August 09, 2011


Have to go to work in exactly one hour. The monk part of the week begins again. I veg on my 4 day weekends. No that's not true - but time seems to get away from me. It's like a mini-vacation every week but the problem is that the way things are going I'm finding more and more comfort in shutting everything out. It's an interesting head I live in.

Lately I've been thinking about batting orders. You know, because, that's important. But there seems to be a science to setting out a baseball lineup and lately I've been thinking about it. a lot. What can I say?

Relative to how you set your batting order, there are a handful of different theories about how to set a baseball lineup but the most prevalent is the one that goes...

1. Best speed. Low strikeouts.
2. Hit and run guy, some power OK, low strikeouts.
3. Your best all-around hitter. Hits to all fields.
4. Your Home Run guy. RBI's.
5. Second best Home Run guy. Usually a lower batting avg. than #4, but power.
6. Usually lower avg. and power but reliable under pressure.
7. Lesser version of #6
8. Most often slowest, a lot of times the Catcher to spare his legs. You get what you get.
9. In the National League, this is 99.999% the Pitcher unless the manager is Tony LaRussa. In the American League there's two lines of thought - this is either your worst hitter or your "second lead off man" with good speed but hits for a lower avg. than your #1 guy.

And this is pretty much how it has been since time immemorial. And you don't screw with the old heads in baseball, they know best and they know everything and they know it all much better than you, especially if high school was as advanced a level you ended at.

What I've been thinking, though, is that this is bullshit. The standard baseball lineup only really makes sense if every inning you start with the number 1 guy again. Because by the second or third inning the guy who will be batting first for you isn't your best speed and may be your slowest guy. So unless you start every inning from the top THE FIRST INNING IS THE ONLY POINT OF THE GAME THE LINEUP REALLY MATTERS.

And as the game progresses the lineup becomes less consequential. At any given inning any player may be asked to start things off, and the #1 lead-off man could really only lead-off once the entire game. Worse than that his speed is often negated by the conditions of the game. For example if the #8 batter gets on base and the #9 batter makes an out and fails to advance him, your best speed guy, should he get on base, can only go as far and as fast as the guy ahead of him. A single by the #2 batter in this case may only send your slow Catcher to 3rd, forcing your speed guy to stop at 2nd.

Anyway the point being is that as the game goes on you never know where in the lineup the next turn starts.


So what you're doing is giving more at-bats to good players, but not your best players. Objective reason will tell you that the #1 batter will come to the plate more times than anyone else. So to put a speed guy / singles hitter in that spot immediately cuts down your chances for getting the most out of your talent. If you put your best hitter in the lead-off spot, regardless of his speed, he will bat more times than anyone else during the course of the season, which improves your overall offense immediately.

There, that's a half hour of baseball. I've got to make my lunch. See how interesting it is in my head? Yeah right.

I'm going to do more on this. But for now, we've established rule #1. Your best hitter (the guy you usually place #3) should be your lead-off man. The whole idea is getting on base and putting pressure on the other team.

More to come...


sybil law said...

This would've been some excellent points to make on Earl's show! GAH!
Otherwise, I agree with you, but mostly because I got a little confused and tuned some out. Like, if we were drinking somewhere - that's what I'd do, and then I'd change the topic altogether.
So what'd you pack for lunch?

B.E. Earl said...

You made exactly the correct point. Over the course of a season, hitters who bat higher in the lineup have more plate appearances than those who bat lower in the lineup.

If it's me, I set my lineup this way:

1. Your 2nd highest On-Base-Percentage player. I'll explain in a bit.
2. Some combination of high OBP and highest SLG. Your Miguel Cabrera/Albert Pujols/Prince Fielder/Jose Bautista type.
3-9. Look at #2 and then slot the rest of your guys in the lineup in descending order based on those two metrics.

OK, I have maybe your 2nd best player batting leadoff. Why? Well, even if it's only for one inning, I want to give that #2 guy (the best hitter on your club) a chance to come to bat in the 1st inning with a runner on base. Because they are run producers. Those guys also generally have the highest OBP on their teams because they are walked so often. Losing a handful of PA over the course of a season is worth the benefit of runs in the first inning.

And don't diminish the importance of scoring runs in the first. Teams who score in the 1st inning win more often than they lose.

But that's really a simplistic look at it. Other things have to be a part of the decision. A balanced lineup that slots lefties and righties alternatively do very well. Look at the most recent Yankee and Phillies lineups. The Yanks have 3 LH, 3 RH and 3 switch-hitters. The Phillies have 3,3 and 2. And they alternate them. Makes it more difficult for opposing managers to stack LHP against them. Or RHP, for that matter.

It's a great and an interesting question. I generally agree with you, but I could see some swing based on other factors.

Oh, and check this fun internet tool. A lineup optimizer using actual stats. Fun!

There are others out there, but they all work similarly to this.

B.E. Earl said...

Or to put it another way, my optimal lineup for your 2011 Chicago White Sox would go a little something like this:

1. Carlos Quentin
2. Paul Konerko
3. Alexei Ramirez
4. AJ Pierzynski???

5. Juan Pierre?

Ah all falls to crap after that.

2001-2010 Adam Dunn probably bats 3rd in that lineup.

2011 Adam Dunn doesn't belong in the majors.

Ditto for Alexis Rios. Ugh..awful!

Gino said...

i stick with the old formula. your best, most reliable-to-get-on-base-and-advance guy should see the plate most often.
(when i was watching baseball, this was Davey Lopes, who never seemed to fail.)

and having him start off every game, and posting an early lead, helps a lot.
like earl said, the team that scores in the first usually wins.

RW said...

gino two things about Davy Lopes right off the top of my head... starting with "seemed" not actually being a statistic but a subjective quality. That leaves it sort of nebulous and unsubstantiated. But the second thing is that who is to say Lopes wasn't the best hitter on the team, and you could have been merely subjectively observing my point.

Well when all is said and done there is a way to measure it, and that would be by adding the player's On Base Percentage (OBP) and Slugging Percentage. Players with the highest couplings of these two stats will always end up being the "best" hitter on any team. The OBP tells you what percentage of the time he gets on base and the Slugging Pct. tells you when he gets on base, how much damage did he do to the opposition. But without being able to quantify and measure this we're just guessing, and your stated preference for the standard lineup is based on emotion alone.

earl I see what you're saying but there are questions about the validity of the RBI relative to how it measures a value. Bill James, of course, negates it completely as being an unfair stat. It's an achievement given to one player that requires the cooperation of everybody else. Plus it is, again, only useful with certainty in the first inning. After that he may lead-off, or have a slow guy at the bottom of the lineup on unable to advance from first to third or second to home which would materially "rob" him of an RBI. So we're back to the problem of the batting order being intrinsically different each new inning.

My point is that if your best batter bats first over the long term he will bat more times than anybody else. My lead-off for the White Sox would be Konerko, not Quentin, who is hugely overvalued and whose OBP is significantly lower than Konerko's. Plus, if you lead off with Quentin, you are also not doing much to set up your "best hitter" with an opportunity because Carlos fails a lot.

And I keep looking at his home runs per plate appearances; one homer run in roughly every 19 plate appearances. That translates to a career mark of 26.3 home runs every 500 plate appearances. Which means if he led off every game, where players easily get well over plate appearances a year, he'd add 11 homeruns a season to his totals. Exponentially apply this to doubles and triples and singles and you see what kind of production increase batting first does. It seems to me if you are looking for a lineup that creates opportunities for runs, the thing you'd want to do first foremost is have the guy who gets on, beyond just singles, for others to pick up - especially since the RBI is a stat that requires someone else to do something in order to get the maximum effect.

But wait till I get to the issue of "pitch count." If you're looking for bullshit, there it is. And what's worse is that if you took it out to its logical conclusion, which may not be too far off, you'd have the same people who tout it as important abandoning it for the sake of the game. But more about that later.

sybil - I'd buy the next one.

RW said...

"easily get well over plate appearances" should be "easily get well over 700 plate appearances..."


RW said...

I have spent the morning looking to substantiate the claim that "Teams who score in the 1st inning win more often than they lose." I can't find anything. That's not to say it isn't there, I'm only saying I can't find it & I have to get some sleep.

What are the stats for when both teams score in the first inning? Does it matter if one scores 2 and the other scores 3? If you can lead me to that I'd appreciate it.

In the meantime, here's a guy who is saying it all better than I am...

Brian said...

Man I am waaaay out of my depth talking baseball, but I do have one thought.

If you had your very best hitter at the top of the lineup, though he'd be up to bat more, it seems that he would also be coming to bat with no one on base once per game for sure (1st inning) and then very likely most of the other times, too, since he would be following the bottom of the lineup (pitchers and catchers, generally). Unless you tweaked that end of things as well...

My guess is that whatever benefit you'd derive from having him bat more would be largely offset by limiting his opportunity for RBIs.

I have absolutely no data to back this up.

Gino said...

brian: his job is not to produce RBI's as much as it is to get on base and keep the offense moving, or begin an offense after the bottom of the lineup has faltered.

RW: division of labors. lopes may not have been the best hitter, and he possibly could hit as well as a clean up guy, but his other talents (speed, cunning etc...) combined to make him a lead off man.
i'll never know cause he was never given clean up duties, although he was quite capable of the long ball at key times, too, if i remember correctly.

can we talk football now?

Brian said...

Yeah, see...out of my depth.

The Freakonomics podcast had an interesting discussion of penalty kicks in soccer a few months back. If you look at it statistically, the most likely way to succeed at a penalty kick is to kick it as hard as you can, straight down the middle. But nobody does this.

I don't remember the whole discussion, but I think it came down to, nobody wants to be the guy that misses the kick right down the middle.

I wonder if that is why nobody screws with this decades-old formula for batting lineups, as well? If you do it, and have a shitty season, you're the idiot that screwed with the system that everyone knows works.

Gino said...

no Brian... i meant the other kind of football.

RW said...

We did address speed here already; and the only times you can count on it being of an advantage is either in the first inning alone or, once the batting order ostensibly changes just because the same guy doesn't lead off every inning, he doesn't have a slow-footed guy from the bottom of the order ahead of him. So once again it is kind of illusive, and only comes into play if he's on first or second with a clear path ahead of him to home. Or perhaps if he is on first with second base open and anybody on third. Pooint being that by putting a good hitter with speed up first you are simply putting your best hitter (determined by adding OBP and Slugging Pct.) less times up at the plate for the season - which doesn't make sense. "Oh hai, I'm going to take the best hitter I have and make sure he doesn't bat as many times as possible." The logic gets faulty at that point.

Brian you have the point of points in suggesting that nobody tries it because if it doesn't work they look like an idiot. Which is so very true, and yet kind of odd because a team like the Cubs has been doing it the same way for 100 years and failing all the while; which sort of puts the emphasis on "if you do the same thing over and over and keep waiting for different results..." etc.

It's hard to improve on what's winning too. Take for example the Boston Red Sox who employ the guy who started this whole mess (Bill James) of alternative thinking on these kinds of subjects and - while they don't give in to the whole hog - never the less have made some small advances in using "sabremetrics" if only to build the team. So it's hard to improve on success, and their field manager (Terry Francona) usually caves to the customary methods.

Just for the sake of argument look at the Boston stats for this year and concentrate on the determinate measurement of "best hitter" the OPS.

If you use that the top of your lineup reads


which immediately makes every old head in baseball nuts because you're putting your "speed" behind Gonzalez and Ortiz, who move like WWI tanks. But if you look at their homeruns nothing is saying the "speed" Ellsbury and Pedroia, can't also supply the power. They are on the same homerun pace as Gonzalez and Ortiz. So is Youkilis. So what would be wrong with that magic first inning everybody keeps talking about if your two "best hitters" bat first followed by guys who can also generate power? I'll tell you why... because anybody presenting a lineup card with Adrian Gonzalez as lead-off and Dustin Pedroia as "clean-up" would have the scouts peering over the top of their reading glasses to see if Francona was nuts.

But if the batting order doesn't make sense anymore as soon as the first inning is over, what are they (theoretically) so suspicious about? The order means nothing after the first, and for the rest of the game you have improved your chances of having your best hotters appear the most times.

RW said...

that would be "hitters"... unless we've just coined a new term...

RW said...

Even the terms "lead-off man" and "clean-up hitter" have an inherent error, as the guys who are designated to those roles are really only that once a game that you can count on, and if the first three guys go 1-2-3 and out in the first your clean-up hitter is your new lead-off man. If that happens, why doesn't anybody cringe and say "OMG you've got your slowest guy leading off! What an idiot!"


Gino said...

how would the opposing teams respond to your new idea? wouldnt it change the way pitchers plan their game?

B.E. Earl said...

Gino makes an excellent point. A lot of the other team's pitching plan depends on the order of the lineup. Lefties, righties, ability, etc...

A "classic" lineup features the more dangerous hitters in the 3-5 slots. This affords the 1-2 hitters more opportunity because teams aren't going to purposely pitch around them because of what's lurking.

Maybe if you move those 1-2 hitters down in the lineup, while moving the 3-5 hitters up, you actually change the production that these players have traditionally seen.

Maybe Dustin Pedroia and Jacoby Ellsbury would suffer in the lineup batting 3rd and 4th instead of 1st and 2nd. Because instead of the juggernaut that is A-Gone, Ortiz and Youk behind them, they now have Youk, Carl Crawford (this year's version) and Josh Reddick. Not nearly as fearsome to opposing managers. Same with A-Gone and Ortiz batting 1st and 2nd. Maybe your gameplan changes with them too based on Ellsbury and Pedroia batting behind them instead of in front of them.

I don't know the answer, I'm just agreeing with Gino that you can't look at current player's stats in their usual lineup spot and project the same production if you move them around. Those numbers don't exist in a vacuum.

Look at a guy like Pedroia.

His OPS (OBP + SLG) based on his position in the batting order for his career:

1 .693
2 .855
3 .784
4 1.584 (!!!)
7 .393
8 .658
9 .811

Now, this doesn't mean a whole lot. He's been a #2 guy most of his career, so all those other numbers represent small sample sizes. VERY small sample sizes. But what if the reason was the opposing team's gameplan? Hard to tell.

But I'm gonna look into that "teams that score first win more often" thing for ya. Least I could do. ;)f

B.E. Earl said...

Check out Fangraphs (a great baseball stat site!) and look into Win Probability a bit.

For example, look at tonight's Yankee game.

In the first inning, Brett Gardner and Derek Jeter started things out with back-to-back walks. Believe it or not, those two events with no outs in the first already upped the Yanks Win Probability from 50% to over 60% based on historical averages. Then Granderson hit a 3-run bomb and all of sudden their Win Probability for the game was up around 80%. Sure, the Angels could have come back. But historical probabilities only gave them a 20% or so chance to do so.

Check it out:

RW said...

But earl you're still talking as if every inning starts off with the #1 man in the order. I'm thinking this is exactly how insidious the old thinking is. If the first three men make outs the lead-off hitter in the 2nd inning is your power guy. I'm not sure folks are checking out the math & graphical example I pointed to.

The standard approach that pitchers and catchers have toward "defending" against the opposing batting order is to know "the book" on each batter. They take each guy in the lineup separately and have the knowledge of what he's a sucker for, what he's good at hitting, and what are his traits when he comes to the plate. "Pitching around" someone is a situational thing that occurs in specific circumstances, but then again what does it mean, exactly? It's basically saying it's OK to walk someone in order to get to a weaker hitter or, intentionally, to not give the guy anything good to hit so if you walk him so what you'll get the weaker guy behind him. First of all it doesn't always pan out and secondly saying a top-to-bottom batting order would encourage that ignores the fact that it already happens with the standard lineup as well.

Two or three times I've mentioned that the batting order only makes sense in the 1st inning and thereafter the only thing that can be said about it with perfect certainty is that the guys at the top of the list will bat more times. And that's all. It seems to me that some of the arguments opposed to that rely on stressing what can potentially happen in the first inning. That's great, but what about the second and third? What about the highly touted "speed guy" having to run behind a slow catcher or pitcher ON PURPOSE because that's how it's done?

I've pointed out that giving Paul Konerko 100 more at bats a season by leading him off will more than likely result in more home runs.

"Pitching around" one batter to face a weaker one isn't something that will suddenly appear in baseball as if by magic if anybody uses the alternative lineup because it happens in just about every game already.

The basis of my argument is that when you make a batting order with your best hitter batting first and your second-best batter hitting second and so on, you end up having your best guys bat more. And since the entire purpose of batting is to get on base you automatically increase the likelihood of that happening more often because of it.

RW said...

Now I do know about win probabilities but I also know of a team on the north side of this city that can score first all they want, it still isn't going to matter.

The percents of probabilities change by the pitch as well. But I don't see axioms in probabilities. I think we can safely say it is axiomatic that the tops of batting orders come to the plate than other parts of the batting order over time. And i8f the single, solitary goal of batting is to get on base, I'd want my best hitter batting the most times. I certainly wouldn't put him in the part of the order that would guarantee he comes to the plate less often and say that's the best way to use him.

And in all honesty that chart and graphic on fangraphs doesn't actually *prove* that teams that score first usually win. I think probably nobody has that because you'd need the box scores of many seasons to show it accurately. It may be so, but even if it is I don't see how having people get on base more times is going tohamper the effort. Not trying to be obtuse, I know you're all saying you need RBI guys behind on-base guys. It's just that it doesn't have to work that way. Question everything, is how I see it. It's like that de-motivational poster of the guys getting run over by the bulls in Spain that says "TRADITION: just because you've always done it that way doesn't mean it isn't incredibly stupid."

But I go back to Brian's point here too... we'll probably never find out because no one - even if they become convinced of it - would ever dare try it.

Maybe this needs some trials in strat-o. Is Spike Lee doing anything for the next few months?

B.E. Earl said...

Actually, I'm not thinking of the first inning at all. I'm thinking about the rest of the game as well with the protection in the order stuff. No matter when Pedroia comes up in the game, he always will have lesser hitters "protecting" him in the lineup if he bats 4th rather than 2nd in this particular lineup. That will always be the case no matter what inning it is and no matter where he bats in a particular inning. The batters behind him, his "protection" will be less dangerous if you bat him 4th or 5th.

That was my point.

As for the Win Probability thing, the data behind the calculations is seasons and seasons and seasons worth of exhaustive, detailed game situations. So it actually does prove that teams who score first have a greater probability of winning. Check out this article on how it was first developed. With boxscores and retrosheet game data for every game from 1979 to 1990. The data pool has been updated since then.

B.E. Earl said...

And one more thing on Paul Konerko. He has seen significant time in each of the 4, 5 and 6 slots in the lineups.

4 3839 PA .892 OPS
5 2152 PA .838 OPS
6 1277 PA .813 OPS

What can we learn from this? Almost nothing, really. Is he a better hitter just because he was moved to cleanup spot, or was he moved to the cleanup spot because he became a better hitter?

And season-to-season offensive production, injuries, teammates (protection) all have an effect.

All I have been saying is that you can't project a batter's production from one spot in the lineup to another spot in the lineup with any accuracy as there are other factors involved. As I've said before, those statistics don't exist in a vacuum. There are many variables.

RW said...

If it is true that "All I have been saying is that you can't project a batter's production from one spot in the lineup to another spot in the lineup with any accuracy as there are other factors involved" then the same can be said about the classic lineup. Using this logic you've just told me that everything you've been saying is wrong. Why does this apply to top-to-bottom and not to the classical lineup?

RW said...

But wait! I read the article at the link you posted and unless I missed the exact reference to what you've been pointing out all I come up with is where he says "An average team, at any point in a game, has a certain likelihood of winning the game. For instance, if you're leading by two runs in the ninth inning, your chances of winning the game are much greater than if you're leading by three runs in the first inning. With each change in the score, inning, number of outs, base situation or even pitch, there is a change in the average team's probability of winning the game."

Maybe I need more background in statistics and probabilities but it seems to me that what he's saying is it's better to lead late in the game than early in the game, which would put a qualifier on scoring first... unless it's in the 9th inning it seems still problematic. And it's not exactly the match to the claim i was hoping for, but I guess it'll do.

And in a larger sense we still can't say what will be more likely to score a run; the classical setup where the best hitters hit less times, or the top-to-bottom lineup where pressure on the pitcher starts right at the top.

RW said...

There is, as you mentioned, also the lefty/righty comparisons; but you do know that even that isn't iron-clad axiomatic, right? That was, as far as I can tell, something that was observed and then finally institutionalized by John McGraw back in the 00's of the last century. It seems intuitive enough and there is data I've seen that "generally" backs it up but there always exceptions. I want to look at this later because even if a stat says a right-handed batter is better against a right handed pitcher than he is a left-handed pitcher it isn't going to matter to a manager hell-bent on platooning without looking at the facts.

What counters that these days is the head-to-head charts that keep tabs on performance when a certain batter faces a certain pitcher. I wonder, sometimes, if a manager could see that his right handed bench guy always does well against that particular right-handed pitcher, but sends a lefty out anyway because it looks like he's following baseball custom that way? Just wondering.

RW said...

Also allow me to point out that using top-to-bottom in no way says that someone like Konerko is always going to be in the #1 slot. Davey Lopes had some pop and wasn't purely a singles hitter - he could very well have had the best OBP+Slugging Pct on those Dodger teams. It isn't in the Batting Avergae - which doesn't tell you all you need to know. It's OPS - On-Base Ples Slugging.
It doesn't always end up a power hitter. Take the happy circumstance of Brian Roberts from Baltimore - in 2006 his combined numbers were over .900 - and he also happened to be the top base-stealer on the Orioles, and he was arguably a better base-runner than, say, Luis Matos from that roster (maybe not faster but smarter. Matos was as dumb as a bag of hammers).

So to isolate, for example, a player like Konerko and then think that he'd be the only kind of player the strategy would bubble up to the top of the order isn't iron clad either. In fact, taking it a step further, a top-to-bottom batting order for the 2006 Orioles would have been

Brian Roberts
Miguel Tejada
Jay Gibbons
Melwvin Mora
Rafael Palmeiro
Javy Lopez
Larry Bigbie
Sammy Sosa
Dave Newhan

And outside of the culture shock of seeing Sammy Sosa in the 8th spot (but wtf he only batted .220 that year) it more or less looks like a usable classic lineup.

Now if they hadn't had Lee Mazzili as manager than year, and had a few pitchers, they might have even made something out of it!

RW said...

that would be the 2005 Orioles... sheeesh.

B.E. Earl said...

This is fun.

I'm actually agreeing with 95% of what you are saying. I do believe your best hitters should be at the top of the lineup. With my one caveat being that the best hitter should bat second, the second best first and the rest in declining order from 3rd to 9th.

Why? I think the advantage of having your best hitter come to the plate in the 1st inning with about a 40% chance of having a runner on base (my ideal leadoff guy has a .400+ OBP) outweighs the advantage of 50 or so extra PA a year. That's an extra 50 from the 1st to the 2nd lineup spot. I don't know the exact figure, but let's call it it 50.

So 162 games * 40% = 65 times during the season he will come to the plate in the 1st with a runner on. Versus 50 extra times he will come to the plate late in the game with maybe nothing to gain.

That's a tradeoff I'm willing to take. Especially since I believe that scoring runs early is extremely important, even though I haven't exactly found the way to prove it to you via the Win Probability stuff.

But yeah...FUN!

FYI - been having so much fun with this here, me and my buddy are going to discuss it on our baseball internet radio show on Sunday night. You inspired us! :)

B.E. Earl said...

And I would have picked the same lineup for the 2005 Orioles as you. Especially given which way each batter hits.

Roberts SH
Tejada RH
Gibbons LH
Mora RH
Palmiero LH
Lopez RH
Bigbie LH
Sosa RH
Newhan LH

A nice, balanced lineup. No lefties or righties stacked in a group. That would keep opposing managers from stacking LHP or RHP against them based on splits. It's never gonna stop a manager from using a LOOGY (or heaven forbid, a ROOGY), but it's a start.

Like I stuff!

B.E. Earl said...

One last thing.

We may be over-estimating the number of additional PA a player gets when you move them up in the lineup.

In 2006, Grady Sizemore and Mark Teixeira each played in 162 games. Sizemore was a leadoff hitter in all 162 games. Tex was a 3 hitter in 103 games and 59 in the cleanup spot.

Sizemore had 751 PA.
Teixeira had 727 PA.

A difference of just 24 PA.

AND, Texas and Cleveland were fairly similar offensive teams that year. Texas sent 6272 batters to the plate that year, while Cleveland sent 6302. So the comparison between Sizemore leading off and Tex in the 3/4 hole holds up.

So is all this much ado about nothing much at all?