The New Ballgame: Understanding Baseball Statistics for the Casual Fan (20 page)

With no superior rating system and no consensus about how to evaluate defense statistically, a new book has hit on the novel idea of presenting as
much data as possible. In this way, The Fielding Bible seeks to provide an accounting of each fielder's strengths and weaknesses, relative to the skill of the
average player in a given position through an innovative plus/minus system.
Rather than a single rating built on debatable assumption, The Fielding Bible
presents a composite view of each player.

In the introduction to John Dewan's book, Bill James explains:

We are trying to ask specific questions about the player's defensive skills, so that, in time, we can create enough of an image
of the player's defense that we will be able to "see " intuitively
what we have left out. How well does this second baseman
turn the double play? How well does that third baseman field
a bunt? How well does that shortstop go to his left? How well
does he go to his right? ...It is my belief that the simple accumulation offacts, in time, will help us to understand more than
we now understand.

The Fielding Bible does this-and more-for each player and each
team. The basic stats (e.g. putouts, assists, errors, fielding percentage) appear
next to unique data: how each first and third baseman does on bunts; how
each middle infielder does on double plays; how all infielders do on balls to
their left, right and "straight on;" how strong outfielders' arms are (both in
throwing runners out and preventing advances).

By providing a three-year data set for each player, we begin to see certain patterns about a player's assets and his deficiencies.

The team profile includes a visual representation of where on the diamond all hits landed-balls down the line, between infielders, in front of outfielders, over outfielders, in the gaps, off the walls, over the walls. Looking at
Boston, for example, we can see the influence of Fenway Park-significantly
more balls were hit off the left field wall than the league average. We also can
see the influence of suspect Sox left fielder Manny Ramirez-significantly
more hits down the left field line and in the left field gap.

The data is the product of analyzing video of every play of the season.
So the Fielding Bible's own catch-all stat, a plus/minus system of "expected
outs," might be subject to challenge and debate, but it has the virtue of transparency-The Fielding Bible supplies the ammunition to attack (or support)
any of its findings.

Best of all, The Fielding Bible is forthright about what we do not yet
understand, and therefore remarkably free of dubious assertions like how
many hundreds of men Andruw Jones has thrown out at home plate.

ow that you have gotten this
far, soon you will be aware of
other statistics-some thought-provoking, some thoughtless. We are going to
draw your attention to several statistics that have gained momentum in recent
years-enough so that you might want to add them to your vocabulary once
you have become comfortable with the other statistics in this book.

But first, we offer this preface.

Curiosity about baseball statistics has been widening and deepening
each year. More people are interested and they want to know more than ever
before. Like scientific research, a discovery advances the quest for truth but doesn't end it. If anything, the discovery expands the possibilities for new
research, exposing passageways to be explored.

That said, sometimes the supply of new statistics exceeds the demand.

Wider and deeper databases, and new tools to analyze them, make research more plentiful. But just because we can track certain things, doesn't
necessarily mean we should. Consider this item that a wire service sent across
the country in late April 2006:

The New York Yankees joined the 1992 Toronto Blue Jays and
2003 Seattle Mariners as the only teams since 1976 to win
their first eight day games. Seattle started 17-0 in the daylight
that year The Yankees'next day game is today at home against
the Blue Jays.

Wow. Drop everything to make it to the stadium today or risk missing
the extension-or tragic end-to that streak of daytime wins. Talk about drama. Let's not mention that at this point the Yankees have played four straight
night games and, as usual, more night games than day games. I mean, if the
Yankees can win just nine more consecutive day games-check back with us
in June on that one-they'll be just the second team since way back in 2003
to win at least 17 day games in a row.

This sort of statistical noise isn't even worthy of the term trivia. Unfortunately, the search for the useful can be distorted by a cacophony of statistical
irrelevance. Whether researching statistics or studying them, consider this:

• Who cares? A good question to ask before writing, computing, researching, or experimenting on anything.

• Cause and effect. Are two or more simultaneous events linked by
anything more than coincidence? If we could prove that the manufacture of baseballs has changed at the same time that more home
runs are being hit, would that prove the ball is the cause? Not unless
we could isolate it from other changes like the strike zone, pitching
practices, and weather.

• Selective memory. Often used to advocate a player for the Hall of
Fame, this technique finds a stat or two that favors the player, then
argues, "Everyone else who has this many of that" is in the Hall of
Fame-ignoring that everyone else also has a lot more of the other
things missing from the first player's game. A variation on this theme:
Combine a couple of stats that favor the player with a couple that are
mediocre and declare, "My man is one of only 12 players in history to
achieve XX home runs, XX doubles, XX walks and XX stolen bases,"
in the hopes that the whole will look greater than the sum of the parts.
Invariably, none of the thresholds achieved is Hall-worthy.

• Lowest-common denominator. Another Hall of Fame argument:
"He has more home runs (or hits, or wins, or saves, etc.) than anyone
not in the Hall of Fame." If that was reason enough to elect someone
to the Hall, then as soon as that was done, the next guy on the list
would have to get in, and the one after him, and on and on until every
player was in the Hall.

Flaws like these generally betray the presenter's bias. Advocacy has its
own merits, but it's the natural predator of open-minded research.

Researchers like Bill James, John Dewan, the folks at Baseball Prospectus, and a few others have earned substantial and sustained followings
by inviting others along on their pursuit of understanding. The pursuit usually
starts by keeping in perspective the above points and always asking: Is this
information actually helpful?

Some of the statistics developed by the innovative researchers mentioned above have become widely quoted, though not so widely that you will
see them posted on scoreboards or cited on baseball broadcasts. These are
complicated enough that they are resistant to the abbreviation necessary in
those places. But if you continue to pursue your curiosity about baseball statistics, you will soon encounter these terms:

James developed this measure long ago to combine all of the major elements
of a batter's offensive performance. The Bill James Handbook describes Runs
Created this way: "an estimate of the number of a team's runs which are created by each individual hitter." That is followed by a page and a half describing the formula and how to tweak it. To simplify: It's the number of times a
hitter gets on base (minus his times caught stealing and double-play groundouts) multiplied by the effect of his hits, walks and sacrifices to advance other
runners, all divided by his plate appearances. The result is a powerful number
that can also be tweaked to show how often a team full of such offensive
players would score.

In 2002, Bill James and Jim Henzler published a groundbreaking volume that
presents in a single number the contribution made by any player-batter or
pitcher-to his team's wins and includes each player's offensive and defensive roles in that rating. The "short-form method" to calculate Win Shares is a
13-step process. The long-form method is...a book. So Win Shares is beyond
our ability to do it justice here. But James describes Win Shares as "Wins
Created" except that there are three "shares" to each win. Even if the math is
more than you want, the book is a treasure for all who want to delve into the
history of baseball. James has calculated and distributed Win Shares to the
players on every major league team in baseball from 1876 to 2001. Then he
created more lists by recombining all the players by position and by era. The
book has lots of numbers, but it has fascinating essays too.

We're not sure why the folks at Baseball Prospectus didn't call it Value Over
Replacement Player if they were going to use the "P" instead of the incompatible "L" in the acronym, because "player" works just fine to describe VORP.
Regardless, this stat measures how much better a player's performance was
than the typical bench player at his position. Many previous statistical measures compare performance to the average player, but VORP recognizes that
if a player gets hurt, or otherwise misses time, his replacement is not going
to be the average player-he's going to be somebody barely above minorleague level.

VORP is a handy number to have to compare value between two or
more players. If you are a big-league general manager considering a trade, VORP can compare your left fielder's relative worth to that of the other team's
pitcher. Or you can use the VORP ratings for all left fielders to find out how
difficult it will be to get a comparable replacement. If you are a simulation
gamer or fantasy player, you can use VORP to see which players give you the
widest competitive edge at their positions.

Runs Created, Win Shares, and VORP all provide clues about future value,
but they measure past performance. PECOTA (the acronym is the last name
of a former big-league utility infielder) is a forecasting system, not terribly unlike the weather forecasts that tell you there's a 25% chance of rain. PECOTA
takes a player's past performance, factors his age, his position, his ballpark,
and much more, and then compares him to every other player since 1949 to
find the best matches. This is used to project the percentage likelihood that the
current player will have a breakout performance, mere improvement, normal
"attrition," or a collapse in the forthcoming season.