t many cheap 90-95 mph homers to be had (1.1%). There is a sweet spot in between, however, with a 22.4% homer rate on 95-100 mph fly balls. There are some soft spots down both lines for homers to clear the fence.

More interesting is the inflationary effects the park routinely has on both liners (park factors of 116.5, 101.3 and 111.2 from 2015-17) and grounders (114.3, 123.3, 136.7). Balls scoot through the infield and the gaps here. Check out the 2015-17 line drive double (110, 113, 117) and triple (107, 129, 206) park factors.

– HOUSTON – Minute Maid isn’t the most extreme hit-friendly park out there, but it sure does inflate homers. It’s not particularly friendly to line drives (the fences are close down both lines, allowing outfielders to play shallower), but its fly ball park factors have been high at 149.3, 114.7 and 132.8 in 2015-17.

This park might be the cheap home run capital of the big leagues. Its 25.9% homer rate on 95-100 mph fly balls is second only to Cincinnati thus far in 2017, but it doesn’t end there. An amazing 12.1% of 90-95 mph flies have left the building, and three of the only seven 85-89 mph homers hit in all of MLB through the break were hit here.

– BOSTON – Fenway Park has a reputation as a hitter-friendly park, and on balance, it is. It typically does not inflate homers, however, and it is very stingy on line drives. What it does do is dramatically inflate fly ball doubles.

Fenway’s fly ball double park factors has been astronomical at 181, 189, 205, 166 and 170 annually from 2013 to 2017. Its overall fly ball park factor is relatively low in 2017, as the fly ball homer factor has plunged to 77 after ranging from 93 to 104 from 2013 to 2016.

In all of MLB, fly ball doubles aren’t all that common, as the data at the top of this section suggests. Fenway is the outlier, thanks to the Monster. Thus far in 2017, 21.0% of 95-100, 18.7% of 90-95 and 9.4% of 85-90 mph flies have gone for doubles.

– LOS ANGELES ANGELS – Shifting gears significantly here to discuss some pitcher-friendly parks. Angel Stadium is death to line drives. Hitters “should have” batted .657 AVG-.871 SLG on liners hit here thus far in 2017, but are batting just .590 AVG-.732 SLG, a huge outlier. The magnitude is greater this season, but this park sapped liner production in both 2015 and 2016, to the tune of liner park factors of 92.9 and 93.8, and line drive double park factors of 95 and 86.

That latter figure has plunged to 66 this year. Only 16.8% of 100-105 and 10.9% of 95-100 mph liners have gone for two-base hits here this season. These figures further underscore what an incredible offensive player Mike Trout truly is.

– DETROIT – The homer-killing park. Comerica has logged homer park factors of 72, 86 and 70 from 2015-17. The hardest fly balls in the game have been hit in Detroit this season. Hitters should be batting .350 AVG-1.030 SLG on them, but are actually posting a .291 AVG-.811 SLG line, the 5th least fly ball production in the game. This underscores how great Miguel Cabrera has been for so long.

Only 63.4% of 105+ MPH fly balls have left the yard here. Cheaper homers are much harder to find as well, as only 33.9% of 100-105 and 11.9% of 95-100 mph fly balls have left the building.

– SAN FRANCISCO – Now that Safeco Field and Marlins Park have had their dimensions tweaked and become a bit more neutral, AT&T Park stands alone as the single most pitcher-friendly park in the game. It all traces back to homer prevention. From 2013-17, its homer park factors have been 71, 60, 84, 68 and 61. That’s dead last three times, and next to last once. While more 105+ mph fly balls (74.2%) have left the yard compared to Comerica, its 100-105 (30.3%) and especially 95-100 mph (5.6%) homer rates are lowest in the game.

One might suggest that the Giants’ light-hitting ways might have influenced their home park factor this season. That’s simply not true, as hitters “should be” hitting a near MLB average .331 AVG-.878 SLG in the air; the park has cut the production level to .297 AVG-.684 SLG, by far the lowest in either league.

We hoped you liked reading The Nuts And Bolts Of BIP-Based Park Factors by Tony Blengino!

Please support FanGraphs by becoming a member. We publish thousands of articles a year, host multiple podcasts, and have an ever growing database of baseball stats.

FanGraphs does not have a paywall. With your membership, we can continue to offer the content you've come to rely on and add to our unique baseball coverage.

Support FanGraphs




newest oldest most voted
Joe
Member
Joe

Is there any adjustment for defensive skill? It doesn’t sound like there is and while I was looking at the ground-ball specific factors, it appears there are huge swings in that category park-by-park. Because the infield is required to be the exact same dimensions, the only difference to me are (1) the type of grass & how it is mowed, and (2) the skill of the infielders (and maybe the third category of outcomes for ground balls that get out of the infield – which is FAR less often than those that are fielded in the infield).

I know that grass affects ground balls, but it doesn’t seem to me that it would be that big of a difference park-to-park, I expect the larger influence is how much range each home-team-infield has to get to ground balls versus other teams around the league. Please, correct me if I’m wrong on this, or it is just a small flaw in the “park-factor” calculation.

jfree
Member
jfree

Defensive skill is prob the biggest source of adjustment for GB’s. That means it would differ from year-to-year (low YoY correlation) but that is also why a ‘park-factor adjustment’ for GB’s in any particular year yields good info – since what will persist the next year (or in another park if a player is traded) will more likely be the adjusted data rather the raw data.

Joe
Member
Joe

I guess I’ve always thought of park-adjustments as adjustments for the BALLPARK itself, and not including the defense that plays in that ballpark…but, I suppose if you are trying to find the “true-talent” of a hitter, it doesn’t hurt to give him a bonus for playing against above-average fielders, or conversely, dinging him for playing against a bad fielding team. There’s a lot of noise baked-in to these assumptions, but I guess it does make sense. Maybe a different name than “park factor” is in order…possibly “context neutralizing factor” has a certain ring to it, right. Ok, maybe not, but it just seemed strange to me that the ground-ball numbers vary so much.

jfree
Member
jfree

The GB deviations from mean look widely variant. But my guess is that the actual absolute bip differences from mean are much narrower for GB than for LD or FB. A narrower distribution of data can actually produce what appears to be a large SD for an outlier. So even if data is adjusted, it’s not a big actual adjustment.

At least for the ‘red’ home teams – the SDP GB data looks like pure skill problem (a lot of chances to prove that yes they suck on IF defense). TB may be a combo of SSS and skill (not near as many chances – NLEast P’s produce much lower GB% than NLWest Ps do – and so far things look bad too).