THE BOOK--Playing The Percentages In Baseball

Friday, May 18, 2012

Indians Batting Order - Mostly by The Book

Apparently the Indians’ view on batting orders intersect to a good degree with what’s in The Book.

Friday, April 13, 2012

If not Willie Ballgame, then who can leadoff?

The five best hitters should bat one through five, in some order or other.  The DBacks have Upton as the best hitter (meaning he should bat either 1st, 2nd or 4th), with some combination of Goldschmidt, Montero, Kubel, and Young in there.  The author however focused too much on the walks part.  There are several considerations, and walks is just one of them.  I can’t imagine that Goldschmidt would be the leadoff hitter.  Problem is, I can’t imagine anyone in that role either.  I guess Chris Young fits the bill the best, but it’s pretty tough.  Of course, any choice is better than Willie Ballgame.

I’d be interested to see MGL run some sims to see if there’s really anything Gibson can do here.

Monday, March 26, 2012

Why do great players have the most fragile of egos?

Dale Sveum points out something about a typical starting player:

“A lot of guys, you can say ‘OK, You’re going to hit here,’ and they’re like, ‘Oh, God, no, don’t do that,’ ” Sveum said. “(The Brewers’) Corey Hart was a little bit like that. Put him in the one or two hole, and he did unbelievable things. But if you put him in the fifth hole, he didn’t really want any part of that.

“Guys are like that. And you try to find out, through conversation, ‘Where do you seem to be most comfortable to where you can be most productive?’ Unfortunately, we live in a little bit of a world like that. Guys have that idea that no matter where they hit, there’s some significance behind it, instead of understanding that…

But if you had a bench player, a guy who would be appreciative just for being in the lineup, he’s not going to care where he bats, be it cleanup or ninth.  He just wants a chance to prove himself, and he’ll do whatever it takes.

The bigger you are, however, and it’s about status, and ego.  You want the bigger corner office, even if it does nothign intrinsically to your productivity.  You want that parking space, you want that number on your jersey, you want all those things that you’ve earned, even if, going forward, it does nothing at all for you.  Except, maybe, it’s now in your head that it will do something for you.

And if you are now close to retirement, and no team is actively looking for you, and some team offers you a job, and even though you used to be a cleanup hitter, guess what: you’ll be happy to hit anywhere, especially if it means you won’t ride the bench.

Is the player’s ego, the thing that got him there to begin with so intertwined with his confidence level that the better the player is, the more you have to worry about his ego?

I can see why Sveum can’t sleep at night, even though, as he noted:

...you’re going to get pitched the same no matter if you hit leadoff or fifth

Sunday, March 11, 2012

How much can a manager leverage speed and OBP with his lineup?

I was going to address this in the Kinsler thread, but I decided to start a separate one. I have always thought it ridiculous for a team to concern itself with finding someone for a particular lineup slot, as in, “We really need a leadoff hitter,” or, “We are so happy with this acquisition because we now have a proven leadoff hitter,” or some such nonsense like that.

You get the best player you can and then you construct the best lineup from what you have. It really doesn’t matter whether you have a “true leadoff hitter” or a “true cleanup hitter” or not. It doesn’t matter at all.

I am not even sure what that means anyway, so let’s try and be more specific in terms of what question(s) we want answered.

I’ll pose two specific questions which we can answer, more or less, quantitatively and which relates to this issue.

1) If we have someone who has lots of speed, or doesn’t have lots of speed, how much can we leverage that attribute (or lack thereof) by placing him strategically in the lineup?

For example if we want to acquire a speedy player that is worth exactly 2 WAR, how much more would he be worth if we plan on putting him at the top of the order versus the bottom of the order, and in doing so, we don’t affect anything else?

One way to answer that is to see how many runs per game we gain when we go from a slow runner to a fast runner at the top of the lineup versus how many runs we gain when we go from a slow runner to a fast runner at the bottom of the lineup. This is very similar to how we figure LI in a game. We compare the impact of a good player or good event to a bad event in terms of WE, at various points and situation in the game.

I took a typical Rangers lineup from last year and ran my sim with Elvis Andrus either batting second or batting 8th, as a great base runner (which he is, so I left his base running projection alone) and as a terrible base runner, like a Prince Fielder. I actually use a 1-5 scale in my sim, so I went from a 5 to a 1. This is not including SB/CS - only advancing on hits and outs by following hitters when on base. I cannot remember if it includes advances on potential WP and PB. My sim probably captures 90% of the value of base running.

I ran 500,000 games of the Rangers playing a team with a RHP. It doesn’t really matter who they played or who the opposing pitcher is. It might matter a little that it was the Rangers with a good middle of the order.

Anyway, with Andrus batting second, as a fast runner, they scored 4.167 rpg.  As a slow runner, 4.101 rpg, for a difference of 9.9 runs per 150 games. That is around what we expect from the difference between a great and poor base runner in general (an average slot in the order), so I suspect that my sim is undervaluing base running a little, or maybe it will find around the same difference in any slot, in which case, it is probably capturing most of actual base running value in real life.

Nope.

As a #8, the difference between Andrus as a great (5) and terrible (1) runner is only 4.65 rpg.  So yes, you can leverage base running with batting slot, but, this is the most extrme situation possible. My guess is that for an actual team making a decision about a player based on where they think he will bat in the order, or what other players that have slotted in the lineup, we are only talking about 1-3 runs per season.

For example, if a team is indeed looking for a lead-off or second place hitter, and player A has the same value/projection as player B, but player A is a speedster and player B is just average (and assuming that their hitting profiles are the same), how much more should they pay for player A. I think Player B is probably going to be worth only 1 to 1.5 runs more than Player B given that you are leveraging him the leadoff or second slot. I think that most teams is going to way overvalue that speedster. IOW, you should pretty much forget about the fact that you are looking for a player to fill a certain lineup slot. Like many things (clutch, batter/pitcher matchups, etc.) Use it as a tie breaker only.

2) Same question as #1, but what if we changed a player’s OBP by 20 points by adding walks only?  How much can we leverage that by batting him lead off rather than 8th? We’ll use the same method.

I did the same thing with the sim. This time I used Kinsler in the leadoff slot or the 8th slot and I ran the sim (500,000 games each) for his normal projection at the end of last year and with his OBP jacked up by 20 points by adding around 10 walks per 500 PA.

Kinsler batting leadoff, normal OBP: 4.146
Kinsler leadoff and an OBP 20 points higher: 4.201.
Gain: 8.25 rpg (Tango, how does that compare to what you would expect for an average player on an average team in an average slot?)

Kinsler batting 8th, normal OBP: 4.163 (why you don’t bat him 8th, BTW - you lose 6 runs a year!)
Kinsler leadoff and an OBP 20 points higher: 8.55 rpg.

Wow, interesting. You cannot leverage his OBP by batting him at the top of the order. I am not sure why. When my computer frees up, I’ll run some more game. Maybe I’m getting too much random fluctuation in 500,000 games. My guess is that 1 standard error even in 500,000 games is still like 5 runs per 150 games, so really, a comparison of 8.55 to 8.25 doesn’t really tell us anything.

I’ll try and run some more games with the speed thing too.  Even though the results seem reasonable, there is too much sampling error there too for the difference to be reliable to any degree…

Monday, October 10, 2011

Tigers batting lineup

Dave takes his stab at it (ESPN paywall).

Monday, September 19, 2011

Is where you hit in the lineup conducive to better performance?

I’ve been meaning to do this study for a long time. 

All data from 1993-2010.  I also limit the data to players aged 25-29.  I ensure that a player is equally weighted in the two pools being compared (and that weight is the lesser of the plate appearances in the two lineup slots being compared).

(Imagine there’s three or seven paragraphs of a yapfest going on here.  I’m not going to do it.)

Ok, now, thanks for your patience in wading through all that.  Now onto the data.

There were 62,894 matching plate appearances for the leadoff slot and the #2 slot.  That is, I’ve got the same players represented in each of the two pools, and each are weighted appropriately.  Their wOBA in the leadoff slot is .332 and it’s .333 in the #2 slot.  That difference is less than one standard deviation.  In all cases, except two, the difference in the leadoff slot and the other 8 slots was less than one standard deviation.  And what are those two slots?

Players who hit leadoff and who hit cleanup: .352 in the leadoff and .368 as cleanup, which is 2.7 standard deviations.  And even bigger gap are those that hit leadoff and bat sixth: .334 leadoff and .344 batting sixth (3.3 standard deviations).

I did this for all the batting slot combinations.  Let’s take them one at a time.

Batting 2nd and everywhere else, except 3rd, shows less than 1 standard deviation of difference.  With with batting 3rd: .362 batting 2nd, .357 batting 3rd, 1.7 standard deviations.

Batting 3rd?  There’s the aforementioned batting 2nd, as well as batting sixth: .359 batting third, .354 batting sixth (2.3 standard deviations).

Batting cleanup?  That one has alot of differences.  None to match the leaoff, but also batting 4th/5th: .365 batting 4th, .362 batting 5th, 1.7 standard deviations.

Batting 5th?  In addition to the cleanup, we have these guys batting 9th as a gap.  Batting 5th, .335; batting 9th, .323; 2.4 standard deviations.

Batting 6th?  Many differences.  There’s the two already mentioned (leadoff, cleanup), and also batting 7th: .340 sixth, .337 seventh, 2.0 standard deviations.

Batting 7th?  There’s batting sixth, and also batting eighth: .321 seventh, .326 eighth, 2.8 standard deviations.

Batting 8th?  Only what was already mentioned with seventh.

Batting last?  Virtually across the board, batted slightly worse there than in all other batting slots.

If I take a simple average, this is what I get:

We see that hitting leadoff and batting ninth is “harder”, while batting cleanup confers some sort of advantage.  We’re not talking about a huge deal of course.  Up to a five point difference in wOBA (which translates to three runs over the course of a season).

We can come up with plausible scenarios as to the reasons.  By the time the pitcher gets to the ninth hitter, he’s been cruising a bit, but when he gets to the cleanup hitter, he’s a bit worn out.  Or, perhaps the reason that the batter was placed cleanup is because he actually was a bit better hitter that day, and he was placed ninth because he was a bit worse.  (Made me wish I would have controlled for platoon advantage too now that I think about it.) It’s possible there are other selection biases to account for.

Anyway, I quite enjoyed doing that, and if someone else likes it, great, welcome to my world.  This is what makes me go to bed after midnight.

All hail Retrosheet!  Now, off to fight that war for Oceania.

Tuesday, July 12, 2011

Ducks on Pond

Nice idea from Poz, to look to see which batters are given the most runners on base.  These are the guys you want to drive in your runners.

Can we do the opposite?  How about which batters are given most opportunities to get on base and score?  I’ll propose one: most plate appearances with no outs.

The chance of scoring when you reach base with 0 outs is 45%; it’s 30% with 1 out, and 15% with 2 outs.  So, if you have one guy that reaches base .400 times and the other that reaches base .300 times, then you want the .400 guy on base with 0 outs (45% chance to score) and the .300 guy with 2 outs (with a 15% chance to score).  That’s going to give you .4x.45 plus .3x.15 runs, or .225 runs.  If you reverse it, that’s .195 runs.  So, you get 15% more runs if you can slot them the right way (under this limited illustration).

Obviously, whoever is your leadoff hitter is going to lead.  But, who is your second guy with the most chances to get on base with no outs?  Do your team, and post the results, and running commentary like Poz if you like.

Saturday, May 28, 2011

Recreational sabermetrics

Finding the barest of anomalies 13 times out of 600,000 situations.

Love that term used by Phil. 

Friday, May 27, 2011

Keeping your batting order static

Players, we are told, love to know their exact role.  They want to know the one position on the field they’ll play, they want to know the one inning they are brought out of the pen, they want to know how many pitches they are allowed to throw, and they want to know the one lineup slot they’ll occupy.

So, if you have Drew Stubbs leading off, and Joey Votto always batting third, and you have Jay Bruce always batting fifth, then when the cleanup slot opens up for a day, it’s easier to simply slot someone else there.  The Reds for example started the season with Scott Rolen there.  Rolen of course is not the healthiest of players, and so, we’d expect that slot to open up.  And when it did, Jonny Gomes filled in.

But then, Dusty Baker decided he needed to have a new static lineup, given that Rolen was out.  So, he puts Brandon Phillips as his cleanup hitter, moving Jay Bruce to second, and then Gomes + whoever rotating in the 5th slot.  Presumably, he figures, better to have the rotation in the 5th slot than the 4th slot.

He had enough of that, so then he moved Bruce back to the 5th slot, keeping Phillips as cleanup, and making the 2nd slot the rotating slot.  So, we are now here: Stubbs still and always leadoff, Votto still and always third, Phillips the permanent replacement at cleanup until Rolen comes back, and Bruce reinstalled at fifth.  The second slot being the rotating slot, you have Renteria there, and Paul Janish(!).

Then, Rolen came back, but he’s eased into the sixth slot (which was Cairo’s semi-permament slot). 

Finally, he’s given back his cleanup slot, Phillips is given back his #2 slot, and now we are finally back to where we started the season.

Except now, Rolen again didn’t play, and so, Baker wanting to keep his lineup fixed, simply rotated in the third baseman (Cairo) into the cleanup slot.

Does all this make sense?  Who knows.  But, why is it important to give the great player a fixed role, while rotating the lesser players around the 6th through 8th slots (presumably, I didn’t check, to break up lefty/righty based on opposing pitcher)?

If anyone needs confidence, it’s the lesser players.  Does Joey Votto really need to know he has to bat third, regardless of the context?  Is his ego that fragile?  Or, is Votto so confident since he first picked up a baseball bat that he’s prepared to hit anywhere in the lineup, and be awesome doing so?

Do we need real pyschologists weighing in here, or is dime-store pyschology what we need?

Reference: B-R.com

Thursday, April 21, 2011

Nice article on ESPN Insider about batting order…

http://insider.espn.go.com/mlb/insider/news/story?id=6395656

Same stuff we’ve read scores of times, but nicely reasoned and explained by the author, Matt Meyers.  Here are some highlights (the bolding is mine):

For years, baseball lineups followed a pretty basic logic: fastest guy hits first, contact and/or fast hitter is next, high-average guy bats third, power bat in the cleanup spot and the next few guys in descending order of general skill. There was nothing scientific about this method—it seemed tied to some old-school vision of small ball—but many managers remain obsessed with it despite evidence that refutes the usefulness of such strategies as bunting and base stealing.

Frankly, it’s somewhat astonishing that clubs have not been quicker to rethink how they construct lineups. Would Albert Pujols really be offended if Tony La Russa said to him, “Research shows the best hitters should hit second or fourth, so I’m going to put you in one of those spots.” If anyone, you’d think La Russa, who has experimented with the pitcher batting eighth, would be open to such an idea. Shunning such comprehensive research in favor of tradition is both arrogant and ignorant. It’s like walking into your shrink’s office and he or she saying, “I know we’ve been treating you with Prozac, but I think we should try a prefrontal lobotomy like we used to do in the old days!”

You might dismiss my rantings by saying, “it’s just 11 runs.” My response: Isn’t that kind of the point? A huge focus of sabermetrics is finding small marginal advantages, and hoping then when you add them up they amount to a legitimate edge.

Sunday, April 17, 2011

Rays fans pick lineup

Or at least Maddon is giving them credit for it.  Not just the order, but also who the players would be.

Wednesday, April 13, 2011

Is Jason Heyward the new Freddie Lynn?

In 1975, rookie Fred Lynn batted mostly 4th.
In 1976, he batted mostly 3rd.
In 1977, he batted mostly 2nd.
In 1978, he batted mostly 6th.
In 1979, he went back to batting mostly 3rd.

In one sense, it doesn’t really matter where a player hits.  Each batting slot means 18 fewer plate appearances.  So, if the choice is the 2nd or 6th slot, that’s 72 fewer PA for Heyward, and 72 more for whoever.  If a pretty good hitter is about +35 runs above average per 700 PA, that means each PA is worth an extra .05 runs.  At 72 “lost” PA, that’s 3.6 lost runs.  A little lost leverage as well, and we’re talking about 4 runs.

When I’ve done the studies in The Book, that’s pretty much what I would find.  Swap the traditional #2 and traditional #3 hitter, re-run my simulator, and I get a 2 run gain.  (If I were to swap the #2 and #6 hitters, 4 runs sounds about right.) Move the PITCHER to the cleanup slot, and it costs “only” 16 runs.  This is really what we’re talking about.  A bigger cost is made in not identifying the 8 or 9 best hitters to put in your lineup.  But, once those are identified, it’s hard to do a bad job with a batting lineup, just as it’s hard to do a bad job with your fielding lineup.  Unless you do something silly, like putting the pitcher in the cleanup slot, or putting Ozzie Smith at 1B.

That said, you don’t intentionally put Jason Heyward where he can do less damage.

That that said: players are humans with egos, and more important than fitting in players in their ideal mathematical slot is fitting them in their ideal egotistical slot.  I doubt, however, that Heyward feels the most comfort in the 6th slot.

Unrelated story: ESPN and Neyer.

Monday, March 28, 2011

Your team’s optimal batting order

Rather than create a thread for each team, I should have just started one thread for all the teams.

Bloggers speak:
Phillies.

Tuesday, March 22, 2011

Batting Order Lineup Tool

I put my comments in this thread.

Friday, March 18, 2011

Marlins batting order

Great job here.  Separate forecasts for LHP and RHP, which means separate lineups. 

Good job bringing up the DP, and I’d like to see a forecast for Stanton on DP:

I have no major concerns with Stanton in terms of grounding into double plays, as Stanton only grounded in double plays in nine percent of his opportunities last season despite hitting grounders on 43 percent of his balls in play

Wednesday, March 02, 2011

M’s batting order

USSM has the Mariners batting order for you.

Monday, February 28, 2011

Rays optimal batting order?

It would be more useful to see the forecasted wOBA for each hitter against each hand.  Then we can discuss the assumed talent that went into the thinking behind the order.  As it stands, without that, the reader has to assume his own talent levels, thereby making your own conclusions moot.

I’d love to see this done with more teams.  I think a month or two ago, someone did the Cubs and Cards?

Batting orders: fix what’s broken

I agree with Girardi, generally:

No matter who hits where, most analysis says the same thing: unless a manager does something dramatically wrongheaded, like batting his best hitter 9th, most lineup tweaks don’t make much of a difference. From a statistical sense, studies have shown that lineup optimization leads to perhaps one extra win per season.

While a little more production might be squeezed out of his league-leading lineup, Girardi seems conscious that if it ain’t broke, don’t fix it.

“You know you have your middle-of-the-order hitters and then you have your guys who are going to be your supporting cast,” he said. “How we kind of divvy up that supporting cast remains to be seen. But our lineup scored lot of runs last year, a lot of runs last year. And maybe we tweak it and maybe we don’t.”

Not so much his reasoning here, but elsewhere, he said he had to manage them as people and if he moves players around, the players may think something of it.

I agree that because you only gain a couple of runs for each switch you make, that that gain might be eroded if someone is thinking about “does he still love me?  what did I do wrong?  why won’t he call” lineup drama.

It would be great if players simply did not treat the lineup slot as status.  But, perhaps because they DO treat it as status, it gives them more confidence.

And so, teams should play up that the #2 slot in the batting order is very important.  As it stands, it’s treated as “oh, not good enough to be a leadoff hitter and not good enough to be a #5 hitter, eh?”.  And so, it’s self-defeating.

Wednesday, February 02, 2011

Cardinals 2011 batting lineup by The Book

Erik gives us his list (note 1st chart against RHP, 2nd against LHP).

I’d love for every team’s blog to try their hand at this…

Thursday, January 27, 2011

Batting Order for Redsox 2011

A reader wrote this up and asked me to post it.  I’m doing this blind, and will read it in a second.