THE BOOK--Playing The Percentages In Baseball

Is it actually more difficult to hit off the bench than it is to hit while playing the field?  This seems like one of those old cliches that sabermetricians would have debunked by now.

It’s true.  See The Book, p. 111-113.

You can also see the followup to that study here:
http://www.baseballprospectus.com/article.php?articleid=5404&mode=print&nocache=1172518606

Thanks.  Eventually I will get around to buying The Book.

I’m sure I should have been able to find this but couldn’t.  Where is the calc to get the PosAdj. for each position.  Thanks.

Regarding the DH (and PH) penalty - shouldn’t teams be trying to eliminate this somehow? If they think the cause is that the DH gets stiff just sitting on the bench, why not have him stretching/jogging/biking/throwing with a trainer in the clubhouse? Or hitting off a fancy pitching machine?
Or if they think it’s a matter of mental focus, why not come up with something that forces them to be mentally engaged?

cannatar: I agree.  There’s a huge leverage possibility here.  Andy said this is the wOBA difference for a guy as a pinch hitter, and overall:

“...the average pinch hitter was .329 for the season, and the actual performance of the pinch hitter was .298.”

A 31 wOBA points difference translates into 7 runs per 270 PA (the number of pinch hit PA per NL team in 2008).  And remember, PH come into higher leverage situations, so it’s probably an LI of 1.4 or 1.5, so these 7 runs have an impact of 1 win.

Is it possible that there is something that players can do so that they can maintain their effectiveness?  You get a free win here if you can figure it out.  A win on the open market costs over 4 million$.

Tango, after our lengthy discussion about positional weights in which I accused you of overrating CF, it seems you’ve seen the light! Again, here are my rankings for the 2001-2005 period (the last years I have calculated):

DH 0
1B -.9
LF/RF -1.5
CF -2.1
2B -2.3
3B -2.4
C -3
SS -3.2

which, translated to your scale, would be

DH -2.25
1B -1.35
LF/RF -.65
CF -.45
2B -.25
3B -.15
C +.75
SS +.95

Furthermore, I would note that the strong trends in the years leading up to 2005 were a convergence of SS and C (SS replacement level going up, C replacement level going down) and a greater separation between CF and the corners.  It would be completely reasonable, if that trend continued, to see C pass SS and establish itself as the scarcest position by some margin, while CF would fully catch up to 2B/3B.  In which case, we are now in complete agreement!

The translation of yours into mine is to reverse the sign, and subtract 1.925.  That gives us this:

1.3 SS
1.1 C
0.5 3B
0.4 2B
0.2 CF
-0.4 LF/RF
-1.0 1B
-1.9 DH

Your gap between CF and the corner OF (0.6 wins) is a little too tight for my tastes.  It’s not that crazy though.  I would have accepted as low as a 0.75 gap.

Generally speaking, your 3B, 2B, CF numbers pretty much align to my 0.25 numbers.  Your SS numbers I simply can’t accept for reasons we’ve discussed in the past, and I’d be happy to continue that discussion if someone wants to.

So, we’re in a basic position of agreement, outside of the SS.

And yes, I’ve seen the light.  You may have missed a recent thread here on the topic in which I try to handle the issue of the difference between IF and OF as a handedness issue in a somewhat logical manner:
http://www.insidethebook.com/ee/index.php/site/comments/fielding_differences_in_the_positions_take_2/

When I run the Fans’ Scouting Report by position, these are the overall valuations I get:
rating Pos
64.4 CF
58.0 SS
56.4 2B
56.2 RF
53.5 3B
47.2 C
44.7 1B
43.1 LF

Consider each rating point as worth roughly 1 play.  Note that there are no lefthanded infielders.  So, a lefty CF who ranks as a 64.4 would probably be 24.4 as an infielder.  We have about 14.5% of all players as lefties.  I’ll presume that means we have close to 30% of our CF as lefties.  So, 70% of 64.4 and 30% of 24.4 is 52.4.

So, we see that by this measure, the CF and 3B are pretty even, as fielders.  And, historically, CF and 3B are pretty even as hitters too. 

***

Anyway, just look at the SS, 2B, and 3B.  There is simply not that big a gap between all those fielders.  There is certainly not a 1 win difference in the fielding value of a SS compared to the other two.  The guys you find at 2B are very similar to the guys you find at SS.  Not as good obviously.  But, not a 1 win difference.

***

Also note the enormous gap in 2008 between RF and LF in fielding.  Just enormous.  LF are simply horrible fielders in 2008. CF is 8 “points” ahead of RF and 21 “points” ahead of LF.  That averages to 14.5 “points” which is roughly 1 win.

So, the identity of the players, their toolsets, tells us exactly who is playing out there.  So, to compare CF to the corner OF, you need this kind of adjustment.  And in 2008, you need a special adjustment for LF.  Guys who used to be considered below average fielding LF are now considered above average fielding LF, simply because the talent base has been lowered severely.

***

This is why I trumpet my fielding spectrum.  It is derived from UZR of players who play multiple positions.  And I get the sanity check from people who actually watch baseball and tell me what kind of players are there.

***

Catchers of course are their own universe.

Tom or Justin, where/how do you apply those Position Adjustments to a player’s value rating? It surely can’t be division, right? Thanks.

Add ‘em.  Whatever you get for Jeter, add +0.75 wins per 162 G played (or 700 PA).

Ohhh. Got it. Now Justin’s pro-rating position adjustments makes much more sense, using 10 RAR/W and such. Thanks.

Thanks for the link, Tango--there’s little I’ve had time to do lately, but at the least I can post these data.  I’m sure the methods behind them aren’t perfect, but I think they’re pretty good.

It does seem that I have DH’s too high--they were rated at -15 runs/season.  I’ll switch to -17.5 runs in the next update.  Not that big of a difference.  I still have some other things to do as well before I consider these things “final,” like using custom linear weights that include the 2008 season, and using patriot’s park factors. 
-j

Justin, is there a specific reason as why you prefer to use runs instead of wins?

Part of it’s personal preference.  I don’t like dealing in numbers where the range of values for most players is 1-5 or so.  I find it easier to think about a compare players using numbers that vary from 10-50.  Small headed, I guess.

Also, the conversion from runs to wins is just an approximation in my methods (everything would be converted to runs first, and then to wins), and therefore just adds another layer of uncertainty.  People have enough trouble understanding where I’m getting the runs data from, especially (for whatever reason) with fielding.  Explaining the 1 win equals average total runs per game approximation just makes it more complicated, and arguably makes the meaning of the data less precise.  So I personally just prefer to just use runs.

You can always divide by 10 in your head to convert to wins, of course.
-j