Jimy Williams’ deployment of Astro relievers has been
a source of consternation for many of us here at BP.
Not long ago, he declared that he would begin using
his closer and best reliever, Octavio
, to protect four-run, ninth-inning
leads from time to time. Suffice it to day, that’s
about as efficient as your garden-variety
doobage-addled record store clerk. Calling on your
closer to protect three-run leads in the ninth is a
bad idea, so clearly what’s being referred to as a
“Jimy Save” runs even further afoul of common sense.

As Joe Sheehan observed in a recent group e-mail, this
practice reached its nadir last week when Williams,
unwilling to go to Dotel, who’d worked the three
previous days (twice logging eminently undemanding and
wasteful Jimy Saves), saw inferior relievers fritter
away a ninth-inning lead to the Braves and eventually
lose it in the 10th. Thanks to the Jimy Save, the
Astros’ best reliever was a feckless observer to the
conflagration unfolding on the field.

All of this leads me to wonder just how poorly
Williams has managed the Houston bullpen. In short,
Williams is riding the early-season hot streaks of
Brad Lidge and Dan
, both of whom have been excellent in
the early going, by giving them more high-leverage
innings than he’s giving Dotel. Because Lidge and
Miceli are presently pitching so well, this seemingly
hasn’t exacted much of a price. Seemingly. Still,
Williams’ misuse of Dotel is manifest, and if it keeps
up, it’s going to hurt the Astros over the course of
162 games in what figures to be a hotly fought race in
the Central. It’ll only get worse if and when Lidge
and Miceli return from the firmament.

First, let’s take a look at what the PECOTA
weighted-mean forecasts say about the Houston coterie
of relievers. Below, I’ll rank every Astro pitcher
who’s made a relief appearance as of May 10 according
to their projections. To remove workloads from the
calculus, I’ll slot them based on projected Value Over
Replacement Pitcher (VORP) per inning:

Pitcher             Projected VORP/Inning

1. Octavio Dotel      0.292
2. Ricky Stone        0.155
3. Chad Harville      0.150
4. Brad Lidge         0.139
5. Dan Miceli         0.129
6. Brandon Duckworth  0.098
7. Brandon Backe      0.049
8. Jared Fernandez   -0.042
9. Mike Gallo         No Projection

PECOTA, coming into the 2004 season, saw Dotel as
handily topping the field as the best Astro reliever.
I agree. Since becoming a full-time reliever in 2001,
Dotel has been among the game’s elite. The
aforementioned Lidge and Miceli place near the middle
of the pack. Although Lidge isn’t the caliber of
Dotel, I think PECOTA’s forecast for him–77 innings,
4.57 ERA–is a bit gloomy. Lidge showed poor control
last season, but he fanned 97 in 85 innings and did a
fair job of keeping the ball in the park considering
the environment. Still, he’s no Dotel. Miceli’s
projection, on the other hand (61.1 innings, 4.48 ERA)
seems about right. He had a nifty 2003, but the
preponderance of the evidence suggests he was pitching
over his head.

As for who’s garnering the most critical innings, I’ll
use run differential as the determinant. Below the
relievers are ranked by the average absolute value of
the run differential for the games in which they’ve
appeared. Allow me to parse that clunky syntax for
you…I’m using absolute value to eliminate the
problems that ensue when negative values are
introduced. For instance, without using absolute
values, a reliever who enters one game with his team
down by seven runs and one game with his team up by
eight runs would have an average differential of +1,
which would lead you to believe he’s been used in
high-leverage situations. By using absolute values of
run differentials, his average is 7.5, which more
accurately reflects the quality of his innings. In
terms of leverage, the lower the average differential,
the more important the appearances:

Pitcher (Appearances)      Avg. Run Diff.
1. Lidge (16)                 2.81
2. Miceli (17)                2.94
3. Dotel (15)                 3.00
4. Gallo (10)                 3.20
5. Backe (15)                 4.33
6. Duckworth (5)              4.60
7. Harville (8)               4.75
8. Stone (6)                  5.00
9. Fernandez (1)              9.00

As you can see, Lidge and Miceli are being used, on
average, in ballgames that are tighter than those in
which Dotel is appearing. That shouldn’t be the case
if the Astros are concerned with optimal usage. Over
the long haul, Houston would be well served to have
the run-differential list more closely resemble the
VORP list above, so that the better relievers are
working the crucial innings. Still, if Lidge and
Miceli continue outperforming projections, having them
pitch in situations of critical mass is certainly
defensible, but they still shouldn’t be ahead of Dotel
in the leverage queue.

For this next list, I’m classifying a “high-leverage
inning” as one in which the opposition has the tying
or go-ahead run at the plate or on base. It’s more
narrowly defined than the save rule, but that’s by
design. Below, I’m ranking Astro relievers by the
percentage of their 2004 appearances that qualify as
high-leverage, according to the definition I’ve just

Pitcher (Appearances)       High-Leverage %
1. Lidge (16)                  31.3
2. Dotel (15)                  26.7
3. Miceli (17)                 23.5
4. Duckworth (5)               20.0
5. Stone (6)                   16.7
6. Backe (15)                  13.3
7. Harville (8)                12.5
8. Gallo (10)                  10.0
9. Fernandez (1)                0

Dotel fares better than Miceli by this measure (as
does Harville), but Lidge remains the reliever of
choice (whether Williams realizes it or not–bet
liberally on the latter) when the game is most in
peril. To cast this another way, I’ll look at what
percentage of appearances came in “close game”
situations, which I’ll define as games in which the
score was tied or the Astros were up or down by a
single run:

Pitcher (Appearances)    Close Game %
1. Dotel (15)               33.3
2. Lidge (16)               31.3
3. Harville (8)             25.0
4. Miceli (17)              23.5
5. Duckworth (5)            20.0
5. Gallo (10)               20.0
7. Stone (6)                16.7
8. Backe (15)               13.3
9. Fernandez (1)             0

Finally, Dotel comes out on top. Exactly one-third of
his appearances are of this nature, but Lidge isn’t
far behind. Lest this sound like modest praise for
Williams, I’ll point out that Dotel should be
comfortably atop all three lists, not just narrowly
ahead on one.

Here’s a couple of examples of when and how Dotel
should’ve been used differently. While hindsight may
be 20/20, the idea is to identify suboptimal usage of
the Astros’ best relief pitcher, so that hopefully
it’ll happen less often going forward.

  • 5/3 – The Reds score two in the eighth off Lidge to
    take a one-run lead. Dotel would work the ninth (so he
    was available for duty) and surrender a homer, but if
    he’d been used to protect the one-run lead in the
    eighth, the outcomes are likely altered. Reds win,

  • 5/8 – The Braves tie score in the ninth off Lidge and
    beat Stone in the 10th. As detailed above, Dotel had
    been used in two of the previous three games to
    protect four-run leads in the ninth. Apportion those
    to lesser relievers, and Dotel would’ve been available
    to defend the one-run, ninth-inning lead May 8 and the
    one-run, ninth-inning lead the next night (which he
    was). Braves win, 5-4.

That’s only two games to date that meet this pair of
conditions: a) the Astros lost, and b) Dotel was
improperly deployed. On the other hand, if Williams,
in his use of Dotel, is going to bungle things this
badly twice for every 31 games, he could possibly cost
the Astros 10-and-a-half games in the standings by the
time the season is done. And that’s only the games
that the Astros lost. In others, Dotel has been used
unwisely, but Houston, sometimes by happy accident,
prevailed nonetheless.

Mismanagement of the middle innings is by no means
indigenous to Houston; it’s a widespread and
long-standing trend in baseball that’s underpinned by
the easily debunked notion that the game’s most
critical outs occur, ipso facto, in the final frame.
But that misunderstanding in tandem with the noxious
Jimy Save makes this perhaps the worst-run bullpen in
the game today.

From my purview, the Astros are the most talented team
in the Central and perhaps the National League as a
whole. I think they’ll go on to win the division, and
I even picked them to make the World Series. If those
things do come to pass, it’ll be tempting to conflate
team outcomes with managerial acumen. Don’t do it.
Williams may have other merits, but in terms of
division-of-labor in the bullpen, he’s striking a blow
against his team’s chances.