Aaron Hill Follow-up
There were so many good questions raised in the comments section of my post yesterday on Aaron Hill that I thought addressing some of them deserved a follow-up post. The bulk of the questions come from former BtB overlord Sky Kalkman:
Would love to see some further breakdowns. Did Hill see more or fewer pitches over each part of the plate (or outside)? Did he see different types of pitches, and different locations of different pitches? Did he swing at certain pitches/locations more often or less often? Did pitchers’ approaches change throughout the year?
Addressing Sky’s first point:
Yes, it does seem that pitchers pitch farther outside to Hill. The peak location is a good third of a foot farther away from Hill than to the average righty. Based on the results of yesterday’s post, this is a good idea: Hill crushes those inside pitches.
Skipping to Kalkman’s question about whether Hill swings at certain locations of pitches more than others:
Generally Hill has a higher swing rate than average righty and this is partially true for middle-in pitches, but as pitches get farther away from him he swings at about league-average rate. So it does look like Hill swings more often at the inside pitches that he has the most success with.
Sky’s last question was, “Did pitchers’ approaches change throughout the year?” This is very interesting. If you look at Hill’s FB% numbers you can see that he has been seeing fewer and fewer fastballs since 2006. But if you split it out by month for 2009:
March/April .525 May .550 June .519 July .572 August .596 Sept./Oct. .590
For some reason, in July, pitchers reversed this trend. After he had already crushed 19 HRs through the first half of the year pitchers started throwing fastballs to him more often. I poked around some more, but could not find a big change in where those pitches were thrown. This is an interesting trend and something that could be looked into further.
Ewan, another commenter, suggests:
One thing I noticed about watching Hill a lot last season is he is very good at fouling off pitches on the outer half for someone his size and who used a heavier bat. Because of this he almost forces the pitcher to throw him something a bit more hittable.
Here is the fouls per pitch by horizontal location:
This does not look to be the case. Hill does foul off a lot of inside pitches — which makes sense because he pulls them so much — but on outside pitches he fouls off pitches at about a league-average rate.
Finally, Vivaelpujols had a great suggestion:
Maybe you could also chart the vertical location in the strike zone, as a function of the depth of the plate. So home runs on pitches around 3.5 pz would extend to the furthest part of the square part of the plate. Home runs on pitches at 1.5 pz would only extend to the front of the plate.
What do you all think? I like it. Most of his HRs are in a fairly narrow band, but he has that one HR that was relatively up in the zone.
Thanks to all the commenters for their insightful observations and questions. Obviously even this still just scratches the surface. I have ignored an analysis of the height of the pitches, beyond this most recent HR graph; looking at performance and location of each pitch type separately; and a host of other questions.
Dave Allen's other baseball work can be found at Baseball Analysts.
The outlier homerun is truly an outlier. It was a Yankee Stadium RF homerun and wouldn’t have been an homer in any other ballpark.
http://www.hittrackeronline.com/detail.php?id=2009_52&type=hitter
Yeah Jessef, I address that in the yesterday’s post. Here I was talking about the one HR up in the zone (closer to the bottom of hte plate on the diagram), not the opposite field one.
Yeah, just saw that now. Sorry about that one!
No problem Jessef, just clearing it up.
One thing that I’m not sure if it’s been covered is if he’s hitting fastballs almost exclusively for home runs or is it hanging breaking balls? Also, if he’s hitting fastballs is he hitting them off of weak fastballs or is he cheating and pulling inside fastballs from guys like Verlander?
The latter case seems unlikely as it should leave him very vulnerable to change-ups but he straight-up murdered change-ups last year. It looks like the only way to get him out is to pound him with breaking balls.
Yeah, good question. If he’s not hitting fastballs for home runs, give him more, right?
In 2009 his HRs were:
19 off fastballs
7 off changeups
5 off curves
4 off sliders
1 off a cutter
Wow, knowing that, why would you ever throw him a fastball above his waist? I’d pump fastballs low and away all day and get easy fly outs. It’ll be interesting to see if the league adjusts to him this way.
Keep in mind those HR by pitch numbers do not reflect the frequency that they are thrown at.
HR% – Pitch Type – Pitch %
53% – FB – 56%
19% – CH – 11
14% – CB – 11
11% – SL – 15
3% – CT – 5
Sounds like he launches change ups (shocking for a pull hitter) and curve balls.
Great point Steve.
If you look at the color-coded HR chart below you will see that those changes and curves were mostly away, and they represent the bulk of his HRs on pitches on the outer half of the plate.
If only I had those graph making skills…(a couple great datasets would be nice too, but I hardly feeling like building one after working 7:30-4:00 everyday doing data entry). I like the response to the comments with a follow up post. Keep up the great work.
The pitchf/x database is available here.
Beautiful. Do those graphing skills come in a database too? 🙂
My readers will hate this.
On the last graph, I like the idea of showing the pitch height in that way in theory, but in practice it is very hard to read. Perhaps it is harder to read since they are all clustered around the same vertical point generally, but i’m not so sure that is the reason. It would be interesting to see something like this for someone who hits HRs all over the plate to see if this is the issue. Either way, one solution I can think of is to make the lines over the plate transparent and mark the points of contact with dots, although this sounds quite complicated.
Good call, this makes it much more readable and shows lots of HRs on low pitches.
Dave, one more think could make this perfect. Actually two. 1) You should make the plate show numbers like a graph (so on the sides, it’s .5-4.5 or whatever, and at the tip of the plate it’s -2 to 2. That way it would be easier to see the actual data, instead of just inferring from the visual (and it would do that too). Secondly, you should color the dots on the plate differently by pitch type, following the FanGraphs color scheme.
Thanks a lot for listening to my last suggestion!
And you should make the lines colored by pitch type to help connect the dots.
14 minutes, Dave? You’re slipping.
This might really just be me, but I don’t have the FanGraphs pitch F/X color scheme off the top of my head, and I didn’t really have any success finding a good example in your recent articles. Having some kind of internal legend would be great.
Also, you’re basically God, Dave. Awesome stuff.
Sky, no one is more disappointed in me than I am in myself.
Aah, Dave you are my hero. Just two more things ;). You should put the color scheme key on the bottom right hand corner of the chart, and instead of doing y=, just put the 3.5 ft, and then Y as the axis label. Then next to the color key, put y = vertical, x = horizontal. thank you.
Have you considered trying a view angle like Gameday where the location is pinpointed in a strike zone box? We wouldn’t get the angle of the homerun, but it would make a nice complement to your graphs.
Perhaps executing that is impractical?
Thanks, Dave. Appreciate the excellent analysis and responsiveness.
On the fouls graph, might it make more sense to do fouls per swing, not fouls per pitch?
Sky,
Yeah I played with the idea (in my head) of going with fouls per swing or fouls per contact. For no reason in particular I settled on fouls per pitch. Anyway the differences are much more drastic on fouls per swing:
Wow, Dave bringing 3-D to that last chart is pretty impressive. Kudos.
Hill also tended to be very aggressive with fastball early in the count last year. Any chance you could look into that a bit more? I rarely remember him taking a fastball for a strike on the first pitch last season.
I’m not sure that I like the approach of plotting vertical height as distance towards the front or back of the plate. Every time I look at the chart I need to force myself to not interpret the origin of the line as the depth of the pitch in the strike zone across the plate. So that one outlier homerun at first blush causes me to think how he pulled a ball to left when the pitch was that deep in the zone.
If I have to continually force myself to adjust my thinking on a graphic, it causes me to rethink the graphic.
If you concur in this, my thought would be to gradations of color to indicate the vertical location in the zone. For example, a pitch on the ground could be red, a pitch six feet high could be blue, and everything in between scaled on a red-to-blue spectrum.
That would also make it easier to evaluate the lines themselves. For example, right now if one wanted to assess whether he pulls low pitches more than high pitches, you have to start tracking individual lines back to the origin. Using a color scheme it would be obvious, as red would predominate on home runs hit closer to the foul pole.
Still not sure why I got no credit for brining up Aaron Hill’s power splits in the first place.
Just because (A) you brought up Hill’s splits and then (B) Dave wrote about it does not mean that A caused B. It is the introductory week for splits, after all, and it’s natural that readers and writers would be interested in some of the same players(Hill, Mauer, etc).
I love how Aaron Hill is such an intense case-study here. haha
Personally I think that we should petition Dave Appelman to code in an interactive set of these graphs to let people see the dimensions they want. Allow people to look at pitch type, height in the zone, etc. along with greying out unnecessary lines and the like to see what they want better.
If we do need to stick to a 2D graph, then could we change the plate to a regular XY coordinate graph of where in the zone the pitch was? Adding a batter standing next to the plate (to show the strike zone) would help the visualization (but muck it up for switch-hitters).
With respect to the increased rate of fastballs after June:
Is this perhaps explainable in terms of the quality of pitching Hill was facing later in the season? Presumably, Hill was seeing a greater percentage of less experienced pitchers as the season progressed. Perhaps the jump we see in the second half of the season has to do with pitchers who either didn’t know how to pitch to him, or else were not confident enough to stick with a set game plan.
I realized you likely don’t have the data available to offer a quantitatively concrete answer to this. But it’s a hypothesis that dissolves the apparent anomaly you cite in the post.
I’m not 100% sure on that, but my gut reaction is that it’s not right. The Jays offense as an whole was very good through mid-to-late May and then crashed hard in June and through the summer. They also played outside the division (weaker teams) at the beginning of the season and started facing AL East teams later on.
Dai,
Good call. The these data are actually available on FanGraph: Under Teams, then pitch type and then split by month.
So just like Hill the Jays saw more fastballs as the year went on (I wonder if this is true for all teams?). But Hill’s jump in fastballs seen, 53ish% pre-All Star break to 58ish% after, is a little bit larger than for the Jays as a whole.