A Pitch f/x look at Bugs Bunny

Review: Mohegan Sun Sports Bar Seats at Yankee Stadium
Update on Chien-Ming Wang

Thanks to a trio of abysmal Chien-Ming Wang starts, the Yankees’ bullpen has thrown 46.2 innings this year, fourth most in the AL. Edwar Ramirez has bit the bullet twice in relief of Wang, throwing a then-career high 51 pitches on the 13th before establishing another career high with 58 pitches on the 18th. All of that extra work gave us a meaningful enough sample of pitches that we can use to take a closer look at Edwar’s one trick pony act.

Let’s get it started with the usual, the pitch trajectories. I’m only going to look at Edwar’s fastball and changeup, but he does through the occasional slider and it’s clearly his third offering. In fact, he’s thrown just eight this year according to Pitch f/x’s classifications, and that’s out of 162 total pitches. As with all of our Pitch f/x graphs, you can click these for a larger view. Let’s start with the bird’s eye view:

Bird's Eye View

This view is pretty useless for fastballs and changeups, but you can get an idea of how quickly each pitch gets on a hitter. Each tick mark along the flight path represents 1/100th of a second, so the farther apart the ticks, the faster the pitch. Edwar’s changeup takes about 4/100ths of a second longer to get to the plate than his fastball, and you can see that difference in velocity by looking at the tick marks. Here’s the view from first base:

First Base View

This view is always fun because he can get a real good look at how the pitches move vertically. Obviously Edwar’s changeup has a considerable amount of drop, and that sink really starts to kick in when the pitch is more or less halfway to the plate. Let’s check out the view from behind the dish:

Catcher's View

Edwar’s changeup has fades down and away from lefthanders considerably, which leads you to believe that he’d be more effective against batters of the opposite hand. The stats don’t jive with that though:

Career vs RHP: .219-.318-.438 against, 33.3 K%
Career vs LHP: .262-.368-.459 against, 22.4K%

They’re both pretty significant samples (171 plate appearances vs RHP, 201 vs LHP), so there’s no funny business going on. I dunno what that’s about, maybe Edwar is just more comfortable throwing to his arm side, which allows him to more effectively bust righties inside. Just a guess on my part.

Lastly, here’s his release points:

Release Points

That’s a pretty tight set of release points, generally about 8-inches tall and 10-inches wide, save for a few extraneous pitches. Edwar has one of best righthanded changeups in the game, and there are two main factors contributing to it’s effectiveness:

  1. Movement: You can see it in the graphs above, Edwar’s change moves both horizontally and vertically. It’s hard enough to hit a pitch that moves in just one direction, nevermind two.
  2. Deception: This is the big one. Edwar throws his change from the same arm slot with the same arm speed as his fastball, but there’s no way for Pitch f/x to measure this stuff (or maybe there is, I dunno). By throwing his two main weapons the exact same way, hitters have no idea what’s coming, and have to figure out whether the pitch is a high-80’s fastball with little movement or a high-70’s change with redic fade while the pitch in mid-flight.

We’re all familiar with the concept of Good Edwar and Bad Edwar; Good Edwar is the one that’ll march out of the pen and strikeout the side, Bad Edwar is the guy that’ll walk the farm before serving up a gopher ball or two. It seems like we either get one or the other with very few in-between outings. Later on in the season when we have more data I intend to breakdown Good Edwar and Bad Edwar side-by-side to see if there’s a discernible difference.

Review: Mohegan Sun Sports Bar Seats at Yankee Stadium
Update on Chien-Ming Wang
  • Will (the other one)

    Any chance we could also get a “Glasses Edwar” vs. “All-Natural Edwar” breakdown later in the season too?

  • Oscar

    Given that there is a Good Edwar and a Bad Edwar, would you say he’s like the “little girl with the curl”?

    (Sorry, couldn’t resist…).

    • http://www.blogtalkradio.com/pics/hostpics/5e4be77c-8e57-4d7a-8533-add8ed030ad5JerkStore.jpg Slugger27

      my god that saying got old fast

  • Chazzy

    His lower armslot for his fastball has always been visible to the naked eye when watching games, but the catchers viewpoint solidifies this. If I were a major leaguer I might be able to pick up on this as a sort of ‘tell’ that the fastball is coming. It is exceptionally hard to sit fastball and adjust to the change given the speed difference and maybe he may get in trouble with this approach, however the 2 plane movement will allow him to get away with such strategies employed against him, most of the time.

    • BJ

      It does seem like the change is coming from higher up and further out than the FB. I guess pinpoint release location is tough to identify and the identical armspeed is enough fool batters.

    • http://twitter.com/tsjc68 tommiesmithjohncarlos a/k/a Ridiculous Upside

      His lower armslot for his fastball has always been visible to the naked eye when watching games, but the catchers viewpoint solidifies this.

      Huh? Am I looking at the same graphs as you?

      The bird’s eye view shows that he releases the changeup slightly further away from his body than the fastball, and the view from first base says the changeup is released ever so slightly higher than the fastball, but the view from the catcher’s perspective has the two pitches coming from the exact same spot.

      These graphs seem to indicate there is little to no “lower armslot… visible to the naked eye”, particularly from the catcher’s (and batter’s) viewpoint.

      • BJ

        You can actually see where the blue starts is not where the red starts, it is less clear because the red is on top of the blue, if it was the other way arround you would see it more clearly. Your eye assumes the blue line keeps going with the red but it actually starts from a lower and more inside spot

        • http://twitter.com/tsjc68 tommiesmithjohncarlos a/k/a Ridiculous Upside

          Yes, but those spots, while technically different, are negligibly different from the viewpoint of the batter.

    • Ed

      His lower armslot for his fastball has always been visible to the naked eye when watching games, but the catchers viewpoint solidifies this.

      The catcher’s view? On that graph the two release points are identical. The other views are the ones that show a difference.

      The only graph with markers on the axises is the scatter plot, so you can’t judge anything by the others. If you look for the center of the clump of red dots and the center of the clump of blue dots, you’ll see that his average release point looks to be about 1/4″ higher and 1/2″ to the left for the fastball. From 60′ away, there’s no way you can notice that.

      • BJ

        I think the scale on the side is in feet, not inches.

        • Ed

          Yeah, you’re right.

          Still, 3″ vertically isn’t going to be noticable. 6″ horizontally may be noticable, but the clustering is less defined in that direction. You might be able to guess fastball on the extreme cases, but in general there isn’t much difference.

  • http://twitter.com/tsjc68 tommiesmithjohncarlos a/k/a Ridiculous Upside

    “Edwar’s changeup has fades down and away from lefthanders considerably, which leads you to believe that he’d be more effective against batters of the opposite hand. The stats don’t jive with that though.”

    I believe you mean “The stats don’t jibe with that, though.”

  • http://twitter.com/tsjc68 tommiesmithjohncarlos a/k/a Ridiculous Upside

    Edwar throws his change from the same arm slot with the same arm speed as his fastball, but there’s no way for Pitch f/x to measure this stuff (or maybe there is, I dunno).

    Would initial pitch velocity work? As in, if he throws both the change and the fastball with the same arm speed wouldn’t it indicate similar initial pitch velocities, and the change just arrives slower because it has a lower end velocity?

    Or am I wrong with that and his change somehow has a noticeably slower initial velocity despite similar arm speed?

    Because I’m guessing you could take the release point graphs and add in the pitch velocity to them via an intensity factor (greater velocity, bigger dot) to compare both horizontal and vertical release point and pitch velocity at the same time on the same graph.

    • http://statspeak.net dan

      No, the changeup starts slower than the fastball, too because of the grip. The grip is what makes it slower than the fastball despite the same arm speed, not the spin or whatever. Both will lose around 10% of velocity on the way to the plate.

      • http://twitter.com/tsjc68 tommiesmithjohncarlos a/k/a Ridiculous Upside

        Yeah, figured it was too good to be true. But changeups probably do lose slightly more of their velocity mid-flight, since they have less backspin and generate more drag coefficient, no?

        • http://statspeak.net dan

          His 20 slowest changeups on April 18th started at 75.9 mph, it crossed the plate at an average of 69.8 mph.

          His 20 fastest fastballs on the same day started at 87.4 mph (ouch), and crossed the plate at 79.9 mph.

          CH velo loss: 6.1 mph (8.0% drop)
          FB velow loss:7.5 mph (8.5% drop)

          • http://twitter.com/tsjc68 tommiesmithjohncarlos a/k/a Ridiculous Upside

            Huh. Freaky.

            I figured the change, with more biplanar break, would have had to have slown down more on the way to the plate. Perhaps that’s why it’s as effective as it is, it moves but doesn’t slow much (and, conversely, why is FB is so pedestrian, it doesn’t move but does slow quite a bit).

    • CB

      In general, the velocities recorded by radar guns are the initial velocities or as close to it as possible.

      The standard way that pitch velocities are measured it to have the gun record speed as close to the point of release as possible. So what you’re seeing is generally the velocity of the pitch as it comes out of the hand. This measurement is generally made within the first 10 feet of the release.

      After that the ball usually slows down around 8 mph or so due to drag if I remember correctly.

      Where that initial measurement is made however is operator dependent so this is one of the reasons for error in radar gun reading.

      For example one of the reasons why there are “slow guns” is that the operator is simply holding it or is positioned at a place where they are recording the velocity too far away from the release point. Same thing is conversely true for “fast” radar guns.

      This is an issue for pitch f/x data. As I understand it the system to measure velocity/break is standardized and fixed in each park.

      However, different pitchers have different release points and his can cause errors in the readings. It’s probably not a big deal but it’s a source of error.

      • http://twitter.com/tsjc68 tommiesmithjohncarlos a/k/a Ridiculous Upside

        Ah, okay. So, to get the start and end velocities that Pitch f/x lists, does it just measure the average speed of the pitch, determine what type of pitch it is (to calculate the percent loss in velocity), and then reverse-solve the equation for probable start and end velocities?

        Is that how they do it?

        • http://statspeak.net dan

          No… if you download the actual data, there are separate lists for start and end velocity.

          • http://twitter.com/tsjc68 tommiesmithjohncarlos a/k/a Ridiculous Upside

            Yeah, but I’m asking how those are actually measured.

            Does Pitch f/x have a radar gun measure the actual velocity of the ball twice, at the beginning and end of the pitch, or is one measurement used and then calculated into two?

            I had always assumed they gunned each pitch twice, but the post by CB seems to say the pitch is just measured once and then physical properties are used to judge the rest.

            Just asking how it gets the numbers it gets.

            • http://www.riveraveblues.com Mike Axisa

              Pitch f/x is a series of two cameras. One looking at the pitcher and one looking at the plate. Each captures the location and spin of the ball as it crosses it’s plan. I assume it records the velo as well.

              • MattG

                So is there any truth to the theory that Mariano Rivera’s fastball maintains its velocity much better than average? Or is that a myth? I had read somewhere that Rivera’s ball typically loses only 4 MPH in transit.

              • huuz

                i seriously doubt that the cameras “measure” spin.

                location is all you get.

                • http://statspeak.net dan

                  Download the data. There are columns for spin rate, spin angle, and I think a couple others involving spin. I don’t know if it’s measured directly or derived from the angle and force of the break.

            • http://statspeak.net dan

              Oh, I see. There’s no actual radar gun making these readings. It’s two cameras that take 25 “pictures” of the ball on its way to the plate. From these 25 points in time, it connects the dots and creates all these flight paths. So the “end velocity” is just the distance between the last two dots divided by the time elapsed between pictures.

              So, theoretically, they could provide the velocity of the ball from 26 feet to 27 feet if they wanted to, but nobody cares about that.

              Actually now that I think about it, there’s an incrementally more precise way to measure velocity, but I’m not which of these two they use. Once all the dots are connected (from the 25 “pictures”), you could just find the distance function and take the derivative of that. No clue how to get the distance function though, we never dealt with air friction and such in physics. Mike might know.

              • http://statspeak.net dan

                Derivative of that function at a certain point**

                • Accent Shallow

                  I’d rather not math up the thread if it can be avoided.

            • CB

              I was primarily commenting on how velocity is generally measured.

              With a radar gun it’s usually only measured once – which is at the “point of release.” Generally with a hand operator that can be done within 10 feet of release.

              Pitch f/x doesn’t use a radar gun at all. It literally takes a series of photos from 2 cameras as Mike described. They calculate velocity using the photos as a kind of timer over the fixed distance the ball travels.

              But in general with pitch f/x I believe the velocity that they report is the velocity that corresponds to what the radar gun should be reading.

              That is the pitch f/x velocity you get reported on game day is the velocity “out of the hand” which again is recorded within the first 10 feet.

              • http://twitter.com/tsjc68 tommiesmithjohncarlos a/k/a Ridiculous Upside


                Thanks, all.

      • http://statspeak.net dan

        Pitch measurements start at 50 feet from home plate in all parks since 2008. In 2007 that point jumped around a lot, from 55-45 feet IIRC, but it’s at 50 now.

  • http://statspeak.net dan

    Edwar throws his changeup with an arm slot that is both higher up and farther out than his fastball, which is extremely weird when you think about it. It’s a very small difference, but it’s there.

    Career vs RHP: .219-.318-.438 against, 33.3 K%
    Career vs LHP: .262-.368-.459 against, 22.4K%

    Two things: One reason this may be is that righties so rarely see changeups from same-side throwers that they’re just really thrown off by it. And a better way to look at this would be to do career splits vs. platoon average. Edwar does better vs righties, but so do all right-handed throwers. How much better/worse is he than expected?

    I guess if you don’t feel like it, I could check it myself later using odds ratios.

    • http://statspeak.net dan

      Forget about it, I’m too lazy.

  • Ed

    Is there any way to compare his pitch selection when facing righties vs lefties?

    Maybe he throws the pitches in different ratios, making it easier for lefties to guess which pitch is coming?

  • Rich

    The Good v. Bad Edwar is why he probably needs another pitch that he can throw more often. It would give hitters another look when his command is lacking.

    • http://twitter.com/tsjc68 tommiesmithjohncarlos a/k/a Ridiculous Upside

      Do I smell a Mariano Rivera teaching him the cutter in our future?

      • A.D.

        If he just went from his normal straight fastball to a cutter, that could be huge.

        • http://statspeak.net dan


          • http://twitter.com/tsjc68 tommiesmithjohncarlos a/k/a Ridiculous Upside

            I mean, seriously, imagine if he had a two-pitch arsenal of a Bugs Bunny change and a boring cut-fastball. That would be sick.

          • Pel

            © Paris Hilton MMIX

      • LiveFromNewYork

        My understanding is that Mo does tutor him.

      • Accent Shallow

        Someone once asked Steve Carlton how to throw his slider. Carlton picked up a ball and said “You hold it like this, and then you throw the shit out of it.”

        I’m sure Mariano’s more intellectual than Carlton, but the great ones can have difficulty teaching mere mortals.

    • Chris

      Except Good Edwar vs Bad Edwar is almost always a matter of his control on a given day (just like Veras). If he throws strikes, he’s generally very effective. Like most change-up pitchers, he’ll give up his share of home runs, but that’s generally not an issue when they’re sandwiched in between three K’s. It’s only an issue if he’s walked people.

  • dkidd

    i wonder if mo ever tried to teach farnsworth a cutter. he seems like the poster child for needing a different fastball look

    • jsbrendog

      0-3 in kc. haha

  • Pel

    Cool stuff, Mike.

    If I could do this myself I would, but do you think you could format the x/y-axis of the “Catcher’s View” to be more symmetrical?

    Like -5, 5 x-axis and a -2, 8 y-axis?

    Maybe drop the transparency of each line and marker to 50%?

    I’d just like to see it from a centered view.

  • Raphy

    I must say that I was terribly disappointed by this article. I thought we were going to get actual analysis on those crazy pitches that the actual Bugs Bunny used to through in the cartoons.