THE BOOK--Playing The Percentages In Baseball

You are correct that batting last has little to no impact.  Here’s one study (http://www.sportmedab.ca/pdf/Baseball.pdf), but I know there’s another one that got published in a journal and examined baseball and I believe cricket as well (I don’t know that game at all but for some reason it rings a bell as being in the study).  Not being a student anymore, I don’t have access to any good resources, but I’ll see if I can find it.

Fairly interesting study, although if this is kid’s baseball, which it seems to be (there is no mention of ages in the study I don’t think), I would think that its relevance to MLB might be pretty weak.

I don’t know how many games in MLB have been played at neutral sites over the years, but I would think that including the ASG, it might be several hundred.  Surely someone has looked at these games.

"Howard ... would be a platoon player if performance mattered anywhere near as much as reputation does. Or if he had a competent manager.”

The first explanation seems dead-on.  Does anyone believe that any other manager in MLB would platoon Howard?  Is this a veiled assertion that every manager is incompetent?

When I run a Markov:
http://tangotiger.net/innwin3.html

We see that entering the top of the 9th, the home team has a .522 chance of winning the game, when tied.  Entering the bottom of the 9th, they have a .655 chance of winning.

When I looked at empirical (actual) results:
http://tangotiger.net/innwin2.html

Those numbers are .519 and .649.

If anything, it might be a disadvantage.  In any case, I’ll call it a draw.

***

If the home team is up by 1, meaning that they have their ace reliever on the mound, their win% in the top of the 9th is actually .873 compared to a Markov expectation (with an average pitcher) of .853.

When the home team is down by 1, in the bottom of the 9th (meaning they are facing the opponent’s ace), the actual is .189 compared to the Markov of .204.

In both instances, the ace reliever helps by close to .020 wins, compared to an average pitcher.

The “advantage” in the ninth inning is whoever can get their ace reliever on the mound.

Tango, that’s interesting stuff. It seems to put to rest the notion that batting last gives an advantage - that advantage should be more extreme later in the game and your numbers show the opposite.
How does the Markov chain take into account the HFA? Does it just give a boost to the home team for every at-bat?
The empirical results seem to show that there’s a (relatively) big drop in HFA if it’s still tied after the first inning. Is it possible that starting pitchers have a little more trouble getting used to the mound on the road? But, if that’s the case, relievers should be affected, too.

The empirical results seem to show that there’s a (relatively) big drop in HFA if it’s still tied after the first inning.

I never heard that before.  Can you give us more details?

I am not sure why anyone, let alone one of the main BP guys, would think that batting last gives a strategic advantage. How about “defending last” or “pitching last.” Common sense and anyone who watches baseball should tell us that the defensive team has as many or more options in the bottom of the inning in a close game.  Infield in.  Outfield in.  Changing pitchers.  IBB’s.  Etc.  What exactly does the home team have?  Pinch hitting and bunting or not, and the way managers utilize the bunt in late game situations (too aggressively), they probably reduce their wp anyway.  Whatever changes in approach the batters can make, certainly the pitchers can do the same (opposite, I guess).

Not to mention that road teams typically bring in their closers after going ahead in tie games or games they were losing prior to their last AB.  Home teams generally do not.  If I had to give a strategic edge to one side or the other, I think I would give it to the road team.

The worse numbers in the second inning is probably because there are worse hitters batting in the second inning.  You would hope this cancels out for both sides, but I guess it doesn’t.

The 5th/6th inning change to the 7th inning is probably because of the pitching change.

I’d recommend the paper that can be downloaded here:

http://econpapers.repec.org/paper/wpawuwpga/0309002.htm

According to Ted’s research in the link above, it looks like we are talking about a .001 win difference (in some direction).  Against that is all the other things that are not modeled, as mgl says, based on what the fielding team does.

And, like I said, my empirical results more or less match the Markov expected.

In the Olympic-rule for extra innings, batting last would be preferred, since you have two guys on base, and you have to know how valuable is that runner on first base.