Buzz Bussinger Is A Know-Nothing...

[Mike Donlin]

...idiot and Tony Larussa is the most overrated manager in professional baseball.

Or so I have concluded from reading THREE NIGHTS IN AUGUST.

Larussa needs to learn about sample sizes. He bases nearly all of his decisions on how a player or pitcher has performed against another player or pitcher in the past. So if someone is 4 for 8 against a given pitcher, he must "own" him; similarly, if a left-handed pitcher limits a right-handed batter to, say, 2 for 8, then he (the LHP) must be one of what Larussa calls those "reverse" pitchers who do better versus righties than lefties!

Bussinger, however, is a slightly different breed of idiot. He bases all of his conclusions about sabermetrics on MONEYBALL and, as a result, does not understand it in the least. Moreover, his charge that sabermetrics robs the game of its emotional value is nostalgia and sentimentalizing of the worst sort.

Sorry. I had to vent. This book ruined my vacation.

[mikev]

You should have read Ball Four instead.

[His Own Bad Self]

Originally Posted by Mike Donlin

...his charge that sabermetrics robs the game of its emotional value is nostalgia and sentimentalizing of the worst sort.

As quoted by the good folks at firejoemorgan:

From the preface to "Three Nights in August":

"In this new wave of baseball, managers are less 'managers' than 'middle managers,' functionaries whose strategic options during a game require muzzlement, there only to effect the marching orders coldly calculated and passed down by upper management. It is wrong to say that the new breed doesn't care abut baseball. But it's not wrong to say that there is no way they could possibly 'love' it, and so much of baseball is about love. They don't have the sense of history, which to the thirtysomethings is largely bunk. They don't have the bus trips or the plane trips. They don't carry along the tradition, because they couldn't care less about the tradition. They have no use for the lore of the game -- the poetry of its stories -- because it can't be broken down and crunched into a computer. Just as they have no interest in the human ingredients that make a player a player and make a game a game: heart, desire, passion, reactions to pressure. After all, these are emotions, and what point are emotions if they can't be quantified?"

End quote.

I like that Bissinger can discern the thoughts and feelings of thousands of people who work in or follow baseball and "buy into" sabermetrics. I wish I were that smrt.

And since Billy Beane played in the majors, doesn't he "have the bus trips and the plane trips"? I'm confused.

[Mike Donlin]

Right. Nor do sabermetricians have the "sense of history." I wonder what they're doing, then, when they pore over baseball history in an effort to understand that moment and our own baseball moment.

About the only excitement this book contained is trying to figure out who to hate more: Larussa or his Boswell. One goes back and forth, but in the end it has to be Bussinger. Larussa seems naive and may actually do what a decent job managing his players, which is about all a manager can do, really. But Bussinger is a thoroughgoing fool.

And, alas, I've already read Ball Four.

[Al Golagnic]

I agree completely. This book taught me that LaRussa is a schmuck. At least the half of it that I read before it became such a chore that I could not go on.

[abailey3313]

I've called coaches and team personnel idiots before too -- but those idiots were guys like Dave Wannstedt.

He might be overrated, but I'm not sure how much credibility you have telling a former World Series champion and one of the winningest managers in history how to read into numbers.

[Mike Donlin]

Originally Posted by abailey3313

I've called coaches and team personnel idiots before too -- but those idiots were guys like Dave Wannstedt.

He might be overrated, but I'm not sure how much credibility you have telling a former World Series champion and one of the winningest managers in history how to read into numbers.

I would say quite a lot of credibility. I may have never managed a baseball game in my life, but I know that 8 at bats is not enough to draw any meaningful conclusions from. It's just so strange that Larussa puts all this emphasis on numbers, on match-ups, but has never stopped to consider that the numbers may not mean anything at all.

Now, Larussa's pitching coach, Dave Duncan, actually seems very smart and very shrewd about tracking pitches, locations, first-pitch strikes, etc.

[abailey3313]

Originally Posted by Mike Donlin

I would say quite a lot of credibility.

I wouldn't.

Originally Posted by Mike Donlin

I may have never managed a baseball game in my life, but I know that 8 at bats is not enough to draw any meaningful conclusions from. It's just so strange that Larussa puts all this emphasis on numbers, on match-ups, but has never stopped to consider that the numbers may not mean anything at all.

Why isn't eight at-bats enough to draw conclusions from? If Batter A has only faced Pitcher A eight times in his career and is 0/8 against him, should we be looking at something else to decide whether or not Batter A should be pinch hit for? Same thing if the batter is 8/8. If a player is 0/1 or 0/2 against someone, yeah, the sample size needs to be expanded. But at some point the numbers gain meaning.

And since when do "numbers mean nothing at all"? That's preposterous.

[Mike Donlin]

Originally Posted by abailey3313

I wouldn't.


Why isn't eight at-bats enough to draw conclusions from? If Batter A has only faced Pitcher A eight times in his career and is 0/8 against him, should we be looking at something else to decide whether or not Batter A should be pinch hit for? Same thing if the batter is 8/8. If a player is 0/1 or 0/2 against someone, yeah, the sample size needs to be expanded. But at some point the numbers gain meaning.

And since when do "numbers mean nothing at all"? That's preposterous.

I had Excel produce a random number between 0 and 1. A number .265 or lower indicated a hit. A number above .265 indicated an out. Each batter got eight at bats. I ran ten such groups.

The batter went 2 for 8 (.250) twice. 3 times he went 3 for 8 (.375). 3 times he also went 1 for 8 (.125). Once he went 0 for 8 (.000). And another time he went 4 for 8 (.500)!

Larussa looks at the outliers and assumes that they represent something meaningful about that player's performance. He will pinch hit the .375 batter over the .125 one. But it's the same batter. It's chance.

Now, it may not be completely chance. No player is a number like .265. He may indeed perform well against certain pitchers, certain pitch-types, or he may have improved his timing by taking extra batting practice swings. But Larussa goes purely on the numbers, and they do not give you any basis whatsoever to judge performance by. Even overall, after 80 at bats, our hitter was hitting .250--close but significantly lower than his "true" ability of .265. That gets us closer to a meaningful sample size, but it's rare that a player faces the same pitcher 80 times in his career. Even so, Larussa would conclude that the .250 hitter against pitcher X was good but not great when what he really is is just what he is.

Edit: This is after ten samples. If we did this thousands of times, doubtless we would have players hitting well over .500, even one or two 1.000s. But it's still the same batter. That's what I mean when I say the numbers may not mean anything at all.

[abailey3313]

Originally Posted by Mike Donlin

I had Excel produce a random number between 0 and 1. A number .265 or lower indicated a hit. A number above .265 indicated an out. Each batter got eight at bats. I ran ten such groups.

The batter went 2 for 8 (.250) twice. 3 times he went 3 for 8 (.375). 3 times he also went 1 for 8 (.125). Once he went 0 for 8 (.000). And another time he went 4 for 8 (.500)!

I fail to see how those tests have anything to do with the topic. You can roll a six-sided dice 100 times and report the results, but that doesn't accurately reflect anything close to a batter versus a pitcher.

Originally Posted by Mike Donlin

Larussa looks at the outliers and assumes that they represent something meaningful about that player's performance. He will pinch hit the .375 batter over the .125 one. But it's the same batter. It's chance.

That's illogical.

If it really is all "chance", then what difference does it make if he's looking at numbers or not? I'd rather rely on misleading numbers -- numbers that have been formed some way, somehow -- rather than blind chance.

Originally Posted by Mike Donlin

Now, it may not be completely chance. No player is a number like .265. He may indeed perform well against certain pitchers, certain pitch-types, or he may have improved his timing by taking extra batting practice swings.

Of course it's not completely chance. If it were, Torii Hunter probably wouldn't be a lifetime 0/26 (or whatever it is) against Roger Clemens.

Originally Posted by Mike Donlin

But Larussa goes purely on the numbers, and they do not give you any basis whatsoever to judge performance by. Even overall, after 80 at bats, our hitter was hitting .250--close but significantly lower than his "true" ability of .265. That gets us closer to a meaningful sample size, but it's rare that a player faces the same pitcher 80 times in his career. Even so, Larussa would conclude that the .250 hitter against pitcher X was good but not great when what he really is is just what he is.

I still fail to see what's so wrong about using numbers that have been compiled for this exact reason.

If using a batter's lifetime average against a pitcher is incorrect, what do you suggest using?

[mlyons]

The folks who wrote The Book: Playing the Percentages in Baseball did a study on this topic. They looked at every player who had at least seventeen plate appearances against a certain pitcher from 1999-2001 and at least nine plate appearances afterwards. They then compared those players' performances in the "before" plate appearances to their performances in the "after" plate appearances and found no correlation at all. Zero. As in, those previous plate appearances had absolutely no predictive value as to what the hitter would do against that pitcher in subsequent matchups. Keep in mind, these are what real hitters actually did against real pitchers, not a random number generator.

[Mike Donlin]

Originally Posted by mlyons

The folks who wrote The Book: Playing the Percentages in Baseball did a study on this topic. They looked at every player who had at least seventeen plate appearances against a certain pitcher from 1999-2001 and at least nine plate appearances afterwards. They then compared those players' performances in the "before" plate appearances to their performances in the "after" plate appearances and found no correlation at all. Zero. As in, those previous plate appearances had absolutely no predictive value as to what the hitter would do against that pitcher in subsequent matchups. Keep in mind, these are what real hitters actually did against real pitchers, not a random number generator.

I guess that's what I mean by chance. The best predictor of a given player's performance is that player's true ability--against all pitchers, not just an individual pitcher.

That--a player's true ability, to the degree that that can be determined--is what I would suggest using to determine decisions.