THE BOOK--Playing The Percentages In Baseball

Friday, October 30, 2009

Some thoughts on Game 2, other than the obvious ones, like AJ pitched a good game…

Of course Pedro should not have pitched the 7th inning. In fact, he probably should have come out in the 5th with Damon at the plate and a runner on second and 2 outs.  The reason is the same as I have been articulating for 2 weeks now.

Good move by Girardi to throw Mariano for 2 innings starting in the 8th.  He could have brought Burnette out for the 8th, but that would have been much worse of course.  AJ the 4th time through the order or Mariano?  Not even on the same planet.

We can’t seem to get through a game without bad baserunning (Werth), bad umpiring (Utley was safe on the DP and I THINK that Howard trapped the ball although I don’t think the replay was conclusive - to me at least), or both.  OK, those were obvious observations, like AJ pitched well.

McCarver made an ass of himself as usual trying to predict pitches or by saying that so-and-so “must” throw this pitch. He is obsessed with pitchers throwing fastballs at 3-0 and 3-1 counts (they “must” or they “will").  I have an urgent text message for McCarver:  They don’t always do that and certainly shouldn’t, depending on the game situation of course.  I am usually very good at “guessing” pitches (with is actually a testament to how poorly catchers and pitchers randomize their pitch selection), but with Pedro on the mound, I have no chance.  Virtually at any count he seems to randomly select a pitch and location among several.

Nice job bunting and then not bunting by Melky in the 6th.  That is the way it is done folks!  You keep the defense guessing and you force them to play somewhere in between up and back.  Whether A-Rod was playing too far in when Melky hit away, I don’t know.  But if he was playing optimally, it would not matter whether Melky bunted or not, as I explained in my “article” on bunting and game theory on Fangraphs.  Now, whether a hit and run is ever correct, and if so, when, I have no idea.  There is not much game theory involved in the hit and run since there is really nothing a manager can do to counteract it other than if they knew you were going to do it nearly 100% of the time they could pitch out.  But it is never going to be done that often to ever warrant a pitch out.  And that means, actually, that it is probably never correct.  It is almost never correct to do something only occasionally, unless the correct game situation only comes up occasionally. 

I don’t know if Jeter bunted on his own with 2 strikes of if Girardi gave him the bunt sign (unlikely), or he missed the sign, but it is NEVER correct for a position player to bunt with 2 strikes, even if the defense is all the way back (which they should be).  I suppose that there is a BE point (the percentage of time that you don’t foul it off), but it would have to very high.

The most controversial play was when the runners were not run running on the 3-2 count to Utley.  McCarver of course was going on about how the runners SHOULD have been going.  His reason?  Because if they were, there would not have been a DP!  Yes, he is an idiot.  Actually he talked about how Utley is a contact hitter, the runners were fast, even if Utley struck out, Rollins might be safe at third, etc.  All of which is true, but…

If you are waiting for me to tell you the correct answer, well… Do you remember how Tango says that an opinion without supporting evidence is B.S.?

I am in exactly the same position that managers (and McCarver, and virtually everyone else in the world) are in when it comes to this (and all other reasonably close) decisions.  I don’t know the answer!  Neither does Manuel or McCarver or Bill James (I assume).

Remember when I told you that one of the differences between a smart person and a not-so-smart one is not necessarily in their respective breadth of knowledge (or some such thing)?  That the difference is that the smart one knows when he doesn’t know something and that the dumb one thinks he knows a lot more than he does?

Well, most managers (and commentators like McCarver and people in general) think they know the answer to a question like this (should the runners have been going) and they will do what they think is right and claim that it was indeed right (whether it worked out or not).  How would they (know the answer)? You can’t “reason” it out. It obviously depends on how often Utley would hit into a DP if the runners are not going and how often he would line or fly into a DP with the runners going and how often one or both would be safe in a K, as well as the value of the potential distraction to Utley if they are going.

I don’t know the answer.  I couldn’t even guess.  I could figure it out, though, quite easily.  At least I could determine whether it is almost definitely yes, no, or flip a coin (go with gut), which is all you want to do with most decisions that involve some unknowns.  You know how people sometimes say that there are things that a manager knows that you don’t know that could affect the decision, or that there are some unknowns that you can’t put into your model?  That is true, but…

I’ll give you another important concept, one that I have also discussed before.  Try to remember it this time!

If I construct a model which is NOT perfect, for various reasons, to come up with one of three answers:  Definitely right, definitely wrong, or I am not completely sure, then all the things that I don’t know will ONLY effect the third category.  And I make sure that I construct the model and group the answers so that is true 99% of the time.  For example, let’s say that I want to create my category, “Definitely no.” How do I do that? What I do is I set up the model and assume that all the things I don’t know are in favor of the answer being, “Yes.” If the answer is still, “No,” then I am satisfied that even if the manager knows things that I don’t know, the answer is still, “No.” I hope that makes sense.

Anyway, back to the runners going or not…

How many managers do you think ever picked up the phone and asked someone in the front office, “Hey, can you ask one of our stat guys to figure out if we should send the runners in a situation like that?  I’d really like to get that right next time instead of just guessing.  Obviously I have no idea what the correct strategy is.  How would I?  What do you think - I am a computer or something?”