THE BOOK--Playing The Percentages In Baseball

I’ve read MGL’s comments on managing and pretty much agree with them and a question occurred to me. The first thing I noted was that there are a lot of major-league managers who make many poor decisions. I live in Atlanta and I’ve watched the Braves a lot over the years. Bobby Cox has been one of the most successful managers in the game, but he makes some incredibly bad in-game decisions, and on a pretty consistent basis. Charley Manuel makes a lot of bad decisions, and made at least one in last night’s shortened game, but his team won the World Series. So, the question that came to mind was this. Since bad managers who make bad decisions don’t get weeded out by a Darwinian process, could it be that it’s because the effects of those decisions are relatively inconsequential in relation to a team’s talent level? In other words, is a team’s talent level such a huge factor relative to any tactical decisions that the manager might make that those decisions matter little in the overall scheme of things?

If the answer to the above question is “yes”, then the most important thing to have in a manager is the ability to handle people. A GM might also want someone who had a intelligent understanding of the game, from a sabermetric point of view, but that would be secondary, if not tertiary, and it might not have a lot to do with whether or not the GM ultimately decides to hire the person. I think if I was in the position of a GM who had the choice of hiring a manager who was good at handling his players but had a “traditional” understanding of in-game tactics, and one who had the personality of a dead fish but had a clear understanding of the game based upon sabermetric findings, I would choose the former.

You make a great point.  One would think that the Darwinian process would finally manage to get a manager who is a great tactician.  But, if we don’t get that, then there must be something even more important overriding that kind of manager from being in MLB.

For some reason though, we don’t seem to have this problem in the NFL.  I guess football players need less baby-ing than baseball players.  I thought baseball players were adults?  Maybe I’m wrong.

In the NHL however, coaches are predominantly former players, and the soft skills are far more valued than technical ones.  The Jacques Lemaires of the world (the hat-trick of former great NHLer, high soft skills and high technical skills) is rare among NHL coaches.

For some reason though, we don’t seem to have this problem in the NFL

Not sure I agree with this.  A fair number of NFL coaches over the last 20 years have been more or less relegated to this “leader of men” role.  It is not uncommon in the NFL for tactical decisions to be made by coordinator “gurus.”

In fact, I think this is the model that a MLB team needs to follow.  Keep the manager, hire an MGL type for “play-calling.” I’m a little surprised it hasn’t happened yet, to tell you the truth.

Tom,

I would argue that coaching plays a much bigger role in football and basketball. Take college basketball, for example. I don’t remember the guy’s name, but he coached at Princeton for many years. Even though they never had top tier talent, he’d always win a lot of games, and mostly because of his system. For some reason a remark made by Hubie Brown when he coached the Hawks in the 1970’s sticks in my mind. He said that he really liked baseball better than basketball, but that he’d chosen to become a basketball coach because you can’t coach a guy to first base.

Excellent point all the way around, so far.  I also appreciate that Tango is giving my managerial comments more “air time.” This was a very frustrating post-season for me.  My family was rooting heavily for the Rays.  This morning, my son called me and said, “One more thing (in addition to all the Maddon blunders we discussed).  With the runner on third and one out (in the 8th inning, I think), don’t you want a K guy on the mound?”

I replied, “Yes. you do.  And as you noticed, Bradford is not a strikeout guy.”

I mentioned this on the live blog.  Bradford’s strengths are low walks and low HR’s.  Exactly what you DON’T need in that situation.  Obviously, it isn’t bad, but each situation calls for you to possibly have the opportunity to leverage certain individual skills, and not just a pitcher’s (or batter’s) overall skills given the matchup.  It’s like lwts by game situation (or WPA/LI).

That was a minor criticism of Maddon, and I’m not even sure that he had anyone else in the pen, better suited to that situation.

Heck, if I had to talk about all the minor mistakes that Maddon (and other managers) make, my articles would be 10 pages long!

Interestingly, during the season some time, I wrote that I though that Maddon was a bad manager. Someone asked me why, and I replied that I could not remember but that was my impression after watching him occasionally the last couple of years.

I now realize the dynamic there.  I watch about 200-300 games a year, so I literally get to watch 10-20 games per team per year.  When I watch a game, I mentally and acutely note all the bad and good decisions (well, there aren’t too many good ones actually ) that the various managers make.  At any one time, I usually have a pretty good idea as to who I think are the particularly good and bad tacticians are.  However, I can’t tell you why because I have forgotten most of the instances that formed the evaluations.

I am biased at this point for several reasons (my rooting for the Rays and the “recency” of this post-season), but I don’t remember a manager making so many blatantly poor decisions in such a short time-span.  The guy literally butchered 1-3 decisions per game.

Let’s see how much WE that bunt play by Howell cost the Rays alone:

Top of the 8th, down a run, runner on 1st and 1 out, the home team (Philly) wins 74.4% of the time (wow, that’s a lot with an average batter at the plate of course (and average pitcher, etc.).  Bringing up a good pinch hitter is not a bad proxy for an average batter, I wouldn’t think.

How about a bunt attempt by a pitcher?

6% both batter and runner safe: .655 WE
63% batter out runner advances: .788 (you see how that result alone reduces WE by 4.4%, which is huge for one play!)
26% out no runner advance .816
5% DP .871

Total WE is .818 for the home team.  Wow.  He reduces the Rays’ WE by 7.4%!  You can’t do anything worse than that.  And that is one play in one game!

Anyway, back to the discussion.

There is not necessarily a natural Darwinian process in environments such as this.  As Tango says, there are likely other forces that get in the way.  To say that there should be Darwinian forces that ultimately create the perfect manager, would be to say that every business, every government, every family unit, etc., would eventually evolve into the prefect, optimal, entity.  That makes no sense because it is not true of course.

To some extent that kind of process goes on, but there are many other forces that work against it.

To argue whether in-game strategy skill (it is really knowledge and not skill) or “intangibles” are more important is a specious one on many levels.  First of all, we have to have a baseline for each.  I have no idea what that baseline is.

Secondly, why does it have to be an either/or?  It doesn’t of course, and to reduce it to that it not productive, nor does it even make sense.

Thirdly, one can probably be easily taught given the right manager, while the other one probably cannot.

In other words, is a team’s talent level such a huge factor relative to any tactical decisions that the manager might make that those decisions matter little in the overall scheme of things?

That is definitely true that a team’s talent combined with random fluctuation around a team’s true talent will dwarf the effect of a good or bad manager, both strategy-wise and intangible-wise, but by no means does that mean they “matter little.” The difference between an average manager and MY optimal one is probably equivalent to to the difference between an average player and as star or superstar.  Do star player “not matter” in the grand scheme of things?  Of course not.  It is just that the public, media, and baseball insiders, can readily “see” one and not the other for various reasons, not the least of which is that, for managers, they have nothing to go by.  They obviously don’t know a good decision from a bad one, otherwise there would only be the occasional bad decision.  When a manager makes a bad decision it is not like a player booting a ball or even making a base running blunder. With the latter, the player knows he made a mistake, with the former, the manager thinks he did the right thing (most, not all, of the time - he occasionally admits a blunder), given his limited knowledge and analytical skills.

I also agree that one of the factors that separates the sports is that yes, in baskets and football, it is true and well-known that the manager or head coach has many more things to do than just keep the players happy. Therefore the owners and GM’s know that they have to hire personnel with good strategic and analytical skills as well as good leadership and “people” skills.

And I think that some teams will eventually move in the direction of having a person like myself (and there are 50 or 100 other people that could do the same thing of course) advising the team/manager on things like this, and having a manager who is more than willing to participate in and implement them. It is only a matter of time.  Trust me, there are already teams that would such a thing if they didn’t have certain old-school persons working on the field and in the FO.  Eventually those old-schoolers will be replaced by new-schoolers, but like most things, changes in large, complex organizations take time because you have so many people and forces working against one another and because inertia in any business is a strong force.  Sometimes the presumed “wisdom-of-the crowds” is NOT a good thing.  Sometimes a meritocracy (where the experts rule), especially in “business,” is better.

One other thing, if all you want is for your manager to be better than the other manager, you probably want a similar “mold”, and you would someone who stands out a bit more.  To take a brand-new mold, with completely different ways to evaluate that mold, not knowing what’s better or worse, means you might not know if you got a good mold or not.

All you have to be is better than the next guy that is in the same family as you are.

***

MGL, what’s the WE if the pitcher swings away?

I personally think that a major-league organization would do well to train their own managers in the same way that they train players. First of all, they would create a “book” in which they explain the organization’s way of doing things. All of the orgnization’s managers would be expected to do things by that “book”. Younger guys would be hired to manage in the lower minors. They would be trained and evaluated there, just as the players are. Those who showed promise would be moved up. When the major-league job comes open, the new manager would be hired from within, among those that the organization has trained.

I get around a 4.3% drop in WE, with the pitcher swinging away.  That’s if I give the pitcher 18% for getting a one-base hit or walk, and 82% for getting an out.  That’s pretty much what I said in the other thread (.040 win difference with an LI of 3, which is pretty much what this play was).

Like I said, I agree that these are all jaw-droppingly stupid moves.  But, the impact of the manager is limited in single plays like this.

Even a badly organized batting lineup is going to cost you .010 wins.

It’s all bad, and they all add up to turn one of the best teams into one of the worst-team in the league.  A .01 win here, .04 there, .02 here and .03 there, etc, etc.  You get all these brain-dead moves, and suddenly you put yourself .20 wins in the hole, turning a .600 team into a .400 team.

...take time because you have so many people and forces working against one another and because inertia in any business is a strong force

I agree with this, and I think the effect is exaggerated in baseball (vs business) because it is not as competitive.  Each team has more or less a monopoly in its home city and there are no new “companies” emerging as a threat to that market share.  Any new “competitors” (owners) are vetted by the commissioner and the other owners.  The incentives to take risks are marginalized by this relative security.

This is why I think some form of relegation would only help the competitiveness and level of play in the sport.

Gameday has Howell’s bunt in the top of the 7th with the game tied 3-3.

Tango - Don’t forget to factor in the much higher chance of a double play with the pitcher swinging away with a man on first and one out.

How come there are never any posts about what manager’s do correctly?

Or does that never happen? And if so, why aren’t you advocating the aboltion of human managers and have them replaced with computers?

Who said anything about using computers?  It seems that there are at least one thousand, if not one million, fans that are better tacticians than the current crop of managers.  Why would we need to use computers to do this job?

Maybe managers are better motivators of people.  That doesn’t mean they must ALSO be the tacticians.

Ideally, the tactician would say to the manager:

“Listen, I want to move Utley from 3rd to 2nd, and put Werth in between Utley and Howard.  Do you think Utley will cry like a big baby?  Do I have to explain to him how this helps the team?  Do you think Werth is going to wet his pants that he’s going to be between our two best hitters?  Should we send him up there with a bib?  Let me know when they turn 18 years old, so that I know they are now men, who can handle b.s. like this.”

So, that’s what I want my tactician to do: figure out what helps the team the most, and then let the pseudo-pyschoanalyst manager figure out the best way to sell the plan to his immature players.

Is that too much to ask?

Tango - Don’t forget to factor in the much higher chance of a double play with the pitcher swinging away with a man on first and one..

Peter, pitchers do NOT hit into more DP’s than a typical position player.  The DP rates per PA are about the same, simply because they strike out so much.

Tango, I can’t easily figure the WE for a pitcher hitting.  I have a Markov program that spits out RE but not WE. I’d have to put it in my sim, which is a pain, but could be done. 

So you are saying that the pitcher batting lowers the WE by 4.3% from a non-pitcher batting?  And then another 3% or so for the pitcher bunting?  So the pitcher hitting is better than the pitcher bunting?

I am not surprised at that, although I think it is closer.  In my research for The Book, I found that an average pitcher hitting is close to him bunting with a runner on 1st and 1 out, and thus, the manager should bunt them about half the time.  This is another huge mistake that ALL managers make, and that is, bunting their pitchers around 80% of the time with a runner on 1st and 1 out.  And it is not like all managers are doing it 80% of the time and thus “mixing it up.” It is that some managers, like LaRussa, are mixing it up and others are bunting ALL the time in that situation.

How come there are never any posts about what manager’s do correctly?

I guess if you want to know all the correct ones, just subtract all the incorrect ones I mention from ALL of the decisions, and there you go!

What am I, Fox News, fair and balanced?  If I were working with a manager or coach, I would use a more positive approach of course - pointing out what they did right and then saying something like, “Hey, and with this play, have you thought about this?  I did some research and came up with this.  Do you think I made any mistakes in my assumptions in the research?  What was your thinking and your rationale?  I would like to know that so I can consider it as well and perhaps even incorporate it into some of my models?”

If my manager gave me the googly eyes, while thinking, “What the F is this idiot talking about?” well obviously that is not the manager I would want to employ.

You give the manager the proper information ands then you let him make the decisions based on the things he might know that you don’t or are not in the model.  At the same time, you let him know which decisions have discretion (and by how much) and which don’t.  You go over the numbers with him so that everyone is clear about what each one is using to come up with the right decision.  If necessary you pull rank (or fire him).  For example, if your manager makes a lot of decisions based on players being hot or cold, or players being perceived as clutch or not, you tell him that your research and that by others pretty much has proved that clutchness and hotness or coldness does not have much predictive value, therefore he should only use it as as tie-breaker.  If he insists that you are wrong - well, you are in trouble, because he is not going to listen to you about other things as well.

Tango hit the nail right on the head.  The whole point of this kind of thing is to simply give managers and coaches more information in order to make their decisions.  If, with that information, they still continue to make the wrong decisions, you work with them some more, in the best way you can.  If they still insist on doing it their way, then they get canned.

With the pitcher bunting thing, you go to Maddon in a couple of days, and you ask him (nicely, non-threatening or accusatory), “Hey what was your thinking when you had Howell stay in the game rather than going to a pinch hitter?  A few of us were having a discussion about what the best decision was in that situation.  We definitely want your input.”

Then you gently tell him that it is possible, with a computer, to figure out the team’s chances of winning, on the average, with Howell bunting, and with a pinch hitter hitting.  Then you tell him of course that means using a hitter off the bench and needing another pitcher. You ask him his opinion on the effects of that.  You ask him what benefit he saw in leaving Howell on the mound.  Etc.

Then you fire his ass!

Peter, pitchers do NOT hit into more DP’s than a typical position player.  The DP rates per PA are about the same, simply because they strike out so much.

MGL - Yes, the chance of a pitcher hitting into a double play is only slightly greater than a non pitcher, 14.3% vs. 13.3%, but that was not my point.  In Tango’s post #8 it doesn’t appear that he had figured ANY chance of a pitcher hitting into a double play.  And a 14.3% chance when swinging away is much higher than the 5% chance of a DP when the pitcher is bunting.

Peter, sure.  I thought that his swinging away WE was a little too high.  I think that it should be around the same as bunting.  That is for an average hitting pitcher.  Obviously it varies a lot among pitchers.  In this case, the WE for a pitcher hitting is irrelevant.  The 100% correct play is a pinch hitter.  I don’t think any rational person who even vaguely understands the numbers would quarrel with that. I have read several articles and comments and no one yet has even remotely suggested that Maddon made the right move.  It even “sounds” silly when I write “right move.” How can it be the right move?  So how can a manager, who is supposed to be the “expert,” do something that NO ONE thinks is even remotely correct, and it is not that he knows something that no one else knows?  I don’t have a clue.  I would feel better if he admitted that he froze up and blew it.  Of course I don’t know what he really thinks, but reflecting on it and still thinking that he made the right move is the worst kind of stupid.

If I use the following numbers, I get .794 for the WE when an average pitcher hits away.  It is going to be higher, because I am not considering that the defense is actually playing for a bunt, which they would be.  I probably can add around .006 wins to make it an even .800.

s or ROE (10% with normal IF, and 15% with infield playing for bunt, with a corresponding change in the other numbers)
d 1.8%
t .14%
HR .3%
BB or HP 3.8%
K 37.8%
batted ball single out 33.2%
DP 13.0%

I assumed that on a single, the runner advances 1 base 2/3 of the time and on a double, 2 bases 2/3 of the time.  I did not assume the lead runner being thrown out at all.

So I have .800 for an average-hitting pitcher swinging away, .818 for bunting, and .744 for an average hitter swinging away. That is the home team’s WE in the top of the 8th.

From Buster Olney of ESPN.com, here is his take and Maddon’s on the Howell bunt:

Joe Maddon is going to be criticized in some quarters for sending up J.P. Howell to bunt in the seventh inning, but I thought this was a coin-flip move—I could see it either way. He wanted Howell to pitch to Pat Burrell leading off the bottom of the seventh inning, as he told reporters on background before the game, because he liked the matchup of Howell’s breaking ball against Burrell, who has trouble with breaking balls. And the reality is that Maddon didn’t have as much bullpen depth as he did at the beginning of October, because right-hander Dan Wheeler just seemed to lose his fastball as the month went along. “There’s just nothing on it right now,” a talent evaluator with another team said this week.

My own question about that move would have been to wonder how he could ask Howell, who is not accustomed to bunting, to advance a runner. But in the end, Howell got down a great bunt. Aki Iwamura smacked a single up the middle, Chase Utley made a great play, and if Utley hadn’t, Jason Bartlett would’ve scored the go-ahead run, with Howell set to provide the matchup Maddon wanted, against Burrell. So it wasn’t a mistake so much as it was a matter of Maddon’s decision not working out the way Maddon had hoped. Bartlett’s gamble came up short.

Of course, we don’t care what Olney thinks about the play (coin-flip), because he is one of the least analytically literate of the ESPN.com crew. We do care about what Manuel says, if only to muse on his rationale.

So he wanted Howell to face Burrell because he thought that it was a good matchup because Burrell does not handle curve balls well.

Without going into any detail, that is complete BS on many levels.  And even if there is some merit to it, the “hitch” here is that Maddon apparently has no idea of how much WE he gave up with the bunt. If he did, even he would know that no amount of “matchup” between Howell and Burrell could make up for that.

O.K., I can’t resist.  Burrel has trouble with breaking pitches from LH pitchers? Givemn that he has a .966 OPS versus LHP the last 4 years, whose breaking pitch does he have trouble with?  Howell is the only LHP who throws breaking balls to RHB?  He had no right-handed pitchers with a breaking ball (Balfour who has seemed to throw nothing but fastballs this series had already pitched)?  Don’t ALL batters have trouble with breaking pitches, or at least most (they should)?

So while we always like to know what managers are thinking when they make any decision, especially a particularly poor one, in this case, his reasoning is B.S., and more importantly, it really doesn’t matter because the WE he gave up on the offensive side could not possibly be made up for with a favorable pitching matchup.

Why does Olney simply listen to Maddon’s reason and assume that it is correct?  Of course Maddon has a reason (at least he’ll tell you one).  That would be like, I question a person’s decision, so I ask him about it, and then no matter what he says, I say, “Oh, OK. I guess you were right.”

Have I beaten a dead horse enough?  I don’t think so.  I am going to tell anyone who will listen, willingly or not, what a bad in-game manager Maddon is.

If anyone has James Click’s email address, please send it my way.  I want to vent to someone who actually might give a crap.