|
|
Mariners vs. Yankees |
Mets vs. Cardinals
|
|
Tuesday, October 10
Should you always play by "the book?"
By Michael Wolverton
Special to ESPN.com
Editor's note: The team of writers from the Baseball Prospectus (tm) writes for ESPN.com Insider. You can check out their web site at baseballprospectus.com, where you can vote for the Internet Baseball Awards.
Armando Rios is going to have a long winter.
The last big play of the 2000 season for the Giants' outfielder ended
with him heading back to the dugout with his head down, about to
become the subject of nationwide criticism. Rios, you may remember,
was on second base with one out in the bottom of the 10th of Game 2
of the Mets-Giants Division Series, with the Giants down by a run.
When Bill Mueller hit a grounder to shortstop, Rios tried to take
third and was thrown out fairly easily.
The verdict was immediate and unanimous: Rios made a serious
baserunning blunder. After all, baseball's "Book" says you never try
to advance in that situation on a grounder hit in front of you. The
Fox announcers jumped on Rios as soon as the play was over. The
ESPN.com recap of the game called his decision "inexplicable". Rios
himself called the decision a "mental mistake". Even as I write this
several days later, San Francisco radio talk show hosts and callers
are still mentioning the play in their series post-mortem.
But was it really such a huge mistake? As far as I know, my
Baseball Prospectus colleague Joe Sheehan is the only analyst
who has questioned
the conventional take on that play. Joe pointed out that the
worst possible outcome of Rios's decision was the loss of only a
single base: a runner on first with two outs instead of a runner on
second with two outs. But the potential upside of the gamble was
huge. If the Mets infield had failed to get Rios at third -- always a
possibility with a non-force play -- the Giants would have had runners
at first and third and only one out. That's obviously a vastly better
situation for the Giants than either of the two-out situations.
Joe's reasoning makes a lot of sense to me, but you can imagine
someone making a case the other way. For example, maybe having the
runner on second instead of first is a big deal since he can score on
a single. So let's see if we can get a more definite answer about
baserunning strategy in these situations by running some numbers.
We'll do this by starting with the run expectation table. That table
tells us, for each of the 24 possible bases/outs situations, the
average number of runs that score from that situation. Here are the
numbers for the 2000 season:
Bases Outs
0 1 2
empty 0.57 0.31 0.12
1st 0.97 0.60 0.27
2nd 1.18 0.73 0.33
1st, 2nd 1.63 1.01 0.48
3rd 1.52 1.00 0.41
1st, 3rd 1.92 1.24 0.52
2nd, 3rd 2.05 1.50 0.64
1st, 2nd, 3rd 2.54 1.70 0.82
For example, this season when teams had a runner on second and two
outs -- the situation the Giants would have had if Rios had held --
they scored an average of 0.33 runs in the remainder of the inning.
The other two situations of interest in this discussion are runner on
first with two outs (0.27 expected runs), and runners on first and
third with one out (1.24 expected runs). These numbers confirm Joe's
argument: the cost of Rios's gamble was small, and the potential
benefit was huge.
We can evaluate Rios's decision by using these numbers to find a
break-even point for the strategy. In other words, how likely would
Rios making third have to be in order for the gamble to be worth it?
Plugging the numbers into a simple inequality (I won't show the math
here), we come up with a break-even point of 7 percent: if Rios can reach
third base safely more than 7 percent of the time on a play like that, then
he made the right (i.e., run-maximizing) decision.
You might argue that the Giants didn't care about maximizing their run
total in the 10th inning down by one run. They cared primarily about
scoring just one run, the one that would tie the game. After all,
that's why they sacrificed Rios to second in the first place. We can
evaluate Rios's decision with respect to the goal of scoring the tying
run by looking at a different table. This one shows the probability
of scoring at least one run given the 24 bases/out situations:
Bases Outs
0 1 2
empty .30 .18 .08
1st .45 .29 .15
2nd .64 .42 .22
1st, 2nd .66 .44 .24
3rd .86 .67 .28
1st, 3rd .88 .66 .29
2nd, 3rd .85 .71 .29
1st, 2nd, 3rd .89 .69 .34
This table tells us that an average team will score at least one run
15 percent of the time if they have two outs and a runner on first, 22 percent of
the time if they have two outs and a runner on second, and 66 percent of the
time if they have runners on first and third with only one out.
Plugging those numbers into an inequality, we now get a break-even
point for Rios's decision of 15 percent. That is, if Rios thinks he can make
it to third 15 percent of the time or more on a play like that, then his team
is more likely to tie the game if he takes the risk.
That leaves the question of whether Rios had at least a 15 percent chance of
making third on that play. There's room for disagreement here, but I
sure think so. Even though that particular play looked fairly
routine, there's really no such thing as a routine throw-and-tag play.
If any one of a number of things go wrong -- the throw hits the
runner, the throw isn't good enough for a tag to be made, the runner
slides around the tag, the runner knocks the ball out of the glove,
etc. -- the Giants would suddenly have been looking at a
better-than-even chance of tying the game, and the real possibility of
going to New York up 2-0. In short, I think Rios' decision not only
wasn't the major blunder it's widely considered to be, it was a smart
strategic move, whether he knows it or not.
These kinds of controversial baserunning decisions occur every day
throughout the summer, but during October they have the potential to
make or break an entire season. My favorite example of costly
over-conservative baserunning comes from Game 7 of the 1991 World
Series between the Braves and Twins, the classic scoreless duel
between Jack Morris and John Smoltz that the Twins eventually won in
the 10th inning.
Many people remember the "Lonnie Smith decoy play" in the eighth
inning, where Smith failed to score from first on a long Terry
Pendleton double, possibly because the Twins infielders pretended to
have the ball as Smith rounded second. But few remember the amazing
play that happened next. At that point, the Braves had Smith at third
and Pendleton at second with no outs. They had to have been thinking
that one run would likely win the game, the way Smoltz was pitching.
But when the next hitter, Ron Gant, dribbled a weak grounder to first,
Smith held at third base.
I don't know whether it was Bobby Cox's decision not to have Smith
running on contact in that situation, or whether Smith made the
decision on his own, but it was a dumb decision that probably cost
Atlanta a world championship. Put yourself in Cox's position before
the play: if you send Smith on a ground ball and he gets thrown out,
you still have a runner on third plus Gant on first. In other words,
even assuming the worst possible outcome of sending Smith, the Braves
really aren't significantly worse off than they would have been if
they'd held him.
Running the situation through the second table above yields a
break-even point of less than 15 percent -- if Smith can score 15 percent of the
time or more on a ground ball, the right strategy is to have him
running on contact. I figure a speedy runner like Smith could have
scored much more than 15 percent of the time on a random grounder, even with
the infield in. And he almost certainly would have scored on the weak
dribbler that Gant actually did hit.
One last comment about the flip-side of the Rios play: If Rios took a
good gamble by trying to advance to third, it pretty much follows that
Mets shortstop Mike Bordick took a bad gamble by throwing to third to
get Rios. He would have been better off taking the sure out at first.
That questionable decision paid off for Bordick, but the Oakland A's
made two similar fielding decisions in Game 3 of their Division Series with
the Yankees that may have cost them the game and the series. In the
second inning of that game, A's pitcher Tim Hudson tried to get Bernie
Williams at home on a high chopper instead of taking the sure out at
first. He failed, and it cost the A's a run when Paul O'Neill scored with
two outs later that inning. In the fourth, A's catcher Ramon
Hernandez tried to get the lead runner on a sacrifice, and the throw
got past shortstop Miguel Tejada. That also cost the A's a run. They lost the
game by two runs; if they'd won, they'd be playing Seattle now.
So what's the lesson here? It's that outs are precious, much more
precious than extra bases, and both baserunning and fielding strategy
should be based on that fact. Fielders should generally be
conservative. They need to be taught and retaught the maxim: "When in
doubt, take the out. And when you're not in doubt, reconsider and take
the out anyway." On the other hand, baserunners could stand to be
more aggressive. Specifically, on those plays when they're not
risking an additional out, they would often benefit their teams by
trying to take the extra base (possibly tempting the fielder to make a
risky play).
And finally, managers, players, and even we armchair analysts could
stand to apply some skepticism and common sense before blindly
accepting the strategic recommendations of "The Book."
Michael Wolverton may be reached at mwolverton@baseballprospectus.com.
|
|
 |
|
