Baseball for the Thinking Fan

Login | Register | Feedback

btf_logo
You are here > Home > Baseball Newsstand > Baseball Primer Newsblog > Discussion
Baseball Primer Newsblog
— The Best News Links from the Baseball Newsstand

Thursday, June 01, 2006

Daily News: Kram: Just for Chuckles, we ask computer for its Phils lineup

Charlie Manuel vs Mark Pankin’s “sophisticated computer model”.....You’ll scream yourself into a state of shock!

Manuel says he is always scribbling out alternative lineups, especially if he plans to work in some bench players on a given day. He had hoped to do that Sunday, but the team had been losing so he stayed with his usual lineup. When he does juggle it, he says he looks at a variety of statistical information, which he keeps in a closet above his desk; he says he has done that for years. He also consults with his coaches for their opinion if a player happens to be going poorly. But he emphasizes that he does not take into consideration the opinion of fans or others.

Repoz Posted: June 01, 2006 at 11:20 AM | 73 comment(s) Login to Bookmark
  Tags: phillies

Reader Comments and Retorts

Go to end of page

Statements posted here are those of our readers and do not represent the BaseballThinkFactory. Names are provided by the poster and are not verified. We ask that posters follow our submission policy. Please report any inappropriate comments.

   1. Andrew Edwards Posted: June 01, 2006 at 12:28 PM (#2046958)
It's like every 'stats' debate ever, rolled into one article.

In one corner:

"Burrell has a lot of power, so we think of him as someone who should be hitting fourth or third," says Pankin, who received his doctorate in mathematics from the University of Illinois-Chicago. "But Burrell has a very high on-base percentage, which is why the model wanted him higher in the order. If you can prolong innings, you can increase your chance of scoring. So you want players who get on base at the top of the lineup."

And in the other:

That is why even if Einstein himself appeared as an apparition and told him to do it, Manuel would not bat Abreu in the leadoff spot. Even when Rollins is struggling, Manuel says he loves "the electricity" that the switch-hitting shortstop can create.... He likes batting Abreu in the three hole because, while he has an exceptional on-base percentage, he has proved he can drive in runs. Working on the assumption that one or possibly two of the top three batters gets on, Manuel says he then has some "good runners" on the bases for the power-hitting Burrell, which he says helps to prevent the bases from becoming clogged.
   2. Weekly Journalist_ Posted: June 01, 2006 at 01:13 PM (#2046983)
which he says helps to prevent the bases from becoming clogged.

Yes. Getting outs helps keep the bases from getting "clogged."

Aren't "clogged" bases what you want as much as possible?
   3. Crispix reaches boiling point with lackluster play Posted: June 01, 2006 at 01:52 PM (#2047012)
Rollins is a really really really bad leadoff hitter.

Why? He doesn't get on base, and he doesn't take pitches.

Isn't that all we need to know?

Meanwhile, Abreu has the highest OBP and most walks in the league.
   4. Crispix reaches boiling point with lackluster play Posted: June 01, 2006 at 01:56 PM (#2047016)
Oh, one more thing; after playing the Expos this week, I now know that it is completely acceptable for a team to have its best power hitter playing leadoff (Soriano), instead of being obsessed with having your power hitters get the maximum possible number of RBIs.

Abreu isn't even our best power hitter, and his OBP is about 140 points higher than Soriano's at this point. Fans are mystified by why having Abreu lead off is out of the question.
   5. Dan Szymborski Posted: June 01, 2006 at 02:05 PM (#2047022)
That is why even if Einstein himself appeared as an apparition and told him to do it, Manuel would not bat Abreu in the leadoff spot.

This is just silly. If Einstein's actual ghost appears before you, you do whatever the hell it asks you to.
   6. Crispix reaches boiling point with lackluster play Posted: June 01, 2006 at 02:12 PM (#2047027)
I didn't even notice this:

Even when Rollins is struggling, Manuel says he loves "the electricity" that the switch-hitting shortstop can create...

When Rollins is struggling, he produces less electricity than any player on the team except Sal Fasano. He swings at every pitch and usually grounds out on the second pitch.

If Manuel likes "electricity", he should teach Rollins the art of the surprise drag bunt or something, or alternately, how to hit the ball far enough that his speed might enable him to get on base. Or maybe switching from LH to RH in the middle of an at-bat, that might throw the pitcher off.
   7. Crispix reaches boiling point with lackluster play Posted: June 01, 2006 at 02:25 PM (#2047039)
I've been very hard on Rollins in this thread, but it applies to "when Rollins is struggling", which is not all the time. He seems to be a very streaky player. I wouldn't be surprised if he has another 30-game hitting streak in his career. But there are long slumps too, during which Abreu (or Victorino? Burrell? Never thought of Burrell) should lead off.
   8. Brewers Now and 4ever Posted: June 01, 2006 at 03:16 PM (#2047093)
I play in a Strat-o-matic league and have Burrell hit leadoff against lefties because his OBP is so high. He's cracked a few homers from the leadoff spot, but he hasn't "cost" me any RBIs. If I had Abreu, there is NO question he would be leadoff on my team. Against righties, Burrell is usually sixth because I also have A. Jones, C. Jones. Guerrero and Teixeira.
   9. mgl Posted: June 01, 2006 at 06:04 PM (#2047228)
"Some people ask, 'Why not bat Howard fourth?' " Manuel says. "I bat Pat there because it splits up our lefties. Because if I hit Howard fourth, the likelihood is that later in the game, [the other team] would bring in a lefthanded pitcher to face our lefthanded hitters."

This quote and the article in general are ironic. I have touted Manuel as one of the worst tactical managers I have ever seen. And yesterday he was asking for trouble by batting 3 lefties in a row. If he is aware of that "problem," why did he do that? Being able to bring in a lefty reliever against 3 non-pinch hittable batters in a high leverage situation is a huge advantage for an opposing manager. A team should NEVER have 3 lefties in a row in a lineup. Never! If possible, they should not have two.

I am not that familiar with Rollins, but yesterday, with runners at 1st and 2nd (or 1st and 3rd) and 2 outs, he had a horrible at bat. With a 3-0 count, he took a cookie right down the middle (you don't take automatically on a 3-0 count with 2 outs and runners in scoring position!), and then swung at a bunch of borderline pitches after that (lousy swings I may add). Actually it is amazing how manny batter take 3-0 pitches with runners in scoring position. This is a pet peve of mine (along with swinging at bad or borderline pitches with a 3-2 count). That is clearly the wrong strategy for almost any batter...
   10. mgl Posted: June 01, 2006 at 06:05 PM (#2047229)
"Some people ask, 'Why not bat Howard fourth?' " Manuel says. "I bat Pat there because it splits up our lefties. Because if I hit Howard fourth, the likelihood is that later in the game, [the other team] would bring in a lefthanded pitcher to face our lefthanded hitters."

This quote and the article in general are ironic. I have touted Manuel as one of the worst tactical managers I have ever seen. And yesterday he was asking for trouble by batting 3 lefties in a row. If he is aware of that "problem," why did he do that? Being able to bring in a lefty reliever against 3 non-pinch hittable batters in a high leverage situation is a huge advantage for an opposing manager. A team should NEVER have 3 lefties in a row in a lineup. Never! If possible, they should not have two.

I am not that familiar with Rollins, but yesterday, with runners at 1st and 2nd (or 1st and 3rd) and 2 outs, he had a horrible at bat. With a 3-0 count, he took a cookie right down the middle (you don't take automatically on a 3-0 count with 2 outs and runners in scoring position!), and then swung at a bunch of borderline pitches after that (lousy swings I may add). Actually it is amazing how many batters take 3-0 pitches with runners in scoring position. This is a pet peve of mine (along with swinging at bad or borderline pitches with a 3-2 count). That is clearly the wrong strategy for almost any batter...
   11. jmac66 Posted: June 01, 2006 at 06:05 PM (#2047230)
Manuel says he loves "the electricity" that the switch-hitting shortstop can create..

that's becuz the ghost of Benjamin Franklin appeared before him
   12. Crispix reaches boiling point with lackluster play Posted: June 01, 2006 at 06:15 PM (#2047236)
But they always bat three lefties in a row. I'm not going to look at every box score, but every game I've seen this year has been

1. Rollins(switch) - or Victorino a couple times
2. Utley(left)
3. Abreu(left)
4. Howard(left)
5. Burrell(right)
6. Rowand(right)/Victorino(switch) - or Rollins a couple times
7. Bell(right)/Nunez(switch)
8. Fasano(right)/Lieberthal(right)/Ruiz(right)
9. pitcher

That's right, no matter who the CF is he bats 6th; no matter who the 3B is he bats 7th; and no matter who the C is he bats 8th.
   13. Greg Maddux School of Reflexive Profanity Posted: June 01, 2006 at 06:29 PM (#2047249)
So, MGL, what would the optimal, Book-approved lineup be? Abreu, Burrell, Howard, Rowand and Utley followed by the dreck?
   14. Spike Owen 10/2/1987 Posted: June 01, 2006 at 07:40 PM (#2047313)
All these comments attacking Manuel for his lineups are just wrong.

You should be attacking him for his complete mishandling of lefty v. righty matchups when running his bullpen. He seems unable to anticipate the opposing manager just might have good pinch-hitters on his bench.
   15. Crispix reaches boiling point with lackluster play Posted: June 01, 2006 at 08:22 PM (#2047374)
He seems unable to anticipate the opposing manager just might have good pinch-hitters on his bench.

That may be because he himself sees no reason to have good pinch-hitters on his bench. Management gave him Dellucci just before the season began, but TEWAN (the even worse Abe Nunez) is more often his pinch-hitting pick, for unknown reasons.

Another bullpen problem is running through relievers too fast. If a reliever gets through an innning in 7 pitches, he's no more likely to go out for the next inning than if he had taken 30 pitches. Then you end up with Ryan Madson pitching eight innings of relief.
   16. Misirlou is on hiding to nowhere Posted: June 01, 2006 at 08:24 PM (#2047378)
which he says helps to prevent the bases from becoming clogged.

God! I am getting to the point where I absolutely hate that phrase. I understand the sentiment; you would rather have a fast guy on base than a slow one. Who wouldn't? But how many times per year would you rather have 1 out and no one on base rather than no outs an a slow man on first? Or one out and a fast man on first rather than no outs and a slow man on second and a fast man on first? 10? 5? Fewer? I'm not saying that baserunning is necessarily underrated, but that OBP, whatever its form, is underrated.

People will point to "the flip' as a counter-example. But if a faster player than Giambi was (not) on base, the play never happens in the first place. And 99 times out of 100, Giambi scores.

Look, I loved the 80's brand of baseball, and even though I am a Cub fan, I can appreciate the Herzog Cardinals. What many people fail to realize is that for all the SB hullabaloo surrounding them, the three pennant winners all led the league in OBP.

Not that they had any base cloggers except for Clark.
   17. Crispix reaches boiling point with lackluster play Posted: June 01, 2006 at 08:26 PM (#2047381)
Yes MGL! What should the lineup be?

My first instinct would be to leave things the same except:
A) Abreu leads off.
B) David Bell hits 3rd (why not, he's not that bad. and it breaks up the lefties)
C) Rollins hits 7th. Or 6th, and then Rowand at 7th.
   18. The Keith Law Blog Blah Blah (battlekow) Posted: June 01, 2006 at 09:19 PM (#2047432)
But they always bat three lefties in a row. I'm not going to look at every box score, but every game I've seen this year has been

1. Rollins(switch) - or Victorino a couple times
2. Utley(left)
3. Abreu(left)
4. Howard(left)
5. Burrell(right)
6. Rowand(right)/Victorino(switch) - or Rollins a couple times
7. Bell(right)/Nunez(switch)
8. Fasano(right)/Lieberthal(right)/Ruiz(right)
9. pitcher


Burrell batted fourth in five of the six games they played against the Brewers; he didn't play in the other one.
   19. Crispix reaches boiling point with lackluster play Posted: June 01, 2006 at 09:25 PM (#2047436)
You're right; he's always 4th and Howard is always 5th.
   20. mgl Posted: June 01, 2006 at 11:32 PM (#2047597)
Yes, when he plays, Burrell generally splits up the 3 lefties. I don't know what the optimal lineup is according to "The Book." It takes some time to figure that.

If I plug some lineups into my sim (which is actually better than using the methodologies in "The Book"), here is what I get:

Versus Lefties:

Rollins
Utley
Abreu
Burrell
Howard
Rowand
Bell
Ruiz

5.523 rpg

Even having 3 of 4 lefties in a row is dangerous.

If we alternate RH and LH batters, which is quite easy versus a LHP, as almost all the RHB are going to be better than the LHB, we can put up this linuep:

Rollins
Abreu
Burrell
Howard
Rowand
Utley
Bell
Ruiz

That produces 5.502 rpg, about 1 run a year worse than the other one. How hard can it be for Manuel to figure out a reasonable lineup where his RH and LF batters are alternated versus a lefty starter. He is truly a moron.

If we switch Rollins and Rowand above, we get 5.520 rpg a little better (essentially the same given the sample sixe of games I am simming).

If we switch Bell and Rollins (Bell 5th and Rollins 7th), we get 5.471 rpg. If we bat Rollins 6th and Utley 7th (again, versus the lefty pitcher), we get 5.455 rpg.

Let's take Manuel's silly lineup the other day versus Livan, with the 3 lefties in a row:

Rollins
Utley
Abreu
Howard
Rowand
Dellucci
Nunez
Fasano

We get 5.950 rpg

With this:

Abreu
Rollins
Howard
Rowand
Utley
Dellucci
Nunez
Fasano

We get 5.911 runs per game

If we switch Howard and Utley, we get 5.900

If we use this:

Abreu
Rowand
Howard
Rollins
Utley
Dellucci
Nunez
Fasano

We get 5.952, which appears to be the best lineup yet without having two lefties batting together other than Dellucci and Utley, but Burrell would pinch hit for Dellucci anyway versus a LHP (although who knows with Manuel a truly brain dead manager).

Let's take their traditional healthy lineup

Rollins
Utley
Abreu
Burrell
Howard
Rowand
Bell
Liberthal

We get 5.798 runs per game

If we use:

Utley
Abreu
Burrell
Howard
Rowand
Rollins
Bell
Liberthal

We get 5.799, essentially the same.

If we try and alternate RH and LH batters (I don't like even 2 lefties in a row), we can do:

Abreu
Rowand
Utley
Burrell
Howard
Rollins
Bell
Liberthal

We get 5.737 rpg.

Switching Rollins and Rowand, we get 5.795. This may be the optimal order without 2 lefties in a row.

If anyone has any other suggestions, post them and I will run it.
   21. Chris Dial Posted: June 02, 2006 at 12:17 AM (#2047656)
Let's take Manuel's silly lineup the other day versus Livan, with the 3 lefties in a row:
...We get 5.950 rpg...

(other lineups)
5.911 runs per game
5.900
5.952
5.798
5.737
5.795


Most lineups worse - some significantly. You can manage to gain two one thousandths of a run.

Yes, that Manuel lineup was downright silly.
   22. mgl Posted: June 02, 2006 at 12:59 AM (#2047704)
Why don't you drop in and troll Chris. The "silly" refers to the 3 lefties in a row. It has nothing to do with the rpg. Go back to your cave.
   23. GGC don't think it can get longer than a novella Posted: June 02, 2006 at 01:10 AM (#2047716)
If I plug some lineups into my sim (which is actually better than using the methodologies in "The Book"), here is what I get:

Honest question: if you run the simmed lineupp with three consecutive lefties, does the program take this into account and throw more lefty pitvhers at you?
   24. Francoeur Sans Gages (AlouGoodbye) Posted: June 02, 2006 at 01:14 AM (#2047721)
Rather than simply looking at mean rpg, I'd be interested in knowing the variance, and other measures of spread. The Phillies are (meant to) have a big-hitting lineup, so intuitively I think they should try to minimise variance.
   25. mgl Posted: June 02, 2006 at 01:26 AM (#2047734)
The sim does use some bullpen strategy, but I don't know off the top of my head whether it really takes advantage of the consecutive lefties. I doubt it.

I doubt that there is much of a difference in the variances among teams and doubt even more that it makes any difference in terms of win expectancy for a game or for a season.

The Phillies have a very good lineup period.
   26. Chris Dial Posted: June 02, 2006 at 01:54 AM (#2047786)
Why don't you drop in and troll Chris. The "silly" refers to the 3 lefties in a row. It has nothing to do with the rpg. Go back to your cave.

I don't like to do this, but mgl, boo hoo.

One of two things is true - either your sim doesn't properly account for three LHBs, which makes it worthless for this analysis, or batting three LHBs isn't silly, as it generates moreruns than almost any other lineup.

You call it trolling - I call it noting some poor analysis.
   27. Chris Dial Posted: June 02, 2006 at 01:56 AM (#2047788)
The sim does use some bullpen strategy, but I don't know off the top of my head whether it really takes advantage of the consecutive lefties. I doubt it.

I doubt that there is much of a difference in the variances among teams and doubt even more that it makes any difference in terms of win expectancy for a game or for a season.


Well, if it doesn't make any difference in teh run expectancies for bullpen strategies than having 3 LHBs doesn't matter, and your ranting is out of place.
   28. mgl Posted: June 02, 2006 at 02:33 AM (#2047860)
Geez Chris, did you get up on the wrong side of the bed? Having 2 and especially 3 lefties in a row gives the opposing team a great opportunity to leverage the use of a left-handed reliever. We don't need a sim to tell us that. In fact, my 10 year old son can figure that out. The purpose of running the sim was to see in general which juxtaposition of players generates the most amount of runs versus a RHSP or a LHSP. If lineup A generates around the same number of runs as lineup B (which almost any reasonable A and B will) given one and only one pitcher (say a RH or LH starter for 9 full innings), then obviously the lineup with 2 or 3 left handers in a row is going to have a worse win expectancy in the real world.

Here is the important point, for those of you who are lexigraphically challenged:

You don't need an analysis to know that there would never be a good reason to bat 3 lefties in a row unless any lineup without 3 lefties in a row is going to produce significantly fewer runs per game, not including bringing in a LHRP to face those lefties in a high leverage situation.

Since we also know that just about any reasonable lineup is as good as any other, we also know that there is NEVER a good reason to bat 3 lefties in a row. This should have been evident to anyone who watched that WAS game the other night, when Stanton was brought in in the 8th inning of a 1 run game and fanned all three lefty batters.

The purpose of the "analysis" was simply to show that there are plenty of other lineups that do not have even two lefties batting together that are just about as good as Manuel's <strike>silly</strike> moronic lineup the other day.

What particular thing I stated or argued that you disagree with, I have no freaking idea, Chris. You are the King of the Red Herrings. You jump in to a thread completely misconstruing what I said (about Manuel's lineup being "silly") and then with egg on your face you levy some other ridiculous criticism.

One of two things is true - either your sim doesn't properly account for three LHBs, which makes it worthless for this analysis, or batting three LHBs isn't silly, as it generates moreruns than almost any other lineup.

That statment is true, except once again, you are not following the argument, as the "analysis" (the sim) was not meant to figure out the efficacy of batting 3 lefties in a row.

I don't know why I bother. I suggest you and I drop it. I have nothing against you. I merely think that one of us often has a screw loose and I'll let the BTF readers decide for themselves which one (to themselves I hope).
   29. John Lowenstein Apathy Club Posted: June 02, 2006 at 02:41 AM (#2047880)
You know, for someone who dishes out gratuitous insults like they were chocolate candy, you sure can't take it.
   30. Chris Dial Posted: June 02, 2006 at 02:49 AM (#2047895)
then with egg on your face

Why do you think I have egg on my face? Because you called me a few names? Waving your "cred" around doesn't carry any weight with me. Your analysis here stinks.

You don't need an analysis to know that there would never be a good reason to bat 3 lefties in a row unless any lineup without 3 lefties in a row is going to produce significantly fewer runs per game

But the 3 LHB lineup *DIDN'T* "produce significantly fewer runs per game" in your sim.

How do you explain that?
   31. Dr. Vaux Posted: June 02, 2006 at 02:51 AM (#2047903)
How is he supposed to take it? He explained the way that Chris Dial had misinterpreted what he'd said, and I assume that Dial read the explanation and understood it. I also was having a hard time figuring out how the three lefties in a row mattered if that lineup scored the most runs, but now that MGL explained matters, I get it. The bit of information that Dial and I overlooked initially was that the simulation wasn't changing pitcher-handedness.

Case closed.
   32. JC in DC Posted: June 02, 2006 at 02:54 AM (#2047906)
You don't need an analysis to know that there would never be a good reason to bat 3 lefties in a row unless any lineup without 3 lefties in a row is going to produce significantly fewer runs per game

But the 3 LHB lineup *DIDN'T* "produce significantly fewer runs per game" in your sim.


Methinks you're missing the point, Chris. I think the point is, that since any reasonable lineup will produce about the same amount of runs, there's no reason to give your opponent the enormous tactical advantage of having his LOOGY able to face or displace 3 lefties in a row. If, however, the world were such (and it's not, mgl believes) that UNLESS you had 3 lefties in a row, you would score fewer runs, ONLY THEN should you start 3 lefties in a row.

Did I get that right, MGL?
   33. Chris Dial Posted: June 02, 2006 at 03:09 AM (#2047931)
How is he supposed to take it?

Take what? Me saying 3 LHBs wasn't "silly"? Yes, that's cruel. Where did I get those manners?

Then mgl made posted #22.

I'm well aware that plenty of times batting 3 LHBs in a row isn't a good idea. I complain about that type of lineup construction plenty.

But depending on how good the LHBs are compared to how good the available RHBs are it may just not matter very much *if at all*. It isn't "silly" or "moronic".

And using a single inning where a LHP struck out three LHBs as clear evidence is ridiculous.

I'm not missing the point. The case is overstated.
   34. Rob Base Posted: June 02, 2006 at 03:27 AM (#2047971)
Waving your "cred" around doesn't carry any weight with me.

Your manbreasts though, are about 25 lbs. each.
   35. Rob Base Posted: June 02, 2006 at 03:29 AM (#2047973)

But the 3 LHB lineup *DIDN'T* "produce significantly fewer runs per game" in your sim.

How do you explain that?


im-be-cile
   36. North Side Chicago Expatriate Giants Fan Posted: June 02, 2006 at 04:37 AM (#2048057)
Your manbreasts though, are about 25 lbs. each.

im-be-cile


Dude, I assume it is entirely clear to everyone here that Dial either 1) totally missed mgl's argument, or 2) just wanted to fight, but what is with the weak ad hominems? Even if he just wanted to argue with mgl, does that warrant manbreast comments?
   37. Banta Posted: June 02, 2006 at 04:58 AM (#2048077)
Dude, I assume it is entirely clear to everyone here that Dial either 1) totally missed mgl's argument, or 2) just wanted to fight, but what is with the weak ad hominems? Even if he just wanted to argue with mgl, does that warrant manbreast comments?


Come on, don't kill a Base/Dial fight. They're so special and delicious.
   38. The Balls of Summer Posted: June 02, 2006 at 06:30 AM (#2048097)
Dude, I assume it is entirely clear to everyone here that Dial either 1) totally missed mgl's argument, or 2) just wanted to fight,

I don't think Dial missed anything. MGL's argument wasn't clear, and when Dial pointed that out, he called him a troll.

Burley's comment was right on.
   39. North Side Chicago Expatriate Giants Fan Posted: June 02, 2006 at 06:36 AM (#2048101)
I don't think Dial missed anything. MGL's argument wasn't clear, and when Dial pointed that out, he called him a troll.

Burley's comment was right on.


Fair enough. Still, I think mgl's argument was clear enough to understand, once you got past the fact that the rpg expectancy was largely extraneous. I didn't need the later explanation to get that, but maybe that was just me.
   40. SouthSideRyan Posted: June 02, 2006 at 07:15 AM (#2048114)
I agree with Dial, there doesn't seem to be a point to a run expectancy formula for lineups if you're going to ignore the likelihood of pitching changes. For somebody who harps about how useless other models/stats are, it would seem he wouldn't use such an incomplete projection model to prove his point. Based on the idea that two lefties should never hit back to back (MGL's thought) he's choosing a lineup that is .15 R/G worse than Manuel's "silly lineup" as his optimal lineup.
   41. Mister High Standards Posted: June 02, 2006 at 01:18 PM (#2048170)
Dial was absolutly correct - if MGL's sim doesn't account for the platoon issue than it has no point being used as part of this particular analysis.
   42. Russ Posted: June 02, 2006 at 01:31 PM (#2048175)
Burrell is usually sixth because I also have A. Jones, C. Jones. Guerrero and Teixeira.

Ahh... the wonders of the 6-team, AL/NL mixed strat league.
   43. Andrew Edwards Posted: June 02, 2006 at 01:35 PM (#2048181)
I actually agree with Dial here. The sim tells us that the Phillies batting three lefties in a row scores more runs than several other more conventional options. This is a surprising result. Rather than just hand-wave it away and say "but we know that doesn't make sense" it behooves us to investigate.

A few explanations are possible:

1) Conventional baseball wisdom, as held by 10-year-olds, is wrong. This has happened before and would be an interesting discovery.

2) The sim is not reproducing the actual effect of batter handedness in lineup design. This is troubling because the impact of this weakness in the sim is unknown. It means that the sim is mis-estimating certain lineups in ways we don't understand, and should bother the hell out of us because it may be keeping us from discovering true 'optimal' lineups. It is telling us that 3-lefties-in-a-row is better than it really is. Who knows what else it is telling us is better or worse than it actually is.

Maybe having a lefthanded #3 or #4 hitter is weaker than it appears because when there are high-OBP guys ahead of them, runners on base invites the LOOGY and reduces, at the margin, the value of thier at-bats. I have no idea, and if my #2 is correct, neither does anyone else. That's bad.

MGL, Dial was harsh because you're a better analyst than to just wave away a counterintuitive result with a line about your 10-year-old 'just knowing' things.
   44. Barry`s_Lazy_Boy Posted: June 02, 2006 at 01:40 PM (#2048184)
I'm with the Dial camp on this one. If it's so obvious that the 3-lefties doesn't make sense, then the simulation doesn't work well.
   45. SoSH U at work Posted: June 02, 2006 at 02:11 PM (#2048203)
Who is Chuckles? Is that a nickname for Manuel?
   46. fra paolo Posted: June 02, 2006 at 02:18 PM (#2048208)
I'm with the John Lowenstein Apathy Club on this one. mgl treats journalists harshly, when he hasn't a clue about journalism (nor it seems, as displayed in this case, developing clear arguments). Every time someone gives him a hard time on these pages he starts bawling 'troll', and eventually runs home to mommy. Or else he tries to throw up a smokescreen about hyperebole or 'just writing for an internet bulletin board'.
   47. b-ball23 Posted: June 02, 2006 at 04:21 PM (#2048323)
Dial was clearly trying to pick a fight.
   48. SoSH U at work Posted: June 02, 2006 at 04:39 PM (#2048344)
If its possible to think that Dial was trying to pick a fight AND mgl doesn't handle criticism well, I'm in that camp.
   49. JC in DC Posted: June 02, 2006 at 04:42 PM (#2048349)
I wish mgl would return, b/c I just don't think you guys get his point. I assumed his point was that regardless of lineup construction, sims will produce similar rpgs. Thus, there's no reason to bat 3 lefties in a row (or not to bat them in a row). HOWEVER, in the real world where teams employ LOOGYs, if you bat three lefties in a row, you're essentially creating a tactical advantage for your opponent where there should be none. Isn't that what he was arguing?
   50. GGC don't think it can get longer than a novella Posted: June 02, 2006 at 04:48 PM (#2048354)
I wish mgl would return, b/c I just don't think you guys get his point. I assumed his point was that regardless of lineup construction, sims will produce similar rpgs. Thus, there's no reason to bat 3 lefties in a row (or not to bat them in a row). HOWEVER, in the real world where teams employ LOOGYs, if you bat three lefties in a row, you're essentially creating a tactical advantage for your opponent where there should be none. Isn't that what he was arguing?


That's what I understood. The only guy that I thought that was out of line was Rob Base. If I didn't know he was a lawyer, I'd think that that he was a kid; what with the use of l33tsp33k and infantile humor and his feuds with Dial and levski. Can those two be annoying? Perhaps, but it takes to to Tango Tiger.
   51. Rob Base Posted: June 02, 2006 at 04:50 PM (#2048356)
"I wish mgl would return, b/c I just don't think you guys get his point. I assumed his point was that regardless of lineup construction, sims will produce similar rpgs. Thus, there's no reason to bat 3 lefties in a row (or not to bat them in a row). HOWEVER, in the real world where teams employ LOOGYs, if you bat three lefties in a row, you're essentially creating a tactical advantage for your opponent where there should be none. Isn't that what he was arguing? "

He was saying that in his own sim, which he represented would not account for the tactical disadvantage of having 3 lefties in a row, a number of lineups produce similar results. So there's no reason to give your opponent that tactical advantage. Dial was doing his thing where he pretends not to understand just so that he can have an argument on an internet message board.
   52. The Balls of Summer Posted: June 02, 2006 at 05:41 PM (#2048410)
I wish mgl would return, b/c I just don't think you guys get his point. I assumed his point was that regardless of lineup construction, sims will produce similar rpgs. Thus, there's no reason to bat 3 lefties in a row (or not to bat them in a row). HOWEVER, in the real world where teams employ LOOGYs, if you bat three lefties in a row, you're essentially creating a tactical advantage for your opponent where there should be none. Isn't that what he was arguing?

I would say this became clear later in the discussion, and that intuitively, it makes sense. However, I would still contend that it wasn't clear from the start, and when Dial pointed it out (snarkily yes, but he did), MGL called him a troll.
   53. Chris Dial Posted: June 02, 2006 at 07:46 PM (#2048513)
MGL said this:
"And yesterday he was asking for trouble by batting 3 lefties in a row."

I have no problem with that. I think that is an appropriate level of critique for what Manuel did..

Calling the lineup "silly" and "moronic" all the while *demonstrating* that the lineup is nearly optimal against the given SP just makes no sense.

Even if the points aren't related, he either doesn't support this:
" A team should NEVER have 3 lefties in a row in a lineup. Never! If possible, they should not have two."

except with:
"In fact, my 10 year old son can figure that out. "

Now some at this site need to step back from the worship of all stats Mgl puts forth, because when he says this:
"Here is the important point, for those of you who are lexigraphically challenged:

You don't need an analysis to know that there would never be a good reason to bat 3 lefties in a row unless any lineup without 3 lefties in a row is going to produce significantly fewer runs per game, not including bringing in a LHRP to face those lefties in a high leverage situation."


Now, he showed that the 3LHB lineup produced near optimal results. So there might be a slightly better one to remove the possibility of 3 Ks in the 8th inning, but to call the lineup "moronic"? Or "Never should have happened" is just too much.

The first observation was okay, but when you start slinging hyperbole, you better back it up better than that - because unlike some, I don't "Bow Down to the King".
   54. Rob Base Posted: June 02, 2006 at 09:03 PM (#2048576)
because unlike some, I don't "Bow Down to the King".

must ... not ... make ... Burger King ... remark ....
   55. Boots Day Posted: June 02, 2006 at 09:48 PM (#2048603)
This reminds me of the discussion about the last MGL sim we hashed over, the one that measured the result of the Barry Bonds IBB with the bases loaded. In both of those, MGL eventually admitted that the sim omitted key pieces of information that, IMO, rendered them irrelevant.

Here, we already know that different lineup orders don't make much difference. What we don't know -- or at least what I've never seen demonstrated -- is the effect of the multiple consecutive lefties in the age of the LOOGY. Running a bunch of sims on various lineups without taking the LOOGY effect into consideration would seem kind of pointless.

Dial wins this one easily.
   56. Harold can be a fun sponge Posted: June 02, 2006 at 09:52 PM (#2048606)
Dial was absolutly correct - if MGL's sim doesn't account for the platoon issue than it has no point being used as part of this particular analysis.

Why does it have to be at one extreme or another? Because the sim doesn't properly model one case, we should throw it out?

Frankly, I think the sim would be more useful in this case if it were simpler, and didn't model *any* bullpen usage. Just say that these are what these lineups would produce if they faced an average RHP in every PA. Treat those numbers as a starting point. Then we can subjectively judge the effectiveness (or lack thereof) of putting 2 or 3 lefties together. My guess (and MGL's) is that the sim is going to tell us that they're basically even, and a detailed analysis of platoon tactics would show that the platoon factor is more important than the variations that the sim sees.

The real problem here is that the sim is trying to model certain things but not doing them realistically.
   57. Bad Doctor Posted: June 02, 2006 at 10:00 PM (#2048610)
I'd like to see mgl's reply too. Because his defenders seem to think "the simulation wasn't changing pitcher-handedness" and "would not account for the tactical disadvantage of having 3 lefties in a row," when he earlier said, "[t]he sim does use some bullpen strategy." If it was using some bullpen strategy, and the "silly" line-up still added 1 run a week (and that's with Dellucci, Nunez, and Fasano instead of Burrell, Bell, and Lieberthal), well, that's why we tend not to trust common sense over an empirical model. And even if we know that three lefties in a row provides a penalty, would it be a big enough penalty to make up the 1 run (or more once we put the three better hitters back in the lineup) a week? Does that fact that Abreu has pretty tame splits and is breaking up the two more normal-splitted lefties matter? I don't think it's that open and shut based on what I've seen of mgl's model's results.
   58. Chris Dial Posted: June 02, 2006 at 11:40 PM (#2048676)
Why does it have to be at one extreme or another?

As I said, mgl's observation that it was "asking for trouble" was about right. It's the pushing to moronic and "never put 3 LHBs together" that I question.

Those aren't extremes, IMO. I mean, in another thread MGL described his simulator as "the most advanced one ever" (or the best one ever - or something along those lines).

My point was that the case was overstated. Nothing dramatic - and I didn't get snippy until mgl felt like rattling my cave.
   59. Harold can be a fun sponge Posted: June 03, 2006 at 12:05 AM (#2048691)
I agree with you in #59 and #54, Dial.

If it was using some bullpen strategy, and the "silly" line-up still added 1 run a week

Whoa, whoa, WHOA! Where did the "silly" lineup add a run a week? The silly lineup came in at 5.950. There was a completely-alternating lineup at 5.952.

The lineups that came in notable worse (<5.8) used a different set of players. You can't compare across those.

Of course, it does smell a little fishy that he apparently projects Dellucci, Nunez and Fasano to outhit Burrell, Bell and Lieberthal.
   60. Backlasher Posted: June 03, 2006 at 12:23 AM (#2048715)
The real problem here is that the sim is trying to model certain things but not doing them realistically.

Vinay,

Would this in and of itself be a problem. Now you might be able to provide some fresh knowledge on this. Correct me where you think I'm wrong: I'll assert becuase its too much effort to IMHO.

For practical purposes, you usually build a simulation exactly like you would a labratory experiment. You are trying to isolate one aspect of performance. You hold some things constant, you model some of the easier things to model, and you use a randomizer so that over long enough runs that which you modeled would have produce a distribution or approximation of a certain measure.

That measure is then used as an input into another model, or the prior distribution is your first Baysain approximation for your learned distribution.

It would seem what you are demanding (much less so than others, particularly two others) would not be a simulation, it would be an Oracle (or at least an expert system) that would be impractical to code. It would take tons of years. You would always be in development.

Now from my limited vantage point, this is what mgl did. He used a simulator designed to isolate RS against certain lineups. That spit out a number, but practically he treated as a first Baysian approximation. He then took that output T= P(R|O) and said what is S=P(R|T U 3LHB) and Q=P(R|T U ~3LHB), and then approximated the output based on the limits of those two functions to find Q >> T.

Now he did that in his head, and didn't show those calculations or express that notation. ANd he also presumed some common sense on those limit functions.

But it looks like to me, that he arrived at a correct answer, and most of the criticism is based on people not being able to instinctively see the math or not having the result precisely quantified. There may have been other social dynamics at play also.

I'm interested in you showing me where you disagree.
   61. Daryn Posted: June 03, 2006 at 12:31 AM (#2048726)
But it looks like to me, that he arrived at a correct answer, and most of the criticism is based on people not being able to instinctively see the math or not having the result precisely quantified. There may have been other social dynamics at play also.

Or even worse, seeing the math but calling him on the carpet for not expressly describing it (until Post 28).
   62. Greg Maddux School of Reflexive Profanity Posted: June 03, 2006 at 12:40 AM (#2048739)
Of course, it does smell a little fishy that he apparently projects Dellucci, Nunez and Fasano to outhit Burrell, Bell and Lieberthal.

Against a right-handed pitcher.
   63. Srul Itza Posted: June 03, 2006 at 01:09 AM (#2048791)
I don't "Bow Down to the King".

mgl was Tut? I never would have guessed. Don't I feel like an ostrich now.
   64. Daryn Posted: June 03, 2006 at 01:22 AM (#2048823)
Buried with a donkey, he's my favourite honky, mgl...
   65. Backlasher Posted: June 03, 2006 at 01:29 AM (#2048830)
Buried with a donkey, he's my favourite honky, mgl...

Born in Arizona, Move to Babylona
   66. Daryn Posted: June 03, 2006 at 01:30 AM (#2048831)
Got a condo made a stone-a...
   67. Backlasher Posted: June 03, 2006 at 01:34 AM (#2048836)
He ate a crocodile.
   68. Daryn Posted: June 03, 2006 at 01:50 AM (#2048867)
Something rhymes with Nile...
Buried in his Jammies...
   69. JC in DC Posted: June 03, 2006 at 01:52 AM (#2048873)
you people just gobble this #### up like it's gospel


This is the first time I've EVER been accused of gobbling up mgl's or any saberama mama's stuff. You couldn't be more off Base, Toiletman.
   70. Gaelan Posted: June 03, 2006 at 02:40 AM (#2048966)
No one is pretending to misunderstand anything. MGL's post was retarded, case closed. His sim, which is even "better" than the analysis in the book he spent years writing, shows the L-L-L lineup scores more runs than any other. Well, if it doesn't take into account relief pitcher usage against such a lineup, WHICH WAS THE ENTIRE ####### POINT OF THE THREAD, WHICH MGL HIMSELF WAS HARPING ON, then this superfuckingadvanced sim is WORTHLESS. Of course, MGL admits he doesn't even know whether the sim does that, just like he can never remember whether UZR takes into account this or that, or whether Super L Weights includes something else. Yet you people just gobble this #### up like it's gospel, when their own creator doesn't even know what the #### he's measuring. Didn't he rank Jason Michaels as like one of the 20 best players in the majors before the season -- ignoring the fact he hadn't batted against a RHP in a few years?


This might be the dumbest thing I have ever read. This site really is going down hill.
   71. Barry`s_Lazy_Boy Posted: June 03, 2006 at 03:13 AM (#2049005)
I agree with Backlasher's 61. And mgl attempted to explain, but he has the communication skills of a 10th grade nerd, and just flew off the handle with "silly" and "moronic". Although I've noticed his writing on the site is improving quite a bit.

His lack of interpersonal skills makes me think his stats may be right. Other than the fielding stuff.

You must be Registered and Logged In to post comments.

 

 

<< Back to main

BBTF Partner

Support BBTF

donate

Thanks to
The Piehole of David Wells
for his generous support.

Bookmarks

You must be logged in to view your Bookmarks.

Hot Topics

NewsblogThe Baseball Equivalent of Hitting on 16 | FanGraphs Baseball
(26 - 7:46am, Apr 23)
Last: Sunday silence

NewsblogJ.R. Gamble: Albert Pujols' 500-Homer Chase Is A Bore, But That's Baseball's Fault
(29 - 7:39am, Apr 23)
Last: JE (Jason Epstein)

NewsblogOTP April 2014: BurstNET Sued for Not Making Equipment Lease Payments
(2055 - 7:37am, Apr 23)
Last: You Know Nothing JT Snow (YR)

NewsblogRoyals G.M. Dayton Moore believes hitting will come around
(7 - 7:32am, Apr 23)
Last: Walt Davis

NewsblogOT: The NHL is finally back thread, part 2
(191 - 7:20am, Apr 23)
Last: Flynn

NewsblogJosh Lueke Is A Rapist, You Say? Keep Saying It.
(12 - 6:32am, Apr 23)
Last: Bring Me the Head of Alfredo Griffin (Vlad)

NewsblogDaniel Bryan's 'YES!' chant has spread to the Pirates' dugout
(177 - 6:30am, Apr 23)
Last: SouthSideRyan

NewsblogCameron: Numbers don't lie: The decline of Pujols is stunning
(210 - 5:23am, Apr 23)
Last: Bunny Vincennes

NewsblogMartin Maldonado suspended
(34 - 5:19am, Apr 23)
Last: Bring Me the Head of Alfredo Griffin (Vlad)

NewsblogMike Trout And Bryce Harper Are Baseball’s Best Young Position-Player Duo Ever
(9 - 2:57am, Apr 23)
Last: Walt Davis

NewsblogOMNICHATTER for APRIL 22, 2014
(90 - 2:20am, Apr 23)
Last: Joyful Calculus Instructor

NewsblogOT: The Soccer Thread March, 2014
(1048 - 1:51am, Apr 23)
Last: Richard

NewsblogOT: NBA Monthly Thread - April 2014
(468 - 1:05am, Apr 23)
Last: robinred

Jim's Lab NotesWe're Moved! (And Burst.net can bite me!)
(106 - 12:37am, Apr 23)
Last: Phil Coorey is a T-Shirt Salesman

NewsblogESPN: W. P. Kinsella: Where It Began: “Shoeless Joe”
(82 - 11:54pm, Apr 22)
Last: Perry

Demarini, Easton and TPX Baseball Bats

 

 

 

 

Page rendered in 0.8994 seconds
52 querie(s) executed