Asking Brian Dozier About His Missing Home Runs

Last year, Brian Dozier hit 42 home runs for the Twins. He’d never hit 30 before that, so it was fair to wonder if the power was repeatable. Earlier this year, he was on pace for a total much more similar to his career norms… and yet some of the factors we used to look at last year’s home runs suggest that he should have recorded more homers this year already. Notably, Andrew Perpetua’s xStats metric indicated last week that Dozier had an expected home-run tally that was four homers higher than his actual number.

I was curious about it, so I decided to go directly to the source. Before a recent Twins-Giants game in San Francisco, I made a trip to the visitor’s clubhouse with video of the longest fly balls that Dozier has hit this year. “Where are those missing home runs?” I asked him.

In what follows, I’ve presented video — the same video I showed to Dozier — of fly balls he’s hit this year that translate to home runs with regular frequency. With each video I’ve included both exit velocity and launch angle. I’ve also included “home-run percentage,” an estimate (calculated by Perpetua*) of how often a similar batted ball would become a homer.

*For HR% and expected outcomes, Perpetua takes the exit velocity and launch angle and spray angle (pull vs push), and then draws “concentric spheres around that batted ball” (based on EV and LA again) and compares within those spheres, weighting each sphere by their size. The largest sphere of comparison is very large but has the least weight.

April 24th vs. Keona Kela in Texas
Exit Velocity: 95 mph
Launch Angle: 28.7 degrees
Home-Run Rate: 24.5%

Video

Dozier’s Comment
“I thought it was going to be a double, for sure. Fastball up, in the ninth. It was more of a line drive. It didn’t have enough launch angle, to go Statcast for you.”

*****

May 18th vs. Chad Qualls in Minnesota
Exit Velocity: 103 mph
Launch Angle: 28.4 degrees
Home-Run Rate: 46.9%

Video

Dozier’s Comment
“That was at Target Field. That was to center. I thought that it was gone. But I’m telling you, man: left-center to right-center at our place, I’ve seen many of those.”

*****
May 19th vs. Nate Karns in Minnesota
Exit Velocity: 97 mph
Launch Angle: 34.3 degrees
Home-Run Rate: 20.3%

Video

Dozier’s Comment
“I was at home. That one went like 430 feet. Oh, yeah. Hit the right-field wall. Anywhere else, that would have been gone. I’ve seen that wall gobble it up plenty of times.”

*****

May 29th vs. Jordan Jankowski in Minnesota
Exit Velocity: 100 mph
Launch Angle: 38.8%
Home-Run Rate: 31.6%

Video

Dozier’s Comment
“Wish I remembered that one.”

*****
June 3rd vs. Matt Shoemaker in Anaheim
Exit Velocity: 102 mph
Launch Angle: 40.3 degrees
Home-Run Rate: 22.4%

Video

Dozier’s Comment
“That was the other day in Anaheim. In daytime, it’s not bad, but that was a night game. Day games, it flies; night time, it levels off. I thought I might have hit it too high, especially in that park. Off the bat, I thought I did hit it out, but then it kept gobbling up air. Then I thought, ‘Oh yeah, probably not.'”

*****
I used my eyeballs to pick these balls, stupidly, so we don’t have comment on the three balls in play that were the closest to being homers, so let’s just look at those really quickly.

April 3rd vs. Danny Duffy in Minnesota
Exit Velocity: 100 mph
Launch Angle: 34.7 degrees
Home-Run Rate: 71.0%

Video

April 22nd vs. Matt Boyd in Minnesota
Exit Velocity: 102 mph
Launch Angle: 21.5 degrees
Home-Run Rate: 58.0%

Video

May 21st vs. Ian Kennedy in Minnesota
Exit Velocity: 102 mph
Launch Angle: 24.9 degrees
Home-Run Rate: 66.0%

Video

What do so many of these balls have in common? “Welcome to the AL Central,” said Dozier with a smile. “We’d have a few players with 30 or 40 or more if we played in a different division.” So many of those balls would have been out in other parks, practically by definition. “In different parks, when you can go the other way and leave the park, it’s great. You can’t in Minnesota. I’ve seen 430 footers to right-center hit the wall.”

The park is huge, and it molds Dozier’s approach. All those pulled homers that were right down the line and might not have been homers last year by Statcast were on purpose. “I’m a firm believer — and I’ve played in almost every park — the ones that you play in the most, it molds your swing. If you’re a guy that provides power, and not at an elite-elite level, then you find ways to get it to the shorter parts of the field.”

So the shape of a park might alter a hitter’s swing; that seems reasonable. But the park is also the source of error when it comes to our ideas of expected home runs. To that end, Perpetua spent the weekend correcting for some park effects when it comes to exit velocity. It’s not captured exactly the same in each park, so there are slight adjustments you could make to the reported exit velocities in those parks. After running two checks, Perpetua has settled on the ‘new exit velocity adjustment’ below in order to normalize each park to the same level.

Park Factors for Exit Velocity
Park Old Exit Velocity Adjustment New Exit Velocity Adjustment
Houston -0.96 1.41
New York (N.L.) -0.94 1.37
Arizona 2.00 1.36
San Diego -0.45 1.15
Chicago (A.L.) -0.21 0.74
Miami 0.09 0.50
Tampa Bay 0.10 0.43
Cincinatti -0.96 0.38
St. Louis -0.60 0.38
Anaheim -0.19 0.35
Atlanta 0.02 0.02
Toronto -0.05 0.01
New York (A.L.) -0.14 -0.01
Minnesota 0.21 -0.01
Pittsburgh 0.11 -0.04
Texas -0.04 -0.07
Washington -0.20 -0.08
Seattle 0.05 -0.08
Boston -0.02 -0.10
Cleveland 0.32 -0.14
Colorado -0.37 -0.16
San Francisco 0.39 -0.19
Los Angeles 0.26 -0.20
Milwaukee -0.02 -0.22
Oakland -0.02 -0.23
Philadelphia -0.02 -0.27
Chicago (N.L.) 0.18 -0.37
Kansas City 0.46 -0.54
Baltimore 0.98 -0.58
Detroit 0.85 -0.69
Perpetua checked to see if home/road velocity difference matched up with what a physics based model would predict based on temperature and altitude differences, and when it missed the mark by too much, he adjusted it as you can see in the table above under “old”. The “new” adjustment was based on a second check, when Perpetua went through and compared velocities on similarly hit balls in different parks and made adjustments based on that comparison.

The parks themselves affect the exit velocity readings. But it also ignores the park effect that’s costing Dozier home runs — namely, the sheer size of the parks in which he plays. Attempting to adjust for that sort of thing is more complicated than one might expect, however. For one, park effects can be hard to separate from other factors. “The difference between parks is generally subtle, so it is hard to tell [if] this ball fell because of bad defense or park effect?” Perpetua pointed out. “Was that ball a home run due to spin rate or wind or is it the park?”

There are a lot of variables, in other words.

And then there’s the problem of sample, too. In year three of the Statcast Era, we have nearly 13,000 homers on record. That sounds like sufficient volume of data. But when we start dividing by league and park and area of the field, the sample gets very small. And then, of course, there are those home fields that have either changed entirely (Atlanta) or been subject to material alterations (Colorado, Miami, New York, San Diego, Seattle).

How about another approach? Let’s go back to that hit off of Karns from above. Perpetua made a chart of all the balls in baseball that have been hit similarly by launch angle and exit velocity. It’s a decent sample. The result? Similar batted balls have been homers 20% of the time, which is generally how Perpetua arrived at the home-run rate listed above. (Dozier’s ball is denoted by the red mark.)

Want to check that against only balls that were hit similarly in Minnesota? Here’s your sample, with Dozier’s ball in red again.

It’s too small a bin to inform our park-effect-adjusted expected home-run percentage. So for now, Dozier just has to grin and bear it. “This is just the way it is, it doesn’t bother me,” he said of Target’s intricacies.

So Dozier has lost a few homers to his home park. Even if “they definitely don’t even out,” Dozier isn’t concerned about the general pace of things. Those homers last year? “For the most part, a lot of them came during the hot summertime, a lot of day games,” he pointed out.

At this point last season, Dozier had seven home runs. He has 10 now. The weather is turning, and he can still turn on a pitch and take it down the line in Target Field. “That’s pretty much it,” as the player put it.

We hoped you liked reading Asking Brian Dozier About His Missing Home Runs by Eno Sarris!

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




With a phone full of pictures of pitchers' fingers, strange beers, and his two toddler sons, Eno Sarris can be found at the ballpark or a brewery most days. Read him here, writing about the A's or Giants at The Athletic, or about beer at October. Follow him on Twitter @enosarris if you can handle the sandwiches and inanity.

newest oldest most voted
Daniel King
Member
Member
Daniel King

It’d be nice if statcast could pick up the spin on the ball as it leaves the bat, hitting a curveball is going to go farther than a fastball for the same exit velocity and launch angle.

mikejunt
Member
Member
mikejunt

My understanding is that it does and players and teams have this data but it isn’t public.

Andrew Perpetua
Member
Member
Andrew Perpetua

It can, and does, measure spin rate. But only total spin rate, so you don’t know if it is side spin or back spin. This data isn’t public.

Also, if you think about it, spin rate on batted balls is hard to correct for measurement error. With a pitch, you know a pitcher tends to throw within a small range. Maybe all of their fastballs have 2200 rpm give or take a few hundred. Maybe 2050-2350 or something. So if you see a fastball with 500 rpm, you know the data is wrong, and you can ignore it. But with batted balls, how do you know how much spin a batted ball is supposed to have? It is a difficult problem to solve, so batted ball spin rate data is less reliable.

Jetsy Extrano
Member
Jetsy Extrano

Doesn’t Statcast get the whole trajectory? That should be enough data to back-estimate initial spin and at least a constant wind, though variable wind gets tough.

Andrew Perpetua
Member
Member
Andrew Perpetua

You can estimate the spin angle, but if you want to directly measure it you need two pieces of radar.