A Humidor In Chase Field Is A Big Deal

A few days ago, Alan Nathan wrote an article for The Hardball Times about the humidor’s potential impact on exit velocity in Chase Field.  He referenced a physics model that estimates a 3.8 mile per hour drop in exit velocity. He also showed that over the past two seasons the Diamondbacks Exit Velocity is 2 miles per hour higher in home games when compared to away games. I encourage you to read the whole article prior to reading what I have produced here today, as I will be building upon what he wrote.

Measurement Bias?

In the comment section of that Alan Nathan’s article I mentioned a potential batted ball calibration problem in Chase Field, and in fact all parks around the game, which may bias the readings up or down. At the time, I had the signs confused (addition versus subtraction), but the point stands.

Yesterday, I dug through my exit velocity data, and I found that batted balls in Chase Field, during the 2016 season, were roughly 2 mph *too slow* given their game results. In other words, if I raise the exit velocity for all balls in Chase Field by 2 mph, the expected batted ball success rates (xStats) had smaller error when compared to the real game stats. I tested 1.5 mph and 2.5 mph intervals as well, both had increased error.  xStats doesn’t have the granularity for precision greater than a half mile per hour, so narrowing down the number even further is impossible using this model.

Chase Field 2015-17
1B 2B 3B HR
Actual 2060 695 117 390
xStats Raw EV† 2042.4 636.0 122.4 294.5
Error -0.9% -8.5% 4.6% -24.5%
xStats EV +2‡ 2042.8 689.2 129.6 398.1
EV +2 Error -0.8% -0.8% 10.8% 2.1%

† Raw Exit Velocity from Baseball Savant

‡ Chase Field Exit Velocity increased by 2mph

I calculated the Diamondbacks home and away game average exit velocity to double check Alan Nathan’s claims. The data appears to be in perfect agreement; there is a two mile per hour increase in Exit Velocity during their home games.

Home/Away Exit Velocity
Year Home Away Difference
2015 90.0 88.2 1.8
2016 90.9 88.7 2.2
2017 89.9 88.2 1.7
Total 90.4 88.4 2.0

If I take this measured 2 mph difference, and the inferred 2 mph measurement bias I calculated using xStats success rates, I find a roughly 4 mph swing in Exit Velocity between the Diamondbacks home and away games. Give or take a half mile per hour. Again, Nathan’s physics model predicted a 3.8 mph difference, so our two models are in agreement.

My Model

With a humidor, much or all of this difference in velocity (ΔV) may disappear overnight, which will obviously play a large role in future offensive production. Since xStats treats every batted ball on a case by case basis, using exit velocity and launch angle to predict success rates, I can use it to predict how the stats may change, and which players may be impacted the most.

To run this analysis, I merely had to subtract 4 miles per hour from all batted balls in Chase Field, and then run my xStats algorithms on that modified data set. Consequently, I have adjusted stats for every batter who had even one single at bat (or faced a single batter) in Chase Field since 2015. However, I will be focusing primarily on the 2016 season.

Results

2016 Chase Field with -4 mph EV
1B 2B 3B HR
Observed 995 331 62 221
EV -4 Estimate 1006.7 298.1 57.4 142.8
1.2% -9.9% -7.4% -35.4%

As you might expect, the home run rate dropped substantially, by 35%. Which is almost embarrassingly close to what Alan Nathan initially predicted back in 2011.

At this point I should state that I ran this analysis many times, testing out many variables, including EV -2 mph (25% fewer home runs) and EV -3 (I lost the result, sorry). I tested raw EV -4 (raw meaning the number from Baseball Savant), which would reduce home runs by 45%.

In each case, I had home run rate drop between 20 and 50%. However, I believe the version I am sharing in this piece, with 35% fewer home runs, may be the most accurate.

How Diamondback Players May Be Impacted

Exit velocity may decrease overall, but that doesn’t mean every player is impacted equally. If your average home run is has a 108 mph EV, then the lost velocity may not make a huge difference. On the flip side, if your average home run has a 92 mph exit velocity, it could doom you. 

Lost Home Runs 2016
name HR xHR ΔHR Percent HR Lost BIP wOBAcon xOBAcon ΔOBAcon
Jake Lamb 19 9.0 -10.0 -52.4% 183 .505 .418 -.062
Brandon Drury 12 5.9 -6.1 -50.9% 180 .489 .385 -.104
Paul Goldschmidt 15 9.0 -6.0 -40.0% 211 .480 .574 -.062
Yasmany Tomas 16 10.1 -5.9 -37.2% 203 .427 .413 -.013
Jean Segura 12 6.7 -5.3 -44.1% 275 .438 .361 -.076
Welington Castillo 8 4.8 -3.2 -39.7% 147 .423 .383 -.040
Rickie Weeks 6 2.9 -3.1 -52.5% 52 .444 .409 -.036
Chris Owings 5 2.3 -2.7 -53.3% 157 .383 .367 -.016
Yangervis Solarte 4 1.6 -2.4 -58.9% 31 .535 .419 -.117
Ryan Schimpf 4 1.7 -2.3 -58.2% 14 .814 .586 -.228
David Peralta 3 0.8 -2.2 -72.0% 80 .429 .374 -.055
Derek Norris 2 0.3 -1.7 -85.1% 11 .673 .360 -.313
Chris Herrmann 3 1.5 -1.5 -49.8% 49 .478 .392 -.086
Scott Schebler 3 1.5 -1.5 -49.3% 13 .669 .528 -.142
Ryan Braun 2 0.5 -1.5 -72.5% 12 .408 .376 -.032

Unfortunately, according to this model, a few of these Diamondbacks hitters appear to fall into the ‘doomed’ category. Namely, Jake Lamb, who lost a jaw dropping 52% of his home runs in Chase Field. Considering he has hit nearly 60% of his home runs at home, that could be a 30% drop in his home run totals each season.

Brandon Drury is in a similar boat, although he hasn’t necessarily defined his value with the long ball.  Goldschmidt and Tomas should each lose roughly 30-40% of their home runs, which isn’t especially surprising considering the model predicts a 35% reduction in total home runs.

Lost Home Runs 2015-2017
Team HR xHR ΔHR Percent Lost
ARI 190 113.9 -76.1 -40.1%
Opponents 200 137.7 -62.3 -31.2%
ATL 8 5.2 -2.8 -34.6%
CHC 7 4.7 -2.3 -32.7%
CIN 10 6.9 -3.1 -31.3%
CLE 1 2.5 1.5 146.1%
COL 25 18.3 -6.7 -26.9%
HOU 9 6.2 -2.8 -31.0%
LAA 2 0.5 -1.5 -75.9%
LAD 26 18.2 -7.8 -30.0%
MIA 5 2.8 -2.2 -44.3%
MIL 6 4.3 -1.7 -29.2%
NYM 12 4.7 -7.3 -60.7%
NYY 2 2.2 0.2 7.6%
OAK 3 2.8 -0.2 -5.8%
PHI 5 4.0 -1.0 -19.0%
PIT 4 4.1 0.1 2.1%
SD 21 13.2 -7.8 -37.0%
SF 23 16.0 -7.0 -30.6%
STL 10 5.6 -4.4 -44.1%
TB 5 4.2 -0.8 -16.1%
TEX 2 1.5 -0.5 -25.3%
TOR 3 2.5 -0.5 -15.2%
WSH 11 7.4 -3.6 -33.0%

This table represents *all* of the seasons I have on record, 2015-2017. The Diamondbacks have 40% fewer home runs, while their opponents are down only 31%. This may look like a worst case scenario for the DBacks;  Not only are you limiting offense, but you’re disproportionately hurting the home team.

Those are the batters, though, how about the pitchers?

Lost Home Runs 2016
name HR xHR D Percent Lost
Archie Bradley 12 5.7 -6.3 -52.7%
Patrick Corbin 14 10.2 -3.8 -27.2%
Jonathon Niese 4 0.5 -3.5 -87.9%
Robbie Ray 11 8.0 -3.0 -27.6%
Dominic Leone 5 2.1 -2.9 -57.1%
Randall Delgado 6 3.6 -2.4 -39.3%
Braden Shipley 5 2.7 -2.3 -46.3%
Jorge De La Rosa 3 0.7 -2.3 -76.7%
Zack Godley 10 7.7 -2.3 -22.7%
Shelby Miller 10 8.2 -1.8 -17.9%
Edwin Jackson 2 0.3 -1.7 -85.8%
Chris Archer 2 0.3 -1.7 -84.3%
Tyler Lyons 2 0.4 -1.6 -81.1%
Josh Collmenter 3 1.4 -1.6 -53.9%
Casey Fien 2 0.5 -1.5 -76.0%
Matt Moore 2 0.5 -1.5 -74.5%
Adam Wainwright 2 0.6 -1.4 -69.2%
John Lackey 2 0.7 -1.3 -66.9%
Francisco Liriano 2 0.7 -1.3 -64.1%
Julio Teheran 2 0.8 -1.2 -60.5%
Carlos Estevez 2 0.8 -1.2 -58.3%
Jake McGee 3 1.9 -1.1 -37.3%
Chad Bettis 3 1.9 -1.1 -36.2%
Jeff Hoffman 2 0.9 -1.1 -53.9%
Mike Bolsinger 2 0.9 -1.1 -52.9%

Man, Archie Bradley may love the humidor, 52% fewer home runs!  A lot of these DBack pitchers are listed with 2 and 3 fewer home runs.  That is definitely nice, Bradley’s FIP, for example, would drop from 4.10 to 3.66. Having said that, we’re talking about pitchers saving 2-3 home runs while the offensive players are losing 4-6, each.

Finally, the team pitching numbers for the past three seasons.

Lost Home Runs 2015-2017
Team HR xHR ΔHR Percent Lost
ARI 200 137.7 -62.3 -31.2%
Opponents 190 113.9 -76.1 -40.1%
ATL 7 4.3 -2.7 -38.7%
CHC 8 4.0 -4.0 -49.9%
CIN 3 1.9 -1.1 -37.2%
CLE 2 1.1 -0.9 -47.3%
COL 22 13.0 -9.0 -41.1%
HOU 6 3.5 -2.5 -42.0%
LAA 2 1.8 -0.2 -8.9%
LAD 25 14.5 -10.5 -42.2%
MIA 3 2.2 -0.8 -25.4%
MIL 7 4.2 -2.8 -40.4%
NYM 9 4.8 -4.2 -47.1%
NYY 4 2.6 -1.4 -36.0%
OAK 2 1.6 -0.4 -18.7%
PHI 9 6.1 -2.9 -31.9%
PIT 9 4.4 -4.6 -51.2%
SD 22 10.7 -11.3 -51.1%
SF 26 20.0 -6.0 -23.0%
STL 9 5.9 -3.1 -34.7%
TB 5 1.2 -3.8 -76.4%
TEX 1 0.3 -0.7 -73.8%
TOR 1 1.0 0.0 4.0%
WSH 8 4.9 -3.1 -38.3%

Conclusion

If the Humidor is in fact installed in Chase Field, you should expect to see the Diamondback power hitters lose between 4 and 6 home runs, prorated to the remainder of the season. Which, for the purposes of this season, may be a reduction of around 3-4 homers, each.

On the flip side, you might expect their starting pitchers so give up roughly a third fewer home runs. Greinke may seem like an obvious beneficiary, but my analysis suggests he may not be helped very much. Instead, look to guys like Archie Bradley and Robbie Ray.

If the Diamondbacks install a Humidor, I imagine it would be paired with plans for moving in the fences.  It doesn’t make sense to install a Humidor without adjusting the ballpark. You may ask why install the thing at all if you’re just going to bring the fences in afterward. Think about it this way: it is very difficult to make a stadium larger.  You would need to move home plate or tear out seating and demolish walls.  Installing a Humidor is like pressing the restart button.  You get to make the field play as through it were larger, so you can then make fine adjustments to the dimensions to achieve the run environment.  Presumably, an offensively neutral ballpark.

 

Edit: I put more stats in a spreadsheet!  I know home runs do not tell the whole story, but it is hard to fit all of the stats into one article like this.  I encourage you to browse the spreadsheet.



Print This Post

Andrew Perpetua is the creator of CitiFieldHR.com and xStats.org, and plays around with Statcast data for fun. Follow him on Twitter @AndrewPerpetua.

newest oldest most voted
igloo
Member

Is there any affect on doubles? Maybe slight decreases in BABIP?