September 23, 2013

Pebble Hunting

Pedro Hernandez and the Rashomon Project

by Sam Miller, R.J. Anderson, Dan Brooks and Dan Rozenson

In the movie Rashomon, a samurai is murdered. Four witnesses give four accounts of the murder, and out of one scenario come four very different narratives and three different killers. Do more angles get you closer to truth, or further from it? It's not clear.

What follows is an experiment. Four of us took a starter that none of us knew anything about: Pedro Hernandez, a Twins lefty making his 12th career start, on Saturday against the A’s. Without doing any research on Hernandez, the four of us watched the start from four different angles:

• R.J. Anderson watched on TV;
• Dan Brooks dove into Hernandez's PITCHf/x log for the game;
• Dan Rozenson listened to the Twins’ home radio broadcast;
• and I (Sam) watched the game from a seat behind the home dugout at O.Co Coliseum in Oakland.

The idea isn’t to get the most accurate scouting report on Hernandez, though maybe that will happen; I haven’t read the other three yet so I genuinely don’t know. The idea is to see how much our opinions about a pitcher are influenced by the way we observe him.

Now, I didn’t get the starting pitching line that I had hoped for in this experiment. I was hoping for six innings, seven hits, five Ks, two walks—enough ambiguity that we would each be challenged to draw conclusions about the pitcher’s stuff. Instead, Hernandez got knocked out after two innings and six runs. It’ll be hard for any of us to not conclude that he sucks, after that. But, of course, there was no dispute in Rashomon that there was a dead body. We can still find out if we diverge on how he sucks, why he sucks, and other suck-related questions.

A very important thing to remember: each of these evaluations is very likely to be wrong in parts. That’s the point, that each individual angle has blind spots. I’m somewhat terrified to read the other three reports, because I don't even feel confident that I have correctly identified which pitches Hernandez throws. Again: Wrongness is inevitable. Wrongness is what I'm interested in. If you want a rigorous scouting report on Pedro Hernandez, you should go somewhere else. Like a mental hospital?

And here goes.

1. Describe the pitcher’s repertoire and quality of each pitch.
Anderson (television): Hernandez threw three pitches: high-80s fastball, mid-80s changeup, and high-70s curveball. The fastball was his favorite pitch in the early going, but he introduced the change and curve as the game wore on. None of the offerings stood out to me. The A's had seven right-handed batters in their lineup and they were able to barrel the changeup without fail, which left me wondering if he had a large enough speed differential between the pitches. Meanwhile, his fastball didn't have enough oomph to use inside against righties. Neither pitch missed many bats and hitters showed no problems lifting his pitches. The curveball was his best pitch by default.
Rozenson (radio): Fastball 88-90, curveball, and changeup (as described on-air). The curveball was mentioned as a key pitch to help break up the monotony of his fastball. He got a big strikeout on the curveball with the bases loaded, but it was against Josh Reddick, who was hitting .205 against lefties coming into the day. There were no descriptions as to the movement or command of these pitches.
Miller (in-person): Fastball, 89-90 mph, thrown without much downhill plane or movement. A 40 pitch, and that might be generous. Cut fastball, 87 mph, doesn't go inside with it as much as you'd want, used as his primary pitch so seems like he thinks that’s his best pitch. A 40 pitch. Throws one of the fastballs 80 percent of the time.
Changeup, 83-84 mph, arm speed telegraphs it. Not enough separation in velocity from the fastball. A 30 pitch. Curveball, 76-78, not thrown often, but in any count. Comfortable throwing for strikes to right handers, or as put-away pitch to lefties. Can be easy to spot out of his hand. A 50 pitch.
Brooks (PITCH/fx): In a lot of ways, Pedro Hernandez is your worst possible case pitcher to do this kind of exercise, because while some pitchers have fairly distinct “clusters” of pitches, Hernandez is one of the “overlapping slop” guys. It looks like he threw a four-seam fastball, two-seam fastball, cutter, slider, and change. Truthfully, it’s the kind of stuff that makes me think “bleh” more than “wow.” Nothing is clustered well and the release point looks shaky. His four-seamer sat around 91 and topped at 92, and he gave up a homer and a double on it. His sinker was a tick lower at 90; that also got hit. The changeup ranged from 84-87 or so, with fairly decent arm-side run, but nothing spectacular; that also got hit. The slider is uninspiring and at 78 is probably too slow with too little movement for the major-league level (it had one swing and miss for a swinging strikeout); that also got hit (do you sense a pattern here?). The cutter is probably the best pitch, sits 89-90, and was most consistently thrown for strikes (with one swing and miss).

2. How good was his control? How good was his command?
Anderson: Hernandez had subpar control; he was inefficient and worked deep into counts, though he walked just one. He tended to miss low to the arm-side and high to the glove-side. Based on this viewing Hernandez appears to have below-average location skills. I will note his catcher (Josmil Pinto) didn't seem to help him get any additional calls.
Rozenson: Hernandez seemed to really struggle with with his control. He was constantly falling behind hitters and seemed unable to hit his spots regularly. Oakland hitters took advantage of him falling behind and waited for fastballs. He walked Kurt Suzuki on four straight pitches and then started the next batter 2-0.
Miller: Poor control. Threw first-pitch balls to seven of first nine batters; walked bottom-of-the-order bat on four pitches, looked uncomfortable on the mound. Poor command. Put very few pitches on the edge of the zone; batters got good swings against him even in two-strike counts.
Brooks: Pedro left a fastball inside over the plate that was taken deep. He left a fastball inside over the plate that was hit for a double. He gave up several singles on pitches left way over the middle of the plate, and a few more hits on pitches located away but not quite enough away. He looks like he got squeezed a bit on a four-pitch walk on a pitch that could easily have been called a strike, but let’s face it, at least Suzuki didn’t get two bases, so we’ll score that a net positive. It’s kind of hard to judge control and command based on two innings of a performance where a guy gets knocked around.

3. What was the single best pitch he threw all game?
Anderson: His single best pitch was a curveball to Josh Reddick in the first inning.

Rozenson: The single best pitch might have been the curve to get him out of the bases-loaded jam—tough to say without having seen it.
Miller: 2-2 curveball to Josh Reddick.
Brooks: I’m reaching here, but I guess I’ll go with “only good pitch he threw” as the slider to K Reddick in the first. The sequence was:

1. Slider (way up and WTF)
2. Fastball (up)
3. Fastball away but caught a lot of the plate but was fouled off
4. Another fastball located pretty well on the outside edge
5. A swinging strike on a slider down and away.

4. What was the single worst pitch he threw all game?
Anderson: The single worst pitch was a fastball inside to Alberto Callaspo, which was struck for a two-run home run. It felt like trouble was certain whenever he threw a fastball inside to a righty.

Rozenson: Unclear
Miller: 2-0 fastball to Josh Reddick, or 2-2 fastball to Josh Donaldson.

Brooks: There are a lot of pitches here that could probably be “worst”, but I’ll go with the homer in the second by Callaspo. Sequence:

1. Slider down and inside taken for a ball (Why? are you a glutton for punishment Pedro?)
2. Cutter away taken for a called strike
3. Fastball down and away
4. Thigh-high inside fastball that caught a ton of the plate.

5. If you were an advance scout, what would you tell your hitters about him?
Anderson: Load the lineup with righties. He lacks the adequate velocity and command to pitch with his fastball inside against righties, and the A's righties had no issue barreling his changeup, meaning he has no real out pitch. Let him work himself into trouble by nibbling. He's going to make mistakes because his margin for error is too small. I would also note that in this start the catcher was implicitly accusing the A's of stealing signs in the second inning, as he angled his body toward third base when giving signs with a runner on first; after Hernandez started his delivery the catcher then added more instruction.
Rozenson: Don't do him favors by swinging at the first pitch. He fell behind regularly and wound up having to fire fastballs over the plate to try to fall back even. Make him throw strike one. He seemed to be working away more than inside, and I would doubt his ability to consistently work inside given the lack of command or velocity on his heater. Can't offer any advice on the off-speed stuff.
Miller: Dig in; he won't make you move your feet. Be patient; he's too fine with his fastball and cutter and will pitch himself into trouble within at-bats. Even if he does get ahead, he doesn't have a put-away pitch, and with two strikes you should still be able to put a big swing on something. Throws a changeup to right-handers, and a curve to either, but will very rarely throw either early in a count. Will throw secondaries behind in the count or when he's struggling to find command.
Brooks: Don’t waste your time. Study the bullpen. You’re going to see it by the third inning. He has the kind of stuff that can probably attack LHH much better than RHH. He didn’t throw a lot of two-strike pitches so it’s difficult to tell what he’ll come after you with if he gets into favorable counts.

6. If you were a pro scout, what would you tell your organization about his future?
Anderson: His future is in the bullpen or Triple-A. I don't see starter material. That's not because he's a short lefty, either; he's kinda stocky and does have a three-pitch mix. It's just the stuff isn't good enough for me to envision him being effective against righties from the bullpen, let alone after seeing them multiple times in one game. Not to be too harsh, because it's one start (and only two innings at that) but I'd project him as a second lefty or otherwise not a big leaguer on a good team.
Rozenson: He needs some work to stay at the big league level. With improved control, and assuming his curve can get some whiffs, there might be a spot for him as a lefty specialist for a non-contending team. A starting rotation gig seems highly unlikely.
Miller: Should be in the bullpen, where his stuff might have some chance of playing up. Looked exhausted by his 50th pitch; mechanics started to show fatigue. Four-pitch mix but none of the four stands out. Hides the ball acceptably against lefties, and with a couple extra miles per hour his cutter/CB combination might play in a LOOGY role.
Brooks: Call the A's. Maybe Billy Beane sees something in him. Get him to trade something for him. Anything at all, really. Maybe if he gets to face a lineup with more than one LHH, he’ll be better? We’re reaching here.

7. Give me a comp.
Anderson: Jeez. I don't know. A smallish lefty with a mediocre three-pitch mix who lacks gimmicks and belongs in the bullpen.
Rozenson: Bruce Chen a best-case scenario. Randy Keisler might be more realistic, albeit with a less lively fastball than Keisler's.
Miller: Current version of Erik Bedard.
Brooks: I don’t know. Early career Aaron Laffey with a cutter but worse stuff? Does that even count as a cutter? Who knows.

***

The advantages and disadvantages of each seat seem clear from this exercise. PITCHf/x gives precision, and it freezes each pitch in place for careful study, but it makes us imagine a scene instead of seeing it. If you've ever been surprised by the face of a familiar radio voice, you know how far off the imagination can be. Radio provides accounts and descriptions of the game by experts who are doing their very best to help us see and understand what is happening. But it's a very narrow tube, and only a sliver of the relevant information can squeeze through it. Television provides visuals, but usually at a deceptive off-center angle, and certainly not at the angle from which the batter sees the pitch. It probably gives a false sense of confidence in where the pitch is and how it is moving.

From my perspective at the stadium, the blind spots were pretty apparent. It’s almost impossible to gauge movement from anywhere but the scouts’ section, and I was having to use all sorts of context clues to guess even what his repertoire was, let alone how good each pitch was. Inside/outside location, too, are hard to see. Don’t even try to call balls and strikes. The single best clue on pitches—velocity readings—disappear from the scoreboard after about three seconds, which means that by the time Jed Lowrie coasts into first base on his single it’s too late to look up and see which pitch the guy actually threw.

I ended up trying to use all sorts of clues for help. I figured a pitcher with good command would get strike calls on the edges of the zone, so I kept alert for batters looking back at the umpire to express disagreement with calls. For location I tried to watch the catcher’s glove. For control I focused on first-pitch strike rates and strikes thrown behind in the count. For pitchability I wanted to see how balanced hitters’ feet were when they were swinging from behind in the count. The problem is that all these things (except strike rate) happen in a flash, simultaneously. The game just flows right past, mostly unobserved.

The most instructive data, from my seat, were the hitters’ swings. From the dugout angle you can really see a hitter’s swing break down when he’s fooled by a pitch (on a swing or a take), and you can really see it explode when he’s on a pitch. The two advantages I felt like I had over TV were seeing the hitter from this angle on every pitch, and being able to see the pitch out of Hernandez's hand.

Roger Ebert began his review of Rashomon with an anecdote, and we'll end our review of Pedro Hernandez with the same anecdote:

Shortly before filming was to begin on "Rashomon," Akira Kurosawa's three assistant directors came to see him. They were unhappy. They didn't understand the story. "If you read it diligently," he told them, "you should be able to understand it, because it was written with the intention of being comprehensible." They would not leave: "We believe we have read it carefully, and we still don't understand it at all."

Recalling this day in Something Like an Autobiography, Kurosawa explains the movie to them. The explanation is reprinted in the booklet that comes with the new Criterion DVD of "Rashomon." Two of the assistants are satisfied with his explanation, but the third leaves looking puzzled. What he doesn't understand is that while there is an explanation of the film's four eyewitness accounts of a murder, there is not a solution.

Answer Key: Brooks Baseball's Pedro Hernandez page, and his BP player page.

Sam Miller is an author of Baseball Prospectus. Follow @SamMillerBB
Click here to see Sam's other articles. You can contact Sam by clicking here
R.J. Anderson is an author of Baseball Prospectus. Follow @r_j_anderson
Click here to see R.J.'s other articles. You can contact R.J. by clicking here
Dan Brooks is an author of Baseball Prospectus. Follow @brooksbaseball
Click here to see Dan's other articles. You can contact Dan by clicking here