THE BOOK--Playing The Percentages In Baseball

One of these days, someone will quantify to some extent a manager’s contribution to a team’s WE from his in-game strategy decisions.  That might turn out to be plus or minus 1-3 wins a year.

...

Other than that, and at the risk of repeating myself too many times, we have no idea whether, if, or by how much any manager influences his team’s performance, and you will never catch me writing an article trying to figure out if some manager should be fired or not.

Isn’t ±1-3 wins/year a good enough reason for a manager to be fired? ±1 win would have helped the Mets or Padres a lot. If you can point to a manager’s decisions over a 3 year span and say, “He’s lost us 3 games through demonstrably poor decisions” isn’t that a good enough reason to fire a manager?

Let’s say I run a company. I have 30 different managers of 30 different groups all producing similar widgets. Naturally there is some variation in widget-making, and just because one group’s production is down or up 1% from projections doesn’t tell me that it was the manager’s tremendous management.

Now I’m a nice president, and I like happy employees. So I give all my managers beer to provide employees with on Fridays. 28 of my managers feed the employees 2 beers with lunch, and 1 during break time. One feeds the employees 3 beers first thing in the morning. One feeds the employees 3 beers right before the employees drive home.

Now it’s possible that even without 3 beers every morning that said plant would underproduce, or that even with 3 beers every morning that plant could overproduce. And it’s possible that the increase in accidents for the ‘beer before driving’ managers was entirely coincidental.

If my assistant, who’s supposed to be paying attention to performance and all that sees these managers and says, “I can’t make a judgment about which are doing well or poorly because we just don’t know about additional effects” then they would (quite rightly) be fired.

I understand that you are saying, from a purely analytical point of view, that we can only quantify some effects that managers have with any reliability, and that advocating firing on the basis of limited information isn’t sound even if we can point to some reliable data one way or the other.

From a realistic point of view, however, I think it’s pretty safe to say that a manager who loses 3 games above average due to poor bullpen management or pinch-hitting deserves to be fired just as much as the managers who are feeding beer to their employees at irresponsible times.

Even if we could show over the long term that the manager, even though poor management is costing 3 games/year, actually provides a net positive due to the things which are far more difficult to measure, it would take too many games/seasons/decisions to be confident about to risk betting on the data we don’t know trumping the data we do know.

As a mets fan I can say that I agree with him as far as Randolph not being a very good tactical manager.  I dont pretend to know how much in terms of wins this has costed us (I imagine somewhat small).  However, they lost the division by one game.  Im not putting the total blame on Willie, but another manager and things could have been different (or maybe not).

The point, as I made in some other thread, is that the spread in talent is so tight in MLB, that the random variation simply dwarfs it.

The beer analogy doesn’t work, because you’ve got such a huge sample to work with, plus you are introducing something so effective.  It’d be like pinch hitting Albert Pujols for Randy Johnson.  Teams don’t do that.  It’s pinch hitting a guy with a .340 wOBA for a guy with a .320 wOBA with the platoon advantage.  Do you realize how tight it is to have a die that rolls 34% success for one guy and 32% success for the other guy?  How the heck would you even know who has the weighted die?  How many rolls would you need?  Replacing a guy with a 45% success with a guy with 10% success is plainly obvious that you require only a few rolls to distinguish.

Yes, accounting for 1 to 3 wins is enormous.  Teams pay 5 to 10MM on the free agent market for those wins.  But, how the heck would you be able to figure out the impact of one manager has on 13,000 PA that totals 1 to 3 wins?  That would mean adding up 0.0001 wins per PA.  Or, if he only affects 10% of the PA (1,300 PA), adding up 0.001 wins per PA.  Or, if he only affects 1% of the PA (130), adding up 0.01 wins per PA.

The only way to do this is to actually sit down and track every single decision he makes, and compare those decisions to every other manager.

(Andy for example did a great job in tracking pinch hitting by managers.)

Until that happens, this is all yapping.

It’s easy to look back now and say the Mets lost by one game and could have won the division had they won a tough loss back in May or June but the fact is that winning that tough loss likely would not have produced the same post-win results.  It’s not as if everything would remain exactly the same and that one reversed decision would be THE factor.

I hear this all the time with announcers pertaining to runners on base and Harry Kalas did it last night as well.  Phillies up 2-0 and Pedro Feliz knocks in two runs to make it 4-0.  Prior to his double, Ryan Howard was thrown out at the plate (he was clearly safe) and Kalas made the assertion that Feliz’s double should have made it 5-0.  Just because Howard’s run would have been the third does not necessarily mean Feliz would have followed with a 2-run double.  Maybe it does; or maybe Howard scores and Feliz weakly grounds out.  We don’t know for sure because it didn’t happen.

Yes, the whole idea if “this WOULD have happened if this HAD happened, and we would he HERE” is a silly one.  But, that is the way the human mind works.  My mother leaves a poker machine in Vegas and the next lady hits a royal after she leaves.  My mother thinks that if she had played one more hand, SHE would have hit the royal.  That is the way the human mind works.

One issue though, is the use of weaker pitchers in high leverage situations.  Scheoneweis’ usage and Mota’s usage were terrible.  Even just using the LI work, you’d come to the conclusion that Willie doesn’t optimize his team’s chances for success.  *Could* Schoeneweis get RHBs out at a given moment?  Sure.  Is he likely to?  Absolutely not.  Only one person has the ability to set those matchups, and it is the manager (or the pitching coach). 

“On average”, how many games would the Mets won if Willie had used a differnt pitcher?  Well, a lot more.

Willie, in the poor tactical dept., also violates seveeral of MGL’s hard-and-fast rules.  He will slap three or four LHBs in a row.  That hurt the Mets several times last season, allowing the opponent to bring in a LHP to just hammer away at the lineup.  He’s doing it again this season with Delgado, Church and Schneider.  And sometimes Chavez for a fourth.

Clarification - he went: Delgado, Anderson, Schneider, Chavez.

Looking at 2007 batting orders he doesn’t appear to do that much.  But the other thing he does is construct a really weak bench - three second basemen of five possible slots?  I mean, come on.

Chris’ points are exactly the type of points that need to be brought up, things that you can track, quantify and qualify.

***

As for LI:
http://www.fangraphs.com/winss.aspx?team=Mets&season=2008

Wagner’s LI is only 1.45, which is very low, especially since the bullpen overall is at 1.03.

Here’s the Mets:
1.45 Wagner
1.36 Sanchez
1.20 John Doe
1.18 Heilman
0.98 Schoenweis
0.95 Sosa
0.61 Feliciano

Normally, you wouldn’t have it so tightly bunched.  Either the Mets games have been such that there’s been little room to maneuver, or Willie’s not doing a good job of identifying who his better relievers are. 

Is Wagner not at 2.0 because the Mets simply haven’t had enough high LI games for him?

If I look at the Marcels entering, Heilman and Duaner deserve all the setup time they can get.  Heilman is making Willie look bad with all those HR he’s given up already.

Feliciano dropped alot in trust from 2007.

He actually has a pretty good group of relievers.  The spread of LI looks pretty good to me, except for the low score on Wagner.

That’s not to say that the individual decisions are necessarily good or bad in each case.  But, overall, it looks fine.

The pattern in 2007 also looks fine (though he had Wagner at 1.7, which is more reasonable, though still low).  I have to believe that Wagner is coming in into alot of mopup situations, and he’s riding the bench when he shouldn’t.  Such is the curse of the 9th inning reliever.  (Hat tip: Tony Larussa)

Heilman has nearly given me a heart attack this year.  The past two seasons he’s been hurt with the long ball. 

In 2007, the .872 OPS in high leverage situations didnt help either.  Its hard to trust someone when they consistently come in high leverage situations and fail to deliver.

It’s gotten to the point where there’s been talk of sending him down to AAA.  We thought it might happen yesterday when they called up Muniz but they put Wise on the DL instead.

Chris wrote this at BTF, in response to MGL’s claim about teams over or underperforming by 10 or 15 wins:

I’ll need some demonstration, not just hypotheticals. Not that they *could*, but that they, in fact, do. This is just as made up as whatever he’s ######## at Sheehan about.

MGL, I believe, was simply invoking the binomial distribution.  If a team has a true talent of .500, then we expect that one team will end up with a W/L record that is 10 wins above that, just by sheer random variation (luck).  And another will be 10 wins worse than .500 on sheer luck.

MGL’s pronouncement on this seem ambiguous.  It could be read as 1.5 teams being +15 wins and another 1.5 teams being -15 wins, or it could mean that 1.5 teams will be +/-15 wins.

Regardless, I’m getting 1 team at +10 wins and another team at -10 wins.

Wagner is pitching in some mopup situations because he “needs work” because he hasn’t pitched in days because Willie almost never uses him except in a save situation. Tie game in the eighth inning, he’ll use anyone else but Wagner.

Felicano’s low-leverage usage is more baffling. His numbers have been tremendous, and he doesn’t have a massive platoon split like Schoeneweis; however, Randolph uses Feliciano like the LOOGY and Schnoeneweis like the guy who can get both sides out.

Actually, all of this was far more pronounced last year. Willie has gotten much better this year except for the curious favoritism for Jorge Sosa, who is barely good enough to be pitching in the big leagues (and he’s not, anymore).

Re: the original topic. Just because Willie’s bullpen mismanagement cannot be measured in wins doesn’t mean it’s not real, and I don’t see why Sheehan’s claim should be ridiculed.

Btw, here’s the BTF thread (*):
http://www.baseballthinkfactory.org/files/newsstand/discussion/the_book_blog_mgl1/

It’s actually a decent read.  Post 14 is my favorite. 

(*)I’m just waiting for the yahoos to come in and ruin a good thread, by taking the opportunity to slam mgl on something completed unrelated.  “Hey look, mgl said something loud and boisterous.  Let’s get him!” You can almost see it in some posts, but the rest of the posters are doing a great job of sticking to the issues.

Agreed about Feliciano.  He can get both lefties and righties out....Show just lefties.  Feliciano should definitely have a higher LI then he does right now.

I wondered this about the mets bullpen.  Show struggles against righties.  Heilman and Smith have struggled typically against lefties.  Its hard when you have a number of relievers that splits are so wide you can only use them against one side and not against both lefties and righties.  I made this point on a mets board but people stated that all pens are like this.  I wonder if the mets relievers have a wider split then the typical bullpen.  Because while I believe Willie has mismanaged the pen, I think he’s been somewhat limited in what he could do as well.

John/13:

Re: the original topic. Just because Willie’s bullpen mismanagement cannot be measured in wins doesn’t mean it’s not real, and I don’t see why Sheehan’s claim should be ridiculed.

It is always the case that everything is real and does exist if it involves a human being.  This should go without saying.

What is always at stake is the degree to which something is real, and the degree to which we can measure it.

It looks to me that MGL is ridiculing Joe’s analysis (making claims without support).  This is what I call “yapping”.(*) Joe’s a terrific writer with a good baseball mind.  If he doesn’t have anything worthwhile to say, he can just skip posting one day.  He’s done alot (for him) of yapping this year. 

(*)I remember Will Carroll once saying he never gets writer’s block.  He doesn’t say that anymore and I think it’s because I told him it makes a difference between having something to say, and having something interesting to say.  Anyone can yap. If Stephen King says he gets writer’s block, that means we all get it.  I wound’t want to read a Stephen King book or short story where he yaps.  We shouldn’t be subjected to Joe Sheehan or MGL yapping either.  Ironically, I may be doing this right now.

I don’t know if it’s merely “yapping” to reiterate what is a common agreement among intelligent Mets fans, that Willie Randolph mismanages his bullpen. Just because the effect of the manager cannot be isolated from whatever over- or under-performance we would expect due to random variation or other measurable factors doesn’t mean that it’s not real, nor that Sheehan should be barred from talking about it.

I don’t even think it’s at all silly. It’s not the same as clutch, chemistry, and other descriptive and largely meaningless intangible “factors,” and I don’t know why we should treat it as such. Because it is tangible.

MGL’s hypothetical intangibles-free league is very illustrative and interesting. But let’s look at another hypothetical, just for the hell of it.

In this league, there is a team whose manager misuses his bullpen. He has six relief pitchers. Three of them are three of the best pitchers ever. The other three are terrible, terrible pitchers who don’t even belong in professional baseball, let alone the highest level of the sport. But our manager uses the three worst relief pitchers in the highest leverage situations, let’s say averaging an LI of 1.6. The three good ones publish a periodical in which they ridicule their manager and call him names, so the manager only uses them in the most meaningless of mopup situations. Let’s say they average an LI of .6.

Now tell me that the manager’s bullpen management is “intangible” and a worthless consideration.

It’s still just as immeasurable.

Im not sure anyone is questioning whether its real or not.  I think the question is to what extent is it real?  Willie mismanages the bullpen and thats costed us in terms of wins but by how much?

John/17:

I said this: “It is always the case that everything is real “

And you said this: “Just because the effect of the manager cannot be isolated ... doesn’t mean that it’s not real”

I am agreeing with you, as a given.  No one is arguing otherwise.  We’re on the same page.

And yes, it is yapping to reiterate the consensus while offering nothing new.  That’s the very definition of yapping!

“Because it is tangible. “

Yes, that’s the whole point!  It is tangible, so how about we produce tangible analysis?

Now tell me that the manager’s bullpen management is “intangible” and a worthless consideration.

It’s still just as immeasurable.

You are rebutting things that are simply not being asserted by mgl (nor me).  Of course it is tangible.  The issue is with Joe’s analysis.  Of course you can measure if Rivera, Joba get an LI of 0.6, and the schlubs are getting 1.6.  That is horrible.  That’s the point of my post #10.

Let’s presume for a moment, just for the sake of illustration, that Randolph’s bullpen management costs the Mets, say, 2 wins a season.

Even if we know precisely how many wins Randolph’s bullpen management cost the Mets, is that enough to say that he’s -2 wins compared to every other manager? Of course not.

What you’d have to do is:

1) Figure out how many wins every manager is costing/saving their team through bullpen management.
2) Make sure that you’re controlling for the quality of bullpen available.

Because Randolph isn’t operating in a vacuum here; you have to compare him to other managers before you make that sort of an evaluation.

And even then you still don’t know how good of a manager he is; managing a bullpen isn’t a manager’s only role. There’s lineups, pinch-hitters, defensive positioning, rotation management. Sure, you can quantify most of that (maybe all of it), but until you’ve gone ahead and done so, you’re just guessing, really. And the more informed you are, the more likely you are to be correct - I’ll take Chris Dial’s guess over, say, Steve Phillip’s guess every day over the week. But it’s far from being a certainty.

There is a question at BTF about what is “true talent”.  (*)

(*)Since this is a blog, we don’t go and putting a glossary every time we talk about “true talent” or “LI” or “WE” or anything else we talk about here.  If we were to write an article meant for broader exposure, we’d get into that.  Otherwise, this blog is meant as if you are walking into a conversation in the middle of it.  That is the essence of a blog.  It’s a series of journal entries (a log on the web… weblog… blog).

That said, here is my response, which I will also post to the wiki mailbag:

“True talent” is a player’s expected performance level for the rest of the season, given average context.

It is based on his career performance level after accounting for the context, weighting the most recent games the most.

As a result, “true talent” is our best estimate, an estimate which changes day-to-day (not unlike say the stock market, if interest rates were a constant), but it is fairly stable for a large portion of players.

A team’s “true talent” is based on the expected players and expected playing time for those players.  That of course, also can change.

As a proxy, it’s whatever TradeSports.com tells you it is, right now.

Yes, of course I was invoking the binomial distribution.  Every team’s win loss, and the win/loss distribution of all the teams in MLB, on the average, is guaranteed to be a normal curve with a binomial distribution centered around their true talent w/l percentage.  In 162 games with a static true talent level, one standard deviation for a .500 team is 6.35 wins.  Given that teams are generally anywhere from around .400 to .600 in true talent and given that their true talent fluctuates from day to day, I think it is fair to say that one standard deviation by chance alone is at least 7 wins per season.

So, yes it is guaranteed that on the average there will be 1.5 (5%) total teams per season that either under or over-perform by 14 or more wins, which is enormous of course.  Another 5% will either over or under-perform by by between 10.5 and 14 wins.

Dial needs to read a high school level statistics textbook, I guess, for the “support” that he wants for the above proposition.

So it is correct, as Tango likes to say, that even if there were managers who significantly influenced their team’s w/l record above or below their true talent levels (with “average” or zero interference from the manager), we would never, ever, ever be able to tell from the win/loss records of the teams as compared to what we “expect” from the team talent.  The random noise completely overwhelms any effect from managers (and any other tangible or intangible influence on a team’s performance, other than our estimate of their true talent).

We can more tell which managers are influencing their players and their team then we can tell which players are really clutch from their one year (or less) clutch/non-clutch performance, which batter/pitcher matchups are “truly” advantageous for one or the other, from one year or even several years of batter/pitcher data, etc.

Heck a manager who is bad is almost as likely to have his ream over-perform as under-perform.  If there are truly good and bad managers, say to the tune of 3 wins per season total (strategy, leadership, etc.), probably 40% of the bad ones are managing OVER performing teams and 40% of the good ones are managing UNDER performing teams.

Just can’t be done.

So let’s please stop pretending that FLO, HOU, STL, and WS managers, or even TB, are doing a great job this year, and Leyland, Yost, Girardi, Black, and Wedge are doing a poor job.  It just doesn’t work that way.  A lot of people will like to think it does, but it doesn’t.

Now, we can take a detailed look at everything a manager does and try and make some assessments of their strategic skill.  I don’t think you are going to find a whole lot of differences among managers.  As I said, maybe 1-3 wins a year, which is not wood of course.

And, let’s be clear about whether we are going to evaluate Willy or any other manager from a fan/psuedo-analysts/journalist perspective, or from a sabermetric one.

For example, let’s say that Heilman is their second best pitcher from a Marcel (projection) perspective and let’s say that there is no reason for the Mets or the fans to think that he is hurt.

And let’s say that Willy trots him out there in high leverage situations day in and day out and he keeps getting hammered.  And Willy keeps trotting him out there.

The fans, the pseudo-analysts, and the media will call for Willy’s head, and say (scream) that he is mismanaging his bullpen.  I would give him a raise.

Whenever I read about a manager being good or bad, half of it is from a “results” perspective, which is the typical fan/media one, and half (actually, a lot less than that) is from a sabermetric one, which is a “prospective” perspective, and not results oriented.

For example, a results oriented perspective would trot out the list of reliever and their respective average LI and then next to that, put their season to date ERA’s (or whatever stats you want) to justify or not-justify those LI’s.  A sabermetrician, coming from a “prospective” position, would put their average Marcel’s (projections) for this season next to those LI numbers in order to analyze the manager’s decisions so far.

I can give many examples of those, but let’s at least be clear about from what perspective we are evaluating a manager.  If it from a “results” perspective, I don’t want any part of it.

Here are the pitchers above along with their (my) current projections.  I’ll add .5 run to my NERC to make it look like an ERA, where the average pitcher has a 4.50 ERA.

1.45 Wagner 3.03
1.36 Sanchez 4.16
1.20 John Doe (I don’t have him in my database?)
1.18 Heilman 4.26
0.98 Schoenweis 4.81
0.95 Sosa 4.89
0.61 Feliciano 4.24

For a closer anything under 3.5 is good.  3 is about the best you can get.  For a setup guy, you want a 4 or less, around 3.8 or 3.9 is average.  low 4’s is decent for a reliever.  Mid 4’s is an average mid-reliever and high 4’s is basically close to replacement level for a reliever.

And of course, you want to maximize the platoon advantage as much as you can, and that can “screw up” the leverage numbers.

For example, you might bring in a mediocre lefty pitcher to face one or two lefty batters in a high leverage situation.  So it might look like you have a mediocre pitcher pitching in too high leverage situations, but once you factor in the platoon situation, that might not be the case.

Anyway, I stand by pretty much everything I said in my original post.

Tango/19:

I would add more, but I might just be “yapping.” After all, I know no more than anyone else about how to measure the effect of a manager on bullpen performance. I’ll just wait until someone smart figures it out, and then I’ll repeat it endlessly.

And let’s say that Willy trots him out there in high leverage situations day in and day out and he keeps getting hammered.

With the added provision (implied of course, but we’ll make it explicit here) that his forecast is constantly being updated.  At some point, if he keeps getting hammered, then Duaner or someone else will overtake Heilman.

***

“John Doe"… a little tongue-in-cheek.  That was supposed to be Joe Smith, which is as non-descript a name as you will find.  What kind of parents would call their kid Joseph, if the last name is Smith?  The whole point of a name is to uniquely identify someone.  I’d bet Joe Smith’s email address is something like . 

***

I know no more than anyone else about how to measure the effect of a manager on bullpen performance

If you were to write an article, and that would be the only words in that article, you would be lauded as a sabermetric visionary.

Quality trumps quantity.

Let us agree that Randolph’s management of his bullpen in 2007 was far from optimal. 

The general question of how one could quantify the effect of bullpen management decisions for a NL manager seems pretty difficult to me. You’ve got to take account of:

1. leverage vs. pitcher quality,
2. platoon effects,
3. effect of use/overuse/underuse on performance, and
4. interaction with pinch-hitter/bench availability.

I wonder if subjective, anecdotal accounts are strictly inadmissible at all times in the court of The Book blog.

Because sometimes I think they reflect the truth as much as, if not more than objective, statistical methods.

Of course, such was the thinking that kept baseball thinking in the dark ages for so long. But there is a danger in going too far the other way, and I think this blog entry represents that danger. Because there is no data and/or proof available, the account is labeled “ridiculous” and dismissed.

Much like the artificial stats vs. scouts opposition, both prescriptive and descriptive accounts are useful and meaningful. I don’t see why, if someone respectable offers that “Willie Randolph really blew that game by bringing in Sosa there,” we should dismiss him all of a sudden as a prattler who adds nothing to the conversation.

In fact, I find this tension far more interesting than the statistical and anecdotal accounts themselves. Perhaps in some way I find it an image of the world before and after The (so-called) Enlightenment. The world rejects the wisdom of the past, but that past was in some ways more beautiful and less petty.

Here is another example of results-based analysis of the best kind:

Last night in the Tigers/Angels game, Bonderman was pitching a (rare) good game, in that he had not allowed any runs thru 7.  In the bottom of the 8th, the first two batters got on and the next batter sac bunted.  Bonderman had only thrown 83 pitches.  Leyland took him out and Bonderman appeared to be miffed.

Perfect setup for a “results-based” post game or post inning analysis.  If the reliever gets out of the inning Leyland is a genius.  If he doesn’t, Letland looks like an ass with a quick hook for no good reason.

Whether that move is correct or not has NOTHING to do with how the inning or game turns out, of course. (Well, not quite “nothing” but close enough.)

Bonderman is not a very good pitcher to start with.  League average at best, probably worse.  After 83 pitches and the 4th time through the order, there are not going to be many relievers who are worse than that coming in a game.  So going to a reliever is absolutely the right move.  It almost always is in the 6th or later (and often earlier) with anyone but a very good or great starter on the mound.

The reason that managers can’t (and don’t) pull out starters early on a regular basis is that they have to pace their bullpen for the long haul (although other than your 2 or 3 best relievers, the rest are usually fungible withe pitchers in the minor leagues), they have to be sensitive to the “feelings” of the starter (you don’t want them to think you have no confidence in them), and to some extent you might need to “stretch out” your starters now and then, although I am not sure about that.

However, one of the most prolific mistakes that most managers make is keeping a mediocre or bad starter in the game too long when they are pitching “well” (I put that in quotes because, let’s face it, the manager’s definition of “well” is usually how many runs the starter gives up, not how well he is actually pitching).  The (incorrect) thinking is that they will continue to pitch well, at least until the don’t, and I don’t mean that facetiously.  ALL managers do that. 

Which is why #20 above is a good post. Chances are if you “dock” a manager 2 or 3 wins with a sabermetric analysis, most, if not all, managers will get the same docked 3 wins.  If you want to evaluate a manager sabermetrically, from the standpoint of whether he should be fired or not, you better do that for all the managers and certainly the one who will replace the one you want to fire.

#25 is a good post also.

And BTW, if you want to fire or just evaluate a manager for not having his relievers being used optimally, not only are you going to fire almost all the managers, but you are going to fire them based on literally costing his team 1-2 runs at this point in the season, which is .1 to .2 wins.  What do you think you would gain if you took Wagner’s 20 innings and gave him another .5 in LI and took that .5 away from someone else?  About 1 run per 9, which is 2.2 runs time .5, which is 1 run.  1 stinking run.

#26:

I don’t see why, if someone respectable offers that “Willie Randolph really blew that game by bringing in Sosa there,” we should dismiss him all of a sudden as a prattler who adds nothing to the conversation.

But that happens all the time here.  You can probably count on MGL making a “What was Manager X thinking?” post at least once a week. 

There’s a pretty big difference between discounting descriptive accounts that, when analyzed, turn out to be untrue (or perhaps, non-optimal) and discounting description as a means of presentation.  A lot (dare I say most) of my favorite analyses in recent times have begun with some variation of “X recalled *Insert Commonly Held Belief* and I decided to see if it actually holds water.” Sometimes it does; sometimes it doesn’t.  But without/until the analysis, even the true ones are hollow.  And that’s the problem that usually gets called out around here.

Meant to put this with my previous post but forgot.

Regarding Tango’s Joe Smith comment, I don’t know if I agree.  I mean, we’ve already seen the trouble with having multiple Dan’s among the relative handful of commenters here, so add to it that I myself am one of the perhaps millions of Dan Smith’s. 

But I actually enjoy it; as part of the generation that grew up on the internet, I can relate to the sometimes-overblown “Kids today post everything online and that’s bad!” stories, as it pertains to things like employers and other authority figures.  There’s something quietly comforting about knowing that, as far as Googling goes, I’m essentially anonymous.  Sure, the stuff that I’ve done that I’m proud of gets lost, and I don’t actually have anything remotely embarrassing or harmful to my character to be worried about, but still… in a generation whose personal photo albums are on the web instead of a bookshelf, I’m off the grid.

Though yes, it’s a bit disappointing to not be like everyone else and get a firstname.lastname@gmail account.  And no, no way I’m naming my kid Dan or John or Joe, but that’s more my own preference than any kind of “lesson learned.”

That’s probably a bit much of a response to Tango’s aside, but I finally have some kind of authority from which to speak.

1 stinking run.

That’s worth half-a-million dollars in New York. I wouldn’t bother with this as fan or even a consultant with some time on my hands. But every major league front office should be doing precisely this kind of analysis. It can pay for itself and then some.

John/26: what Dan/28 says.

What we’re calling out around here is statements that are unsupported.  If that’s what we wanted, we’d listen to WFAN.  Joe Sheehan is smarter than Mike and the Mad Dog.  That’s why we hold him to a higher standard.

As Dan said, we have no problem at all with descriptive qualitative statements.  But, they can’t just be any yapping b.s.  As Buzz knows all too well, there’s plenty of blogs to get that kind of insight.

It’s the analysis (or lack thereof) that we question.  Whether the statement being made is objective or subjective is irrelevant.

The short of it is: if you can hear it on WFAN, then we don’t want to hear it here.

Let it be clear that I am completely agnostic as to whether Willie (sorry about all those “Willy’s” - I knew they looked a little strange) should or should not be let go. 

If you are the GM or owner or other insider of the Mets, there are probably a million reasons why he should or should not be fired.

I also have no quarrel with Sheehan’s “analysis,” other than the comment I quoted, since I only skimmed the article.  I was merely making the point that it is virtually impossible for us to tell how much, if any, of a manager influences his team’s over or under-performing.

What is always amazing to me is how often a manager gets fired and still gets paid (IOW, they fire him before his contract runs out).  I mean, how tough is it to figure out if the guy you hire for a million plus dollars a year is someone that is capable of doing the job your hire him for?

Can you imagine many other industries where someone gets an employment contract for several million dollars and regularly gets fired and paid to boot.

tango,
I appreciate your concerns of critics at BTF, but:
“Dial needs to read a high school level statistics textbook, I guess, for the “support” that he wants for the above proposition.”

isn’t going to garner you much support. 

The answer to my question that it is guaranteed - it’s that the math indicates it should, given a large enough sample (and I’m thinking in terms of genetics).  Given two 1st generation with XX and XY, and generating a Punnett Square, you get two offspring that are XX and two that are XY.  However, there are plenty of instances where, in four offspring, all are XY (or XX).  Just because the math says something will happen, doesn’t guarantee it will happen.

"Can you imagine many other industries where someone gets an employment contract for several million dollars and regularly gets fired and paid to boot. “

Uh, lots of CEOs?

Chris/33: that wasn’t me.

I think it’s irrelevant how one manager compares to the pool of managers. If the manager is doing demonstrably stupid things in comparison to what projections suggest, he should either get with the program, or get fired.

I know you don’t like the beer analogy because you think the impact would be obvious (I honestly have no idea how much of an impact 3 beers in the morning would have versus 2 beers at lunch and 1 at 3pm on widget-making, let alone the impact that 3 beers would have on a commuting population of employees as far as accidents over the norm for a control group or the 2 beers with lunch/1 at 3pm group). Assuming that the impact is marginal but measurable (much like how a manager uses his bullpen), I would feel justified in firing the manager for that choice even if the other managers are just as bad.

After all, it’s really really easy for the manager to stop doing something negative, just as it’s easy for managers to use their brains picking out pitchers in high or low-leverage situations in a far more optimal way. And there’s no shortage of brainpower in ballclubs who could figure out the optimal use if given enough time and data.

My point is that even if sabermetrics can’t figure out with a high-degree of certainty whether a manager should be fired or not, we certainly do have the ability to see if managers are regularly making poor decisions or not.

And many of those poor decisions are posted on this blog by tango and MGL.

I fail to see why regular poor decision-making by someone paid a 7-figure salary is not grounds for dismissal when we can measure and analyze that negative impact. In reality, if we don’t fire someone we know is having a negative impact because overall the impact may be possible and we just can’t measure it, we’re saying that we trust that the intangibles are having a bigger impact than the tangible negatives.

Can you really justify not firing a manager in a situation where the bullpen use is costing 3 wins a year?

Like I said:

The only way to do this is to actually sit down and track every single decision he makes, and compare those decisions to every other manager.

(Andy for example did a great job in tracking pinch hitting by managers.)

Yes, so let’s do that, and when we find out which managers are making consistently poor decisions, we should suggest that they be fired for someone less incompetent.

If we find that all managers are equally incompetent, then we should suggest that they all be fired and replaced with people who can follow simple rules to manage effectively.

What’s so controversial about that?

Hmmm… re-reading this thread in a slightly different light, I have to make a point, so as to not appear hypocritical.  If there’s one thing I can’t stand is a hypocrite, and I want to be able to stand myself.  WYSIWYG with me.

MGL takes alot of grief at BTF (even when he’s not participating in the discussion).  I’ve posted to people there (in 2, maybe 3 threads) that it is simply unbecoming of intelligent people to take the constant (unsupported) potshot drive-bys of other posters (or potential posters), that if they need clarification that MGL has a blog right here where he can answer your questions.

In post 12, I highlighted Chris’ post, to which I replied that MGL must have meant the binomial.  While it was clear to me that’s what MGL was talking about, it may not have been clear to Chris (or for that matter, hundreds or thousands of readers).

Chris actually came to this blog, so he is (was?) present to engage in a discussion.

MGL/22 gave an excellent post, but intermingled was a one-line potshot at Chris, the very thing that I was railing the BTF posters against.  I missed it, or perhaps even turned a knowing blind eye.  That was simply unfair on my part.  I’m sure most of us probably missed it, but that doesn’t mean everyone did, as Chris/33 rightfully points outs (in a post directed at me) that I may be correct in my view of some of the BTF posters, but MGL’s response basically undermines my entire premise: that we can engage in discussions without making it personal, if I invite posters here.

My position is fairly simple: we’re here to engage in discussions where we can move forward.  We educate where we can, and listen when we can’t.  We give the benefit of the doubt as much as possible.  We don’t make it personal with other posters and we don’t do drive-bys.

I apologize for not having met my duty here.

Consider this post an interlude, and not one that requires a response, as I don’t want this thread derailed, but it needed to be said here.

I apologize for my “pot-shot.” I don’t think that Chris thought that was Tango who wrote it.  He somehow (and for some reason) is holding Tango responsible for that.  I think.  Which is ridiculous.  Tango, there is no need for you to apologize.  You did nothing wrong.

Anyway, nuff said about that.  There is no reason to belabor this issue.  As Tango said, this blog/forum is about intelligently moving forward with regard to interesting issues, and shall stay that way.  We want Buzz to think highly of it!

Sal, we are starting to talk past one another, or you are arguing with a straw man, or something like that.  No one said that Randolph should or should not be fired, nor has anyone even given much support for the proposition that Willie has been demonstrably bad.

That being said, all employees in the course of their employment do some bad things.  Whether and when you fire them or not is another story.

And it is relevant the pool of people you can reasonably and practically get to replace a manager whom you fire.  Your blanket proposition that if a manager does not do everything optimally you fire him and replace him with someone who does is a little (maybe a lot) like saying that you have to get rid of every player who does not have an OPS of at least .800 because they are playing sub-optimally.  Who are you going to replace them with?

Now, THAT being said, if I were running a team, yes, I would NOT tolerate a manager doing much of the things that I think are suboptimal and I would make sure that I hired a manager who would listen and then use the things I tell him to use, and if it turns out he doesn’t, then yes, he would be fired.  No teams, even the so-called sabermetric ones, appear to be following that model.  Whether that is right or wrong, I don’t know.

And there are still people inside and outside of the industry that would dispute a lot of the sabermetric propositions that dictate whether a manager is doing a good job or not.

I would want him to use the best reliever for 90 innings a year and in many non-save situations and not in many save situations, but that “opinion” is by no means universally agreed with even by smart people.  I would want my starting pitcher who is not an ace, to virtually never bat, even if that means coming out of the game in the 5th inning when he has not given up a run, but that opinion is also not universally agreed with.

Etc.

So there is the issue of, “From whose perspective are we evaluating a manager and what level of sub-optimal behavior are we using as a requisite for firing him?”

In any case, all I said that was Sheehan offered little support for the proposition that Willie is a good or bad manager, and that neither he nor anyone else outside of the organization knows whether he should be fired or not.  If it were not for the fact that the Mets have an under-performing record (and not even THAT under-performing for that matter, as compared to, say, SD), it would not even be an issue.  The other point I was making was that even if managers had significant influence on their team records below or above the team’s true talent, you could NOT tell from the team w/l records (other than in the very, very, very long run).

Actually, he understood that you said it.  But, he was pointing out to me that basically I took a firm ground at BTF against potshots, and here it is happening in my own backyard.

His post 33 was initially confusing to me, which is why I made post 35.  Anyway, hopefully, lesson learned.

Yes, the main point that is being made is how wholly unreliable a team’s W/L record, when trying to evaluate a manager.  The W/L record is a byproduct of many things, not the least of which is random variation.  The manager’s influence, relatively speaking is tiny.  One guy having a legitimately bad season (playing through injuries let’s say) can undo completely a manager’s tactics over 162 games.  Imagine a lawyer working on a case for 6 months, and his spouse throwing his files in the garbage the next day.  You lose the case, but it’s not the lawyer that really did anything wrong.  But, you get a tally in the L column for that.

Donald Trump subscribes to this ludicrous theory in The Apprentice.  (Or at least, he says he does while filming the show.)

It’s funny how the one group of people who believes in luck as much as some of us are the players themselves.  The won’t use “luck”.  They’ll take about the breaks, the bad bounces, the crossbar, game of inches, the anything can happen, etc, etc.  All in deference to things that are either outside their control, or subject to such fortuitous timing as to be basically be outside their control. 

The group that hates the talk of luck are the fans.  They simply see W/L as something real, and they live and die on the W/L.  They don’t want to spend the time to analyze things, so they simply quote the W/L record.

Not talking past each other, I was just confused. You clarified it with this statement:

if I were running a team, yes, I would NOT tolerate a manager doing much of the things that I think are suboptimal and I would make sure that I hired a manager who would listen and then use the things I tell him to use, and if it turns out he doesn’t, then yes, he would be fired.

In my mind, when you said:

you will never catch me writing an article trying to figure out if some manager should be fired or not.

Yes, I realize there is a difference between writing about it and believing it, as well as ‘should’ and ‘I would’ but…

And just for clarification, I do not believe a manager needs to be optimal, just not incredibly sub-optimal. For instance, if you have a pitcher with a true talent ERA (or whatever more sabermetrically-oriented stat you have) of 3.0 and another one of 3.2, and he puts the 3.2 player into the higher leverage situation 1 out of 10 times, I don’t see too much of a problem. But when he’s eschewing the 3.0 guy for a 4.5 guy in a close game, that’s when he deserves to be fired in my mind.

Also, I am of the opinion (though obviously it is nigh-impossible to quantify, and therefore it probably isn’t best to make decisions on) that a manager has an absurdly minor positive impact on players, and that it’s more the extent to which he has a negative impact. As a result, I think someone with good people skills and nice note cards telling them who to switch in/out and when would be just as fine as a manager as anyone in the game today.

Players who can OPS .800, however, are a far rarer breed, and that prevents you from just dumping them by the wayside. I see no reason you couldn’t find some likeable third base coach from your minor league system every year to promote and see if he works. If he does, great! If he doesn’t, switch him for someone else.

How about a fans’ manager rating like the fans’ fielding ratings?

That wouldn’t be a bad idea actually.  We can come up with say a good 20 or 30 questions of various things that a fan perceives of a manager (that we can’t figure out from the data easily enough).

In fact, we can also seed the questions with a couple of easily verifiable questions to weed out the yappers from the thinkers.

If a couple of you think it’s a good idea, then start posting some questions that we can ask the fans about their manager.  Let’s see how far we can get…

Tango, I forgot about this until now. I see if I can come up with some questions.