Out on the Data Frontier Where will the numbers game go in the future? Beyond hitting and pitching

April 04, 2004

In the winter of 2002-03, his first shopping season as general
manager of the Red Sox, Theo Epstein found himself competing
against Athletics G.M. Billy Beane for the services of an
injury-prone first baseman who was about to turn 28, who still
had not reached 50 RBIs or 250 at bats in a major league season,
and whom the Diamondbacks were eager to dump. Erubiel Durazo
hardly seemed a prized commodity. Both G.M.'s, however, coveted
him because of his efficiency statistics--a .390 career on-base
percentage and a .528 career slugging percentage--not to mention
his affordability. (He earned $375,000 in 2002 and $1.065 million
last year.)

Beane beat out Epstein for Durazo, constructing a four-team trade
with the help of his friend, former assistant and statistically
like-minded general manager J.P. Ricciardi of Toronto. (Beane
gave the Blue Jays a pitching prospect, Jason Arnold, in the
deal. Durazo, in his first full season, hit .259 with 21 homers
and 77 RBIs, and had a .374 OBP and a .430 slugging percentage.)
"All of us," Beane says of himself, Epstein and Ricciardi, "are
swimming in the same waters."

Those sabermetric waters will only become more crowded this
season, as teams such as the Dodgers, Mets, Indians and Cardinals
begin to rely more on quantitative analysis. As additional teams
take the plunge and once obscure statistics become mainstream,
where will G.M.'s mine next?

"You can keep looking deeper and deeper into [hitting and
pitching] statistics," Toronto assistant G.M. Keith Law says,
"until you start to bump against the wall in terms of what is
really useful. We're just about there. But I do think we can be
better in other areas as opposed to going deeper into what we
already know."

According to executives, these are the new frontiers of
quantitative analysis:

--DEFENSE Several statistical models already are in use to
quantify defensive ability, though none has gained universal
acceptance. "I'm working on one of my own," Law says. "The others
are O.K., but I haven't seen anything yet that really hits the
mark. I don't believe, for instance, that defense fluctuates
wildly from one year to the next, so I have a hard time believing
that Rich Aurilia can come up in one model as a minus-20 one year
and plus-five the next. A lot of it depends on the input data
being flawed. The base hit just out of the reach of the
shortstop, for instance. Was it hit hard, a single all the way,
or was it a three-hopper?"

--CHARACTER "There are probably some things we can learn from the
NFL," Ricciardi says. For instance, the pro football league has
long relied on personal interviews and the Wonderlic test--a
measure of intelligence, cognitive ability and problem-solving
skills--to evaluate prospective draft picks. The Rockies, Indians
and Orioles are among teams trying to identify high-character
players. Colorado, for instance, put together an exhaustive
report on high school third baseman Ian Stewart, which included
personality exams and lengthy interviews with his amateur
coaches, before taking him in the first round of the 2003 draft.
"We talk all the time in baseball about a player's ability to
handle failure," Rockies G.M. Dan O'Dowd says, "when a lot of
times in the big leagues you need to be more concerned with how a
player handles success."

Says Dodgers G.M. Paul DePodesta, "We have more than a thousand
players to consider for the draft. We're not going to run
psychological profiles on all, but we could do it with the top 50
or so."

--INJURIES At what point in their careers do catchers begin to
wear down? Do shortstops and centerfielders become less
productive in their mid-30s than players at corner positions? Are
pitchers who throw sliders more at risk of injury? Those and many
other questions related to managing injury-related investment
risk could be addressed by a kind of actuarial table created from
a database of injury information. "What's going on now,
especially when it comes to keeping pitchers healthy, is a lot of
guesswork," Law says. "We still don't know whether it's worse for
a pitcher to throw 130 pitches in a complete game or 100 pitches
over five innings."

--BALLPARK EFFECT Manager Whitey Herzog might have known
intuitively 25 years ago that ground ball contact hitters with
speed make for a good artificial-turf team in places like Kansas
City and St. Louis. Today's information can be more specific
about what types of skills play best in what parks, for example,
quantifying how a righthanded power hitter would fare in Boston's
Fenway Park, or measuring the value of a fleet-footed
centerfielder in a spacious hitter's park such as Colorado's
Coors Field. "A lot of work on park effect is already being
done," Epstein says. "I think it will become so commonly used
that fans will routinely know and accept it." --T.V.

COLOR PHOTO: JASON WISE (AURILIA) FLAWED DATA Quantifying defense, like Aurilia's, is still out ofreach.