THE BOOK--Playing The Percentages In Baseball

I think his quote is fine, all things considered.  That’s about as far out there that I’ve ever seen a manager talk about the sac bunt.

Baby steps, MGL, baby steps.

Even if he didn’t completely get it, he’s thinking about it. To me, that’s the most important thing. He’s questioning and thinking and not accepting conventional wisdom just because it’s been passed down from generation to generation. To me, that’s what the work you and Tango and everybody else is really meant to accomplish: Get people to start looking at all this baseball stuff and not just going with what we’re told is true.

I don’t know much about Yost, but as has been said before, some of the saber-friendly guys who follow the Brewers closely say that he talks a good saber game, but that’s about it (just talk).  That is just what I heard.

I was just being facetious with my sac bunt comment.

I would like to think that a manager that actually took the time to read The Book would do more than just “not be such a proponent of the sac bunt,” especially since nowhere in the book does it say that.  In fact, quite the opposite - at least the opposite of conventional sabermetric, or pseudo-sabermetric, wisdom, or what people think that sabermetric wisdom says.

Being a Cubs fan, I’ve gotten to see many games the Brewers have played over the years, particularly last season when watching them actually had meaning when it comes to the Cubs.  Yost may understand that the bunt is overused, but that is about it.  Ned Yost was the Cubs MVP in 2007 and had they stayed healthy this year, I have no doubt he’d have been the Cubs MVP in 2008.  I don’t imagine he’ll have a job with the Brewers after this season. 

If he really has read The Book, I’d like to suggest he re-read pages 16 through 334 because he’s not implementing anything from it.

Unfortunately, David is right. Although Ned has read the book, it appears he has not followed through with much of anything in it.

As a Brewers fan, it’s frustrating to see this. Ned doesn’t know how to handle hot/cold streaks, platoons, and does a poor job with the bullpen and pitchers in general. You’d think that after reading The Book he’d come away from it and actually utilize the strategies presented.

Never bunting is better than bunting excessively, correct?

Never bunting is better than bunting excessively, correct?

Not necessarily at all.  The optimal strategy is sometimes bunting and sometimes not, making sure that for every potential bunt situation you randomly mix up your bunts and non-bunts.  As you move away from that optimal mix, in both directions (too much or too little), you lose win expectancy.

I suppose we could figure out how much never bunting costs.  Then, you would still have to specify how excessive the bunting was in order to determine which was better.

Never bunting is a horrible strategy.  The defense would never have to play other than a normal position which would take away some extra hits when you hit away.

The biggest mistake that ALL teams (on offense) make is being too predictable.

Tonight in the Mets/FLO game, Beltran was up in the first inning with runners on 1 and 2 and no outs.  I assume he rarely bunts, especially in the first inning and especially when hitting lefty.

The third baseman was playing back.  Someone is making a mistake (or both).  Either the third baseman should play up a little or Beltran should NOT never bunt or should bunt more often than he does in that exact situation.  Beltran did in fact lay down a bunt and was easily safe at first on a single.  As I said, someone likely made a mistake there - either the defense or Beltran (his manager).  Either the defense should not be playing so far back, or Beltran does not bunt often enough in that situation.  I can’t tell who.  It is possible that both teams were not making a mistake.  The only way that would be possible is if the defense sometimes plays up with Beltran in that situation and sometimes plays back, randomly mixing it up.  That is probably not the correct way for the defense to play optimally though. The reason is that the offense sees the defense’s “hand” before they have to play theirs.  Thus, the defense should “mix up” their strategy by always playing somewhere in between all the way up and all the way back.

In that situation, Beltran must bunt sometimes and the infield must either play up sometimes and back sometimes (randomly) or play in between all the time.  That would be the “Nash Equilibrium” point in game theory.

Sometimes the bunt or no bunt is so lopsided that it is correct for the batter to always or never bunt and the defense to play all the way up or all the way back.  That does not happen as often as most people, including managers, think.  We can easily prove that by showing that with the defense playing all the way up, as you would expect with a light hitter at the plate in a tied game in the very late innings, the result of a sac bunt attempt is not very good - probably equivalent to the weakest batters swinging away. Even a good hitting pitcher fares better when swinging away than bunting with the defense playing all the way up. 

With a good hitter at the plate in the early innings when the defense is not expecting a bunt, we can show that the result of the rare sac bunt attempt (by a decent bunter with decent speed of course) is very good - better than a good batter swinging away.

The result of the sac bunt is so sensitive to where the defense is playing that, as I said, it is relatively rarely correct for any batter, no matter how good or bad he is at hitting, to always bunt or always swing away.  I would say that for the majority of batters they should be in the 25-75% range for bunting or non-bunting.  Just a wild guess, but somewhere in that range.

Yost seems to be willing to at least TRY to not use “traditional manager theory” as a crutch.  He’s batted Kendall 8th for most of the year, so he’s at least willing to try new things (although he’s still using the results of small samples to test them).  Yesterday, he also pulled Sheets (who was at 86 pitches) in the top of the 7th for a pinch hitter (down by 1 with runners at 1st and 2nd).  That might not sound like much but it’s something he usually wouldn’t do.

Like most managers, though, he loves to “make things happen” by attempting double steals, hit and runs (even with pitchers), suicide squeezes, etc… He’s also more than happy to defer to his gut with the game on the line, so what’s the point? 

He WANT’S to be known as a cerebral manager.  I just don’t think he has it in him.

MGL, I understand the necessity of random variation within a range in order to be unpredictable. I guess I should have phrased my question thusly:

Isn’t never bunting (and being thus predictable) better than always bunting when the defense expects it, and so never having the advantage of bunting to a surprised defense?

This is not an irrelevant consideration. Until we get some managers who excel at game theory, they need cut-and-dried strategies to follow.

(Personally, I liked Beltran’s bunt when the defense wasn’t expecting it, but not so much when I consider how terrible the Mets’ 5-9 hitters were behind him.)

Here’s a very broad statement that I feel is necessary just because game theory is talked about so much: It’s hard to use game theory if the other guy is too dumb to realize you’re doing it.

You really don’t have to mix things up as much as logic would dictate because most major league managers aren’t using logic. They aren’t reading trends. They aren’t figuring out how often you do things. It’s sad, but they just don’t work this way.

I would bet a team could predictably follow the numbers (as in never doing something randomly, just doing what they’re “supposed” to do) for at least a full season before any team caught on. It could probably go on for much longer.

When Felipe Alou was asked which manager he hated to be opposed against, he said Bobby Valentine.  Not because he was so good, but because you never knew what he was going to do. 

Either Bobby practiced game theory so well as to “bluff” the right amount of time, or he was so inconsistent as to go by gut instinct all the time, or Felipe was so rigid as to always telegraph his plays.

I love Felipe.  In 1994, he was the most popular Expo, bar none.  We had Moises, Walker, Wetteland, et al.  We loved them all, but Felipe we loved the most.

OK, it’s also now clear that Yost skipped the entire chapter on streaks and has no idea what a run expectency table even is. 

Down by 1 in the bottom of the 5th, with a runner on 2B and 1 out, Yost intentionally walks Chipper.  Of course, Teixeiria follows that up with a 3 run blast, effectively ending the game.  But Chipper is batting .425, so it must have been the right call!

The Book, p.305, for the above situation, Andy doesn’t list anything, and therefore, no walk is allowed according to Andy regardless of the gap between the hitters.  And Chipper and Texeira don’t even come anywhere close to having a meaningful gap anyway.

***

If it was guys on 2B and 3B, then you need to have one batter have a wOBA that is 25% higher than what is due up.  “Due up” is 44% of each of the next 2 batters, plus 12% of the 3rd batter.

If the next three guys up were league average hitters, the guy you are walking needs to have a wOBA of .425, which basically means he’s got to be the best hitter in the league.

Okay, so I got hooked up with pre game BP passes and tickets from a friend last night. He got him at an auction, and the affiliation was Ned Yost. So he leads us around and I had the opporunity to talk to him.

I asked him whhether he has read The Book, he said yes. And here’s the all important Q&A: (rough quotes)

Me: How much of the Book do you incorporate while managing?
Yost: Not much of it.

He then said that he was never much of a fan of the bunt anyway, becuase outs are at a premium. He mentioned the intentional walk and how he doesn’t like it much. He said it’s “interesting stuff,” but it’s clear he thought of it more as interesting than actually useful.

A friend of mine asked his reasoning for the pitcher batting 8th, and he mentioned the DP, the second leadoff, etc., and then said we estimated it would add 35-50 runs over the year. I don’t even need to do any math to know that’s incorrect. Didn’t the Book see a difference of about 3 runs? I asked him if they had a fulltime stats guy, he said yes, so I’m assuming that’s where he got that number from. It’s surprising that a proffesional sabermaterician would find that number.

So ya, that was my convo with Ned. It’s unfortunate that he won’t utilize the Book, becuase he could use the help.

T-Nast, interesting.  ALL managers could use the help.  I watch a lot of games, and I doubt that Yost is any worse than the average manager.  My guess is that every single manager could add between 2-3 wins a year with optimal strategied, the best ones adding 2 and the worst ones adding 3.

As far as never bunting or always bunting, again, it is not really a comparison you can make unless you define “always.”

But, yes, probably never bunting is better than alwsy bunting in a fairly obvious bunting situation.  When the defense expects a bunt 90% of the time, the bunt is generally a really bad play because the results are so poor.

It is not that hard to “teach” a manager to randomize his bunts, for two reasons.  One, I could easily give him a couple of rules of thumb that make it appear as if the bunts were random, but they would not be.  Two, it is not that important to get the optimal mix correct.  A few rough rules of thumb given the type of batter and the situation would go a long way.

When we talk about “game theory” we are talking about using the optimal strategy depending upon your opponents’ strategy and the chance that your opponent changes his strategy in response to yours.  If your opponent is an expert game theorist, that requires a certain strategy for you (basically the “nash equilibrium one").  If he is stupid then that requires a differeent strategy.

Yes, teams are stupid, but one thing they do is a lot of advance scouting.  If you never bunted, for example, teams would know that quickly and they would not play that close in a potential bunt situation.  And if you bunt a lot, like some managers do (not so much anymore of course), teams will know that too, and they will play up alot.  You really can’t fool teams all that much.

Again, with the sac bunt, the offense has the advanatge of generally knowing where the defense is playing so that true “game theory” where one side has no knowledge of the other side’s strategy, does not come into play.  Of course, because of that, it is incumbent upon the defense to choose a position for the infielders that is optimal for each batter and for each situation such that no matter what the offense does (bunt or not bunt) they can gain no advantage.  If the defense does not do that, then the offense can either always bunt or always not bunt in any given situation.

But, in practice, you don’t want to do tjat against a astupid defense, because you will quickly force them to play more optimally and perhaps perfectly optimally.  For example, let’s say that in an “obvious” bunt situation (with a non-pitcher at bat), the defense plays down the batter’s throat, which is typical of many teams.  The correct thing for you to do is to never bunt against that defensive configuration. But if you adopt that strategy, they will quickly or at least eventually stop playing up so far.  In fact, they might end up playing perfectingly in position.  You don’t want that.

So you still want to bunt some times and you hope that they will still play too far in, even though you are not bunting that often.

Teams are much more likely to play too far in in “typical” bunt sitations.  Then again, against very good hitters who never bunt, they tend to play too far back (actually not too far if these batters never bunt).  The mistake is that these very good battes do not bunt enough to at least force the defense in a little.

Again, most of this is common sense, and does not require one to be a mathematician or expert game theorist.  It is just that the inertia of tradition, stupidity, and ignorance among baseball insiders is a powerful force.

I really hope that it wasn’t Dave Lawson (Brewers’ stats guy) who came up with that estimate.  For those that don’t remember, he was mentioned here a few months ago:

http://www.insidethebook.com/ee/index.php/site/comments/brewers_sabermetrician/#comments

I should also mention that, wherever that estimate did come from, the Brewers must not have taken it very seriously.  The pitcher has been batting ninth for over a week now.

30-50 runs is a ludicrous estimate.

I’m sure if MGL runs his sim, he’ll get similar numbers like mine: when I put the pitcher in the cleanup slot, that costs just 0.1 runs per game (16 runs for the season).  To think moving him down just 1 slot would be around 40 runs?? Nope, not even close.

The Brewers were including in that estimate the fact that Kendall would hit into less double plays batting after the pitcher.  What did they think, he’s going to hit into 100 double plays?