Monday, December 27, 2010

Joe Bautista - won't you take me by the hand? **

I still can't believe that Jose Bautista hit 54 home runs in 2010.  Could it be that he was doing something differently?  His OPS sure says so:

2006 - .755
2007 - .753
2008 - .718
2009 - .757
2010 - .995

Let's check out his first 20 plate appearances in 2009 and 2010:

2010                                                        2009

1) 3 - pop out                                           1) 2 - line out
2) 6 - line out                                            2) 3 - 1b
3) 4 - fly out                                              3) 9 - 1b (infield)
4) 4 - walk                                                4) 7 - swinging
5) 4 - called out                                         5) 3 - fouled out
6) 5 - walk                                                6) 8 - swinging
7) 4 - swinging strike                                 7) 4 - fly out
8) 4 - walk                                                8) 2 - sacrifice
9) 2 - pop out                                           9) 4 - walk
10) 7 - pop out                                         10) 4 - swinging
11) 4 - GIDP                                            11) 1 - 1b
12) 2 - flies out                                          12) 3 - foul out
13) 4 - 2b                                                 13) 4 - 1b
14) 4 - swinging strike                               14) 1 - infield choice
15) 6 - walk                                              15) 4 - swinging
16) 3 - 1b                                                 16) 1 - ground out
17) 1 - pop out                                          17) 3 - looking
18) 1 - 2b                                                  18) 1 - 2b
19) 3 - line out                                            19) 3 - 1b
20) 4 - sac fly                                             20) 6 - 2b 

75 pitches                                                  73 pitches

Huh.  Not much of a difference in total pitches.  Maybe he didn't do anything different.  But not so fast -- he hit 73 in 2009 thanks to 3 long at bats (#3, 4 and 6) offsetting 4 times when we drove the first pitch into play (#11, 14, 16 and 18).  By contrast, in 2010 he was more consistent, with only 2 1-pitch at bats and only 1 at bat of 7 pitches or longer.

What do I make of this?  Well, first and foremost, I make that 20 at bats isn't a credible sample.  But given that this is what I have in front of me, I'll make the following posit.

In 2009, Bautista shows a ton of ability to make contact.  He goes 3-4 when he chases after the first pitch and #18 went deep.  Further, he fought hard in those early at bats, earning 27 pitches over 4 plate appearances.  In 2009 he had all of the tools to be a dynamic hitter except one -- patience.

In 2010, he saw pitch #4 in 13 out of his first 20 appearances (compared to just 9 in 2009).  He slowed it down and waited for his pitch.  And did it pay off...

Bautista also benefitted from not having a reputation as a power hitter.  I'm guessing that 2010 isn't easily repeatable, and I wouldn't want to put down money on another 50 home run year.  But I'd be surprised if he doesn't manage one home per 18 plate appearances -- maybe even one per 15 (that's 35-45 home runs if he stays healthy).

I guess you can call me a believer.


** -- Yeah, his name is Jose.  But it didn't scan, you see...


I'd like to tell you how I evaluate the performance of the pitcher and the hitter on the first pitch of an at bat, but first I'd better set the table.

Absolute Perfection

What would a perfect game look like? I'm not talking about Don Larsen, I'm talking about Absolute Perfection. I think that you'd agree it would look like this: 81 swinging strikes. It can't get any better than that.

Think about those swinging strikes: Doc Gooden's split finger fastball, falling off the table. Keith Foulke's change-up, with the batter out in front. Tim Wakefield's 50 mph work of art or Mitch Williams' pitch, knocking both batter and pitcher off their feet. Yeah, that's what I'm talking about.

Caught looking is nice, and fouls get the job done. But the essence of a swinging strike is pitching perfection. He thought he had you, but he never even touched you...

According to the baseball almanac, there have been some pretty perfect innings. 44 times, a pitcher has retired the side on 9 strikes, but only 44 times. And I'm guessing that this was never done on 9 swinging strikes, but I'll get into that in a minute.

The Batter Battle Book

So what does perfection look like from the opposite side? Harder to say. Is it a home run? After 10 pitches? 20? It's a little more open ended. To a certain extent, any non-out is a victory, but not all victories are created equal. Heck, I've watched Youk go down on 12 pitches and smiled watching him do so. What constitutes a victory is ambiguous, so it shouldn't be surprising that there are quite a few ambiguous first-pitch results which feel like wins -- or at least light losses -- for the batter. That will complicate the picture.

Clash of the Titans

Imagine Greg Maddux, one of the craftiest pitchers of all time, going against Rickey Henderson, the best lead-off hitter ever. Before a pitch is thrown, what are they thinking?

Well, Maddux knows what Henderson hits well and how the location of the pitch affects him. Maddux also knows his own limitations. His catcher can probably add some feedback on which pitches are working well today. Based on this information, they select a first pitch. But this doesn't tell us too much about what that first pitch will be. Maybe it will be based on what Henderson can't hit. Maybe it will be based on Maddux throws best. Maybe it will be neither, because that's what we least suspect.

How about Henderson? He knows almost everything that Maddux knows. He knows what he can hit. He knows what Maddux can throw. He probably even figured out what is working well by watching Maddux warm up. What doesn't he know? What Maddux decided to throw.

So what's Henderson's best strategy? Well, I would posit that Henderson has enough confidence to risk going down 0-1. And if he preps to swing at anything, he's going to risk making poor contact, followed by a weak out. So I think that his best strategy is guess a pitch and location. If he sees that pitch, he cranks it. If not, he let's it pass, even if that means taking the called strike.

This jibes with my experience, which is that a disproportionate amount of first pitches either called strikes or balls. You see other outcomes, but these two account for many of the first pitches. Just for grins, I looked at the last game of last season, truly one for the ages. The results weren't quite as clear cut as expected, but only because the 7th, 8th and 9th innings went nuts.

In the first 6 innings, Lincecum and Lee were beautiful, allowing 5 hits and no runs on a combined 12 innings. In those innings, here is the distribution of first pitches:
  • 14 balls
  • 13 strikes (looking)
  • 14 others (including fouls, swinging strikes and hit into play)
Clearly not a uniform distribution. I think that the World Series would conform more to my expectations, because the average quality of both hitters and pitchers will be higher. It would be informative to look rigorously at all of the rest of the games and see which of the other categories increase.

Just for the heck of it, here are the next 3 innings results, including 5 hits and 4 runs in just 6 combined innings.
  • 5 balls
  • 4 strikes (looking)
  • 15 everything else.
Yup. It was that ugly.

Measuring the outcome

So, enough table setting. Let's get on with it. Here's how I judge the outcome of various kinds of first pitch results:

Swinging strike. No matter how you look at it, the batter got owned. If we're talking about a run-of-the-mill Jacoby Ellsbury quality hitter, you can see why I credit the pitcher with a major win. But how much more-so for a Kevin Youkillis quality hitter? He thought it was his pitch and he still couldn't make contact...

Called strike. You have to give the nod to the pitcher -- he got what he needed to get done. But it's a weak victory. The batter may have let it go as a calculated risk -- better to risk a strikeout tomorrow than a ground-out today. And location matters. If it's on the perimeter, I credit the pitcher extra with achieving great location. Otherwise, it just wasn't where the batter wanted it.

Ball. Similarly, the ball has to be a victory for the batter, but it need not be a major one. If it was close to the strike zone, I give the pitcher an A for effort, but the batter still got the job done by not swinging.

In play. In play is tough, albeit not as tough as a foul. If the batter makes a weak connection, and the ball dribbles into play, well that's just one step above a swinging strike. If the batter went chasing a ball on the perimeter, again, you have to be impressed with the pitcher. Of course, if ball rings of the bat, you have to think that the batter won, plain and simple. Even on line-out to an infielder. This could be a slap hitter, able to dig out anything the pitcher throws (so, kudos). This could also be the situation I described above -- Henderson has out-thought Maddux and predicted the pitch. He may fly out to the warning track, but you can't ask him to do better.

So to sum up -- weakly hit -- victory to the pitcher. Well hit, victory to the batter.

Foul. Okay. Here we go. There are at least 3 things that I have to know before I can assign victory points:
  • Quality of connection
  • Pitch location
  • Where the foul ball went
Quality of connection is the same point I made above -- if the batter just barely gets wood on the bat, that's one step above a swinging strike.

Pitch location is also important. If the ball was on the perimeter, or outside the strike zone, the batter was probably looking for it somewhere else. He got owned, even if he made good connection.

Finally, via Keith Hernandez again, there's a big difference between a foul straight back and one off to the sides. According to Hernandez, if the ball goes straight back, the only mistake the batter made was one of timing. He got the location right, but not the speed. The pitcher did alright, but it wasn't a big victory.

If the batter shoots the ball into the (non-fair) stands, well, there wasn't much else the batter could do with the ball. Think about those inside pitches which are ruled self-defense. The batter could have let it go for a ball, but swung and got a strike. Call that a victory for the pitcher. Maybe two steps above a swinging strike.

One more time.

So to sum up, here it goes, from best pitcher to best batter outcome:
  1. Swinging strike
  2. Weakly hit (in play, or foul ball -- regardless of outcome)
  3. Well hit foul (not straight back, perimeter, or outside of the strike zone)
  4. Called strike (perimeter)
  5. Called strike (center of strike zone)
  6. Well hit foul (not straight back, in the strike zone)
  7. Ball (close)
  8. Ball (not so close)
  9. Well hit foul (straight back)
  10. Well hit, in play (regardless of outcome)
That's my story, whether I stick to it or not.


Give 'em Ell 'sbury!

In case you're curious, I count pitches on both sides of the plate.

The signing of Crawford made Ellsbury one of the tradable guys in the majors:

  • He could start on most teams
  • He's cheap
  • The Sox don't really need him

  • Sure, both Cameron and Drew are injury prone. But Kalish is close and both Nava and Reddick can fill in nicely. And don't you think that the Sox could have gotten Grienke for Ellsbury, Doubront + someone?

    But I came to praise Ells, not to, um, bury him.

    Most Boston press has forecast the top of the order to be Ellsbury, Pedoroia, Crawford. That seems wrong to me. According to Keith Hernandez' excellent book, the lead-off position will come to the plate 50 more times than the number 2 position, in a given season. The leadoff will come to the plate 100 more times than the number 3 position.

    So your lead-off hitter really ought to earn those extra appearances. Does Ellsbury deserve 100 more than Crawford? Not a chance. You'd essentially be rewarding Ellsbury for being just as fast, but having less power. Frankly, I'd rather see Crawford lead-off and Ellsbury drop down to 9th.

    Right, right. Not to bury him.

    So if Ellsbury hasn't earned the lead-off spot in the past, why should I praise him? Here's why. From 2008 into 2009, he became more patient at the plate. Far more often, he ran the count full. Far more often, he took the first pitch for a strike. It didn't show up in his OBP, which remained 55 points above his batting average. But it showed up in his pitch count. If he becomes more choosy he will see the benefits for years to come.

    I'm looking forward to seeing Ellsbury's next full season. I don't think that we've seen his best yet.

    Incidentally, if the differential between batting average and on base is any judge of utility as a lead-off hitter, Crawford ain't so hot. He only gets 30 points. So who should lead off? How about anybody:

    Ortiz: 100-120
    Drew: 100-120
    Gonzales: 80-130
    Youk: 80-120
    Pedroia: ~80
    Cameron: 70-90
    Scutaro: 60-90
    Varitek: 60-80

    Truth be told, you probably care more about OBP than differential, but I think that most of these guys still look better than Ellsbury...


    Flowers for Papelbon

    The only thing more meteoric than Papelbon's rise from 2005 into 2006 was his decline from 2008 through 2010. It's almost as if the quality of his pitching deteriorated at a rate, directly proportional to the rate of increase.

    Little relievers ain't like little starters. You don't get the same feedback by counting the number of pitches thrown by a reliever. As my man Witherspoon pointed out, these guys might throw 50 pitches on any given night, they just tend not to. But I do count relievers' pitches, and this gives me an opportunity to talk about one of my favorite subject -- optics.

    If you owned an insurance company (and I don't) you'd care about growing profitably. But profits are a fickle mistress. No hurricanes? Extra profits. One bad car accident can erode several months' worth of profit. So rewarding profits leads to rewarding luck over skill.

    The same thing can happen in baseball. What you care about is wins (or, in this case, saves). But saves are equally luck driven. Even ERA, for relievers, isn't that useful. Even for an entire season.

    Optics are a way that Insurers, and Baseball types, can get a quicker read on quality. They tend to be metrics that

  • gather more quickly
  • are less prone to luck
  • are correlated with what you really care about

  • And therein lies pitch count. If you only expect 2-3 runs per nine innings, you'd have to wait for 30 to 40 appearances before you had some sense of whether a pitcher was living up to expectations. But after just a handful of appearances, you have a sense of how many pitches your reliever has to throw. So "gather more quickly?" Check.

    A single hung pitch, driven into the stands, adds 0.10 to 0.15 to an average relievers' seasonal ERA. If there are runners on base, even more. But you aren't likely to have a quick inning on luck alone. Most hitters are somewhat patient and single digit innings are really rare. So "less prone to luck?" Check.

    Alas, on "correlated with..." I only have intuition. I don't think that there are many pitchers out there who throw 25 pitches an inning, but dominate. The best pitchers dominate quickly.

    A single game from last year (sorry, I don't remember the details) will help explain.

    The Sox were facing a good team. Maybe the Rays. And Papelbon came in to save the game in the ninth. He retired the side, with two K's. If you just read the line the next day, you'd have said "maybe he's turning it around." If you watched the game, you weren't fooled.

    The first batter struck out on 4 pitches. No complaints there. But, things went down hill from there.

    The second batter ran the count full before putting the ball in play. I can't remember what the out was, but it was prosaic. A lazy pop-up to center, or a standard-issue 6-3. It wasn't the play that bothered me. It was the 6 pitches. Papelbon had thrown 3 balls trying to nibble the edges and the batter wasn't fooled. But it gets worse.

    The last batter took 9, count 'em 9 pitches to go down. Again, he ran the count to full. And then Papelbon needed 4 more pitches to finish him off. Now, full count scenarios are special and deserve their own post, so I'm not going to go into details here. But conventional wisdom (with which I agree) says that the longer the at bat, the higher the probability of success. Yeah, Papelbon won the at bat (on a foul tip) but he kind of got lucky. One mis-fire and the batter was walked.

    So what does a good season look like? In 2010, Rivera had a pretty fine season, albeit, not one of his best. Looking at the 53 appearances which were exactly one inning, he kept it

  • under 16 pitches 70% of the time.
  • Under 18 pitches 80% of the time.

  • Those numbers are pretty close to Mo's 2008 (65% and 78%). Consistency -- less driven by luck.

    Okay, fine. Papelbon doesn't need to be compared to the greatest closer ever. How about Jenks in 2010?

  • under 16 pitches 52% of the time.
  • Under 18 pitches 69% of the time.

  • Our boy Papelbon?

  • under 16 pitches 46% of the time.
  • Under 18 pitches 52% of the time.

  • (Down from 56% and 60% in 2007...) If you want to get a sense of whether Papelbon has bounced back in 2011, it will only take a little pitch counting to know. They say it ain't over until the fat closer sings. It may be time to break out the flowers.


    Waiting for Francona

    People used to believe that Pedro Martinez was on a 100-pitch count leash. In 2003, many thought the failure to enforce that leash led to the calamity that followed (and I shan't say more about that today).

    I'd like to see Beckett on a similar leash. Looking at his batting average against since 2002 shows that something was painfully different last year:

    2002 -- .232
    2003 -- .246
    2004 -- .235
    2005 -- .234
    2006 -- .245
    2007 -- .245
    2008 -- .256
    2009 -- .244
    2010 -- .292


    I watched a lot of those games. Some of them never looked good. But for several, they shared the following qualities:

  • Fewer than 60 pitches through 4 innings and very few base runners.

  • One or two massacres among the 5th, 6th and 7th innings.

  • What happened? Was it mental? Physical? Did the opposing batters just get wise to him?

    Darned if know. But I suspect that there is something there. Beckett's 70th through 100th pitches feel less good than the ones that come before them. I don't know if the velocity falls off. I don't know if the aim falls off. I don't really care. Here's what I'd like Francona to say to Beckett:

    You get 85 pitches. That's it. If you finish the 5th inning with under 70 pitches, I let you start the 6th. But you never get the 7th until you've demonstrated consistent success in the 5th and 6th. Beckett gets more wins. The Red Sox get more wins. What's not to like?

    And who gets those extra innings? How about Wakefield? He gets plenty of advanced notice and gets to warm up like a starter. And I like the idea of him having a defined roll -- right now it looks like he's waiting for one of the starters fall prey to an injury...

    Will it happen? Will Francona bring down the boot on Beckett? Of course not. If I expected the Red Sox to try such creativity, I will find that I am forever Waiting For Francona.



    If I have anything to add to the global conversation about the Red Sox, it's this: I am an inveterate pitch counter with a Bayesian / Predictive Modeling background. Hopefully that will provide some entertainment...

    What is a Bayesian approach to pitch counting? Well it starts like this. Before the first pitch is thrown, I make two assumptions:

    1) The average starter pitcher lobs 100 pitches over the plate over course of a game. The best pitchers will go over that -- even into the 120s. A pitcher having a bad day (even a star) will often fall short.

    2) The average inning takes about 15 pitches. 105 pitches, on average, should take you through the 7th. Good innings go under 15 pitches. Bad innings can be interminable. If you only get 100 pitches (on average) to start with, throwing 30 pitches in the first inning means you've just lost the 7th inning.

    The following observation is also important:

    The middle relief staff is the weak underbelly of any pitching staff. The more innings they have to pitch the higher the probability of the other team winning.

    So, with that in mind, every single pitch changes the probabilities of the outcome of the game. If the pitcher gets through the 1st on just 12 pitches, he has increased the probability of making it through the 7th (and dare I say 8th?). But it runs deeper than that. Throwing a ball on the first chance increases the expected total pitches in that at bat. It also increases the probability that future strikes will get hit into play, and maybe even fall for hits.

    As I watch games, I take note of how the game hinges on these individual pitches, and that's what I intend to share.


    Before I begin, I should add two caveats. the first is that, I'm a busy guy. This means that I don't have time to test my theories quantitatively. If someone already has, I'm always happy to change my theories when faced with evidence to the contrary. This also means that I only watch Red Sox games. And "watch" should be in quotes -- I don't own a TV. I follow games on So I only learn from those pitchers I get to see, which can be limiting.

    Second, I've toyed with blogs before. I offer no promises. Maybe this one will last as long as some of the last (2 of which survived more than a year). Maybe I won't make it through the season. If you like what you see, feel free to leave me a comment. Like every pitch, every comment influences the probability of future outcomes.