Matt Harrison's Last Start in Pitch f/x
by Desert Ranger on May 1, 2009 12:41 AM CDT
31 comments
A look at Matt Harrison's monday start using pitch f/x
First and foremost, this is my first post on the board though I've been reading for a while, so I'd just like to say hello. I was very happy to see there are people out there who care about the Rangers as much as I do.
Now then, since that's over, I was going over all the info from Matt Harrison's last start, Monday against the Orioles. His first 2 innings were rough, to be generous (50 pitches), and after that he seemed to settle down (63 pitches over 5 innings). I read everything everyone suggested as to why he settled down and I thought I'd take a look at the Gameday data to see what it suggested.
The first suggestion, put forth by both Harrison and Mike Maddux, was that he was rushing and leaving pitches up in the zone as a consequence of picking up his tempo. After the second inning, he slowed down and pitches started to go down in the zone.
via i41.tinypic.com
Now from this graph it doesn't appear that he did any better after the 2nd inning of keeping the ball lower in the zone, but at the same time maybe he got a few bad hits off balls that were left high in the zone early. I haven't looked into that yet but will if you guys want it.
A few people were asking about pitch velocity as the game went on, so I looked into that as well. I plotted only four seam fastball velocity for the time being.
via i43.tinypic.com
This shows that he kept an average fastball velocity of about 89mph (edit: actually 92mph), but he did absolutely bring it a few times, maxing out at 96. It also shows us that he was able to bring his velocity back up late in the game, even after 100 pitches.
Next, I did a simple graph of pitch selection by inning. I'm a little questioning of some of their classifications; they have Harrison throwing a four seam (FF), two seam (FT), cut fastball (FC) (only 3 of these, I think they are actually changeups), changeups (CH), curves (CU), and sliders (SL).
via i42.tinypic.com
He never got crazy throwing off speed pitches, even in the middle innings when his fastball lost a little life. He obviously threw more fastballs in innings where he was having trouble, but he also threw more pitches in general. What sticks out to me is that he threw more 2 seam fastballs in the longer innings as well. This bears investigation.
Here's the graph of his pitch movement. Sorry for the trouble with some of the colors, excel doesn't give you great options.
via i44.tinypic.com
Curves are off on their own, sliders and what Gameday calls "cut fastballs" are grouped together, and then there's that shotgunned area towards the upper right. Those are 4 seam fastballs, 2 seam fastballs, and changeups... and they all move the exact same way. Average 4 seam velocity is 92.0mph, average 2 seam is 84.9, and average change is 83.1. If we assume that those 2 seamers are actually changeups and gameday just can't tell the difference, then the average "changeup" is 84.5mph. Either way, it just looked like a slow fastball with no movement difference.
Maybe I'm reading too much into it, but it looks to me like once he stopped relying on his secondary pitch (2 seam, change, whatever it was), he was able to improve. I'm probably just reading too much into it though.
Sorry for the long first post, and if you read this far, thanks. If you want any more graphs from this game let me know, I'd be happy to help. Pitch f/x is more exciting than my actual job.
8 recs |
31 comments
|
Email
Print
Comments
Kind of nit-picky...
Almost all of his FF velocities were above 89 mph. How does that wind up with 89 mph as the average velocity?
Good catch
I don’t know what happened there, but you’re right, it should be 92.0mph.
Thanks.
by Desert Ranger on May 1, 2009 1:00 AM CDT up reply actions
Tip
In Excel you can set the intersection point for the axes. Format Axis→Scale… somewhere on that tab, it asks where the intersection point is. You can use that you get the Y-axis out of the middle of the plot area.
Haha
Again, thanks. This is why I shouldn’t do excel late at night without double checking things.
by Desert Ranger on May 1, 2009 10:18 AM CDT up reply actions
Wow.
Why can’t more first time posters be this good? Interesting data, thanks DR.
"That start was like somebody on a deathbed suddenly jumping up and doing the Lindy Hop, then speeding away on a pogo stick while playing the fiddle." - Telegraph
Great first post.
One thing I did notice on your graph of pitch locations: Harrison had about seven or eight pitches right in the middle of the plate during the first two innings, and one after that. A 92 MPH fastball isn’t anything to sneeze at, but I don’t think Harrison can really get away with pitching in that part of the strike zone.
Adding to the chorus
Great first post. I must say that posts like this and Ghettobear’s right below it are awesome. But I’ve noticed that there are many fewer comments than on most fanposts because most of us just aren’t sure how to interpret all this data. I know I sure don’t understand or know exactly what to make of all this stuff. I don’t know nearly enough about either pitching or statistical eval to make anything of it.
I and I’m sure many others would appreciate those of you who know a lot about those things dumbing it down for us. I’m not ashamed to ask.
G G G E-flat_______ F F F D__________....
Out of all these...
I think the pitch movement is the most confusing one, in large part due to MLB’s decision to rate vertical movement differently than horizontal movement. It really, really bothers me.
What questions do you have about the graphs?
Pitch movement graphs
They definitely are one of the more challenging ones. I’ll try to explain it a little better here and probably update the post once I get a chance.
So, if the pitch is in the positive area horizontally, it moves away from righties or towards lefties that many inches during its trajectory. It doesn’t take into effect where it starts or ends, only how much it moves. If it’s in the negative half horizontally, it moves in to righties, away from lefties.
Vertically, if the pitch is in the positive half it means it “rises”, although this is tricky because it just means that there is back spin on the ball causing it to stay up. A pitch that’s negative vertically drops, such as curveballs, which makes sense because they have top spin and therefore dive to the ground.
I can do a more in depth pitch f/x explanation if need be, though I’m still learning too.
by Desert Ranger on May 1, 2009 10:17 AM CDT up reply actions
I understand the concept of positive vertical numbers.
It just bothers me that (essentially) a theoretical pitch completely unaffected by gravity is rated with positive vertical movement on this scale. Not only is their use of positive numbers in a context weighting vertical movement physically impossible given the angle pitchers throw baseballs, it makes it that much harder to know how much true sink exists on a particular pitch.
You did a good job explaining it here — this gets on my nerves on principle because I’m a programmer and a technical writer, and this just reeks of some absurdly stupid group-think.
I think you've misunderstood the vertical component.
First of all, there’s no such thing as “unaffected by gravity.” All pitches are affected by gravity. The vertical component literally measures “true sink” – what the pitch does compared to how gravity alone would affect it.
If want to know how “big” a break is, compare the distance of two pitches on the movement chart. If a fastball is at -10, 10, and a curveball is at 5, -5, then you know that those two pitches, when thrown with the same initial trajectory, will be at least 21 inches (by pythagorean theorem) apart by the time they hit the mit (not counting for the extra time gravity gets to affect slower baseballs).
by NoNameOnCard on May 1, 2009 12:54 PM CDT up reply actions
I understand the vertical component.
I wasn’t clear when I said this originally, but I intended the “physically impossible” statement to refer to the fact that it’s impossible for pitches to rise or not have downward vertical movement.
I’m just saying this: if the vertical portion were on a negative scale which simply represented the number of inches a pitch dropped from release point to catcher mitt, I think it would be far easier to interpret.
That's close enough, I suppose.
Would overhand/three-quarters release points vary the height enough in conjunction with pitcher height that it shouldn’t be constant? (I seriously don’t know the answer to this question)
It's all relative I suppose.
Release point data is roughly captured by PITCHf/x, and it does vary from pitcher to pitcher.
Arm angle has more to do with the release point than a pitcher’s height does. Jeff Nelson was like 6’ 9" or so but he threw sidearm so his release point was probably just under 6’. Brandon McCarthy is about 6’ 7" and his release point is well over 7’ because he throws more over the top.
That isn’t to say that height doesn’t matter, though. There’s basically no way that Danny Herrera will ever have a release point anywhere near 7’…
To really put pitch movement into perspective, you might want to set your brain up to recognize the main fastball (be it 4-seamer, 2-seamer, or sinker) grouping as your mental origin (0,0). From there, the location of the other pitch types on the movement chart should start to make a lot more sense.
Of course, this will inadvertently give you the impression that the main fastball of a pitcher is straight, when it probably isn’t. If you recognize that, though, you shouldn’t get too lost.
Honestly,
I usually think of pitch movement by centering the strike zone around 0,0 and then equalize all the other pitches, reminding myself that fastballs will usually have some horizontal break, and sliders are usually the pitches closest to an actual gravity-only pitch.
So…basically, I’m doing what you suggested, but using the slider instead of the fastball (which I’ve done ever since a BTB (I think?) post about fastball movement vs. slider movement).
One of my friends refers to advanced statistical analysis as “baseball calculus” and usually gets tired-head when I explain this stuff to him (he’s a math major, so it isn’t really that he can’t grasp the concepts, just that he isn’t as familiar with the fundamental pieces). This PitchF/X stuff really gives that idea some legs.
from BtB
Change in location compares a pitch’s actual final location at the front of the plate to where physics equations would have expected it to end up given no spin at all (and no knuckling effects).
http://www.beyondtheboxscore.com/2009/4/17/841366/understanding-pitch-f-x-graphs
Elvis Andrus - 2009 AL Rookie of the Year
I recall the reasoning.
I’ve tried explaining this to co-workers who like PitchF/X, but don’t understand the graphs.
“Well, zero/zero is a pitch with no spin.”
“Like a knuckleball? But those move all over the place.”
“Right, but we’re ignoring wind effects and just concentrating on gravity.”
“Okay…but doesn’t that make it hard to compare faster pitches to slower pitches?”
“Well…yes.”
Where do you go from here? In this scenario, you’re talking about comparing the spin on an 85mph pitch to the spin on a 92mph pitch…but gravity affects the 85mph pitch more, effectively changing the meaning of its 0/0 point.
Do you see my problem here? In these graphs, each pitch’s location changes are comparable only to other pitches with the same velocity. So you’re essentially asking people to compare this 85 mph slider to an 85 mph knuckleball thrown in a vacuum.
Right? Have I missed something here?
understood
I think 0,0 is just a reference point that makes it easy to compare movement on a pitcher’s pitches. The same point is used for all pitchers in the league. It may help a little if we understood how the league (or the makers of pitch f/x) came up with it’s baseline pitch. As for velocity, it has been awhile since my college physic class. Just how much of an effect does gravity have on a 90mph FB vs one thrown at 95?
Elvis Andrus - 2009 AL Rookie of the Year
Not much, but...
When you compare a 95-mph Padilla fastball to a 55-mph Padilla curveball, the gravity difference is pretty large.
Yeah, that's essentially it.
Another reason I’m so hung up on this is that GameDay plots all the pitches completely in a strike zone without any need to compare them to the gravity-only pitch, so I feel like they took a straightforward, easy-to-interpret interface and complicated it by introducing the gravity-only pitch.
To me, the 90 vs. 95 scenario you mentioned isn’t huge, but if I think “I wonder how much his fastball moves in relation to his curveball”, then the number crunching (based on this data, at least) is absurd. We’re talking about three transformations (curveball to gravity-only pitch at curveball velocity, gravity-only pitch at fastball velocity, gravity-only pitch at fastball velocity to fastball) just to extract relative movement differences between two pitches in the same context.
Even if we aren’t talking about huge movement, it’s worth accounting for the changes considering how often we talk about an inch or two of break making a huge performance difference.
The pitch location graph
is confusing. Too many data points spread over too many innings. Maybe it would be more illuminating to see a graph inning by inning, or to find some sort of weighted average each inning and make a simple graph of those averages? Then you might see a bigger trend. But if I have 90-100 data points spread all over a graph it’s useless to me unless the differences are just painfully obvious.
Plus, these charts may not account for the reasons behind different location points and other factors. Was his release point consistent? Do the pitches left up in the zone correspond with higher velocity (maybe overthrowing or rushing) or is it random? How can we be more certain of pitch selection? (this seems like a big problem and constant question with pitch f/x).
Concerning pitches left up in the zone, I’m more inclined to trust the eyes of the pitcher and the catcher than a confusing graph, until and unless I understand the graph better. I love having this data available, I’m just not sure that it’s the right way to answer that particular question.
G G G E-flat_______ F F F D__________....
Pitch location graphs
I think you’re right about this. I haven’t tried this, but I wonder if you could upload these Excel graphs (that’s what they are, right?) to Google Documents and have different sheets for the different innings, or something to that effect (of course, assuming that you could share the graphs themselves). In general, I find that these graphs are information overload.
Perhaps if a breakdown graph by strike zone area went along with this graph? The BTB guys do this now, but I actually find their in-strike-zone/fat/wide charts to be even more confusing.
I think the main problem is that, in order to display easy-to-digest information, you need a wealth of graphs and charts, which could then become daunting because of the amount of graphs themselves.
Release point
Is definitely the next thing to look at, and very simple to do. The pitch location graph can obviously be added onto as well, which is something I can do over the weekend.
Concerning gameday, I’m more putting in those caveats as part of my own uncertainties. I’m not sure that Matt Harrison throws 6 pitches, but I wouldn’t think so. Gameday, as far as I know, just classifies pitches based off algorithms; Harrison could just have pitches that throw the system off.
Lastly, I agree with you that this question probably won’t get answered simply using the data, but it definitely doesn’t hurt. Harrison definitely changed something after that second inning and it made a huge difference. Now, will I be able to find that answer in the huge amount of data that 7 innings of pitching produces? Good question.
by Desert Ranger on May 1, 2009 3:02 PM CDT up reply actions
+ everythinghejustsaid!
"Back on the scene, with a gangsta lean" RW
"you gonna lose your horse. seriously." FX2
Yes we can! November 04, 2008
Over at BTB
http://www.beyondtheboxscore.com/2009/4/20/846320/using-pitch-f-x-to-determine-an
Discussing how to judge an umpire.
They have the ump’s calling 83% of pitches correctly. That’s average. I wonder what the average was for our recent Baltimore series.
Comments For This Post Are Closed
Secondary Sidebar
FanPosts
Community blog posts and discussion.
Recommended FanPosts
Recent FanPosts
In Association With
Big League Stew - MLB - Yahoo! Sports
- John Lindsey couldn't have come any closer to his first at-bat
- The Juice: Swisher ends Orioles sweep dreams with timely homer
- Create-a-Caption: Dusty Baker wants YOU to buy playoff tickets
- Padres' Heath Bell wistful about Trevor Hoffman's 600th save
- Want your first career home run? Face a pitcher from Chicago
Certain photos copyright © 2010 by Associated Press or Getty Images. Any commercial use or distribution without the express written consent of Associated Press and Getty Images is strictly prohibited.
