                     2004 Projection Disk Notes
                     --------------------------

April 9, 2004


Table of Contents
-----------------

Introduction
The players
Ballparks
Projected statistics
Playing time
Real-life injuries
Settings
A few technicalities
The April update


Introduction
------------

Thank you for ordering the 2004 Projection Disk for Diamond Mind Baseball.  
For those of you who are new to our projection disks, we'd like to take a 
little time to talk about how we put it together and many of the fun 
things you can do with it.  

First, however, we'd like to talk about what it CAN'T do for you.  To put
it simply, it can do a lot of things, but it can't predict the future.  No
matter how many times you play out the 2004 season with this disk, you'll
never really know how the 2004 season will turn out -- whether someone 
will take another run at the homerun record, whether another Japanese
league star will take center stage the way Ichiro did in his rookie year,
which rookies will shine and which will fall flat, who will make the 
playoffs, or who will go on to win the series.  Why not?

Because there's a ton of uncertainty in baseball.  Nobody knows which 
players will have career years, and for those who do, whether it was a 
one-time event (Brady Anderson hitting 50 homers) or a persistent change 
in their level.  Nobody knows which post-Tommy-John-surgery pitchers
will return at 100% or better.  Nobody knows who, if anyone, will get
hit by a pitch on opening day and miss the next six weeks.  Or which of
the older players will put up one more good year and which will be
released in May because they just can't get around on the fastball
any more.

This part of baseball -- the unknown -- is a big part of the reason 
why we love to read the boxscores and team notes every day.  Because 
almost every day there are new stories, some inspiring, some tragic, 
but almost always interesting.  We're glad we can't predict the 
coming season, because that would take the fun out of it.

But there are a lot of things you CAN do with this disk.

You can get a good look at how a player is most likely to perform for the 
team he will play with in 2004.  It is well established that most players 
will continue to perform as they have in the past if you adjust for their
age.  But it's often hard to see that, because when players move to new 
leagues (DH effect), new parks, or new levels (AA -> AAA -> majors), their 
numbers change, even if their performance remains pretty much the same.  And 
stats fluctuate from year to year even if the player doesn't change venues
Because our projections are based on three years of data, including minor-
league results, and because they're adjusted for all of these effects
(league, park, competitive level, and age), you can get a good view of
their established level of performance.

You can evaluate the claims of your favorite team's general manager.  One of
our pet peeves are GMs who make a couple of mid-level free agent signings and 
then anoint their club as a contender.  Well, many clubs are more than one 
or two players away from contending, and they're either ignorant of that 
fact or they're lying about it.  Playing out the coming season can tell you 
whether the team's moves have made it better, and if so, by how much, 
assuming each player performs to his established level.

You can play what-if games.  Because Diamond Mind Baseball gives you the 
ability to modify players and change rosters, you can make your own 
assessments about potential player moves or injuries.  

You can also get a handle on how much a player's new teammates might affect
his stats.  Our projections show the effects of aging and changing leagues, 
parks, and levels, but they don't tell you how many RBI they'll gain by 
moving to a better hitting team or changing to a new spot in the batting 
order.  Only by playing out one or more seasons can you find that out.  And 
if you play in a fantasy baseball league, you might gain a small edge this 
way.

You can see how the minor-league stats of some key prospects translate into
expected performance levels in the big leagues.

And you can have fun playing games using rosters that are more current
than those on our 2003 Season Disk, without having to track all the player
moves yourself, and without having to enter the 2004 schedule.  For many of 
you, that will be reason enough to use this disk.

Now, because we know that many of you are interested in learning more about
how we came up with our projected stats and ratings, we'll get into some 
of the details.


The players
-----------

The projection disk includes over 1700 players, including a few hundred
prospects who have never appeared in the majors but who have a decent shot
at making their debut this year.  In truth, some of these guys aren't
really prospects, but we included them because they've played in the big
leagues before.  As you will see from their projections, some of them 
haven't played well enough in recent years to have a legitimate shot at a
big-league job.

You'll see that some players appear in the free agent pool.  Some of those 
players are interested in playing but haven't caught on with a team yet.
Others are players whose careers are probably over but who could show up 
on a roster this year.  A few have recently announced their retirements, 
but we wanted to have them around just in case they change their minds.

We could easily have added another 200 players, guys who have had a taste
of the big leagues but haven't established themselves, at least not in the
last couple of years.  Many of these players are a little older and can no
longer be considered prospects.  But they're still hanging around the
fringes of the big leagues, providing injury insurance on AAA rosters, and
getting invitations to spring training.

We left these fringe players out because they haven't done well enough in
the minors and in their brief big league visits to earn a good projection
for the coming season.

And there are a few hot prospects who get a lot of ink but are not included
on this disk because they haven't yet played enough above A-ball.


Ballparks
---------

The park ratings are based on three-year averages (or less if the park was 
added or substantially renovated in the past three years), except for the 
new parks that will be opening this year and the parks that have been 
modified for the coming season.  

The new parks this year are in Philadelphia and San Diego.  In addition,
Kansas City's park has been returned to its old dimensions.

Because most parks don't change from year to year, we don't ship new park
images with our season disks.  But you can download any images you don't
already have from our web site at no charge.


Projected statistics
--------------------

We've received some questions about how our projections are different 
from those produced by others.  Indeed, they are different in many ways, 
partly because we needed more and different information than the 
others are providing.

When we started doing projections in 1998, we thought about licensing the
projections from one of the other providers so we could focus on the game-
related aspects of the projection disk.  But this proved to be impractical 
for a number of reasons:

- we needed projections for over 1700 players, and many of the other 
sources don't do that many

- to support a full-blown simulation, we needed to project many more 
statistical categories than the others provide

- we needed to project left/right splits as well as overall totals for all
batters and pitchers

So we decided to build our own projection system.  Actually, we expanded on
a system that we originally developed in 1994.  When that season came to a 
premature end because of the player strike, the TOPPS baseball card company
hired us to simulate the missing games so they could produce CyberCards with
full-season stats (real life stats through August 11 and simulated stats for
the rest of the season).  To their credit, they wanted to include prominent
minor-leaguers (Derek Jeter was one) who would have been called up in 
September had the season continued.  So we developed a method for projecting 
major-league performance from minor-league statistics.
                           
The basic ideas for this projection system come from Bill James.  He was the
first person to demonstrate that it was possible to predict major-league
performance from minor-league stats, and he wrote quite extensively on the 
subject in the 1985 Baseball Abstract.  In the same book was a detailed 
explanation of his system for adjusting performance for a player's age.

Our methods are based on Bill's ideas, but we have gone beyond a simple
implementation of his formulas.  The advances in data collection and 
computing power in the intervening period have made it possible to do more.

Here are the key ideas that make up our system:

- we use both minor-league and major-league statistics from the past three
seasons, ensuring that virtually all players have a large amount of playing
time on which to base the projections

- we use both AA and AAA statistics, and adjust both to their major-league
equivalents.  Bill's published formulas cover AAA adjustments only, so we 
created our own AA adjustments.

- all stat lines are evaluated with respect to the league averages.  This
does two important things.  First, it makes sure that stats from hitter-
friendly leagues are suitably deflated and those from pitcher-friendly
leagues are also adjusted to normal levels.  Second, it ensures that
pitchers who faced the DH and those who didn't are evaluated properly.
The DH adds roughly a third of a run per game, and if one doesn't take 
this into account, NL pitchers would be rated better than their AL
counterparts of equal ability.

- all stat lines are adjusted for ballpark effects, including the minor-
league parks.  The published Bill James methods do not take minor-league
park factors into account because they weren't available at the time.

- recent performances are weighted more heavily.  What a player did last
year is more important than what he did three years ago.

- performances at higher levels are weighted more heavily.  What a player
did in the majors is more important than what he did in AA ball.

- stat lines with more playing time are weighted more heavily.  If someone
batted .375 in 24 atbats, that doesn't matter nearly as much as what he
did in 400 atbats at some other stop along the way.

- the individual league- and park-adjusted stat lines are averaged (using
the weights just discussed), then age-adjusted to produce a set of projected
stats that are league- and park-neutral.

- these neutral projections are then applied to the league and park in which
the player will compete in 2004.

- the overall stats are converted into left/right splits based on each 
player's composite splits for the past three seasons.

- the distribution of fly balls and ground balls allowed by pitchers is 
based on actual ratios for the past three seasons.

That's the essence of the system.  The other projection systems we looked at
make some of these adjustments, but we're not aware of any that make them 
all.  And we think it's necessary to make them all in order to evaluate past
performance correctly.               


Playing Time
------------

One thing to keep in mind is that rates matter and absolute numbers don't.
It doesn't matter whether we project a player for 300 atbats or 500.  The
simulation only needs to know that for each 100 atbats, the player should
get 16 singles, 4 doubles, and so on.  It's up to you to decide whether he
plays a lot or a little.  Or, if the computer manager is handling things,
it's up to the manager profile to allocate playing time.  

We could have scaled the projections to 600 plate appearance for all hitters
and 1000 batters faced for all pitchers.  Or 500 and 800.  It doesn't really
matter.  But it would look funny if we did it this way.  The team totals 
would be way out of normal ranges, and some of the numbers on the league
reports might overflow the space allotted to them.

So we wrote a program that figures out from the manager profiles how many 
games each player is slated to start at each position, how much he'll
play versus left- and right-handed pitchers, and how much he'll play in a
platoon or reserve role.  Also, by looking at the saved lineups, we can
see whether he will be batting near the top of the order and therefore
getting more trips to the plate.  With this information, it projects a
total number of plate appearances and divides them among left and right.
And it does something similar for pitchers.

The net result is a set of playing time projections that resemble a normal
team, with some guys playing a major role and some guys getting only a few
appearances.  For players with a limited role, we established minimums of 
75 plate appearances and batters faced.  We know some of these guys won't
play at all and some might only get ten atbats in the season, but if we 
scale the stats down to ten atbats, we end up with rounding errors.

For example, if a player is supposed to get two triples per 100 atbats,
he ends up with zero if we scale his line to ten atbats.  By keeping the
minimum at 75, we can retain the basic shape of a player's stat line.  It
means that most teams have more than a full season's worth of playing
time, but that's okay.  It's not over by too much, it doesn't hurt
anything, and it helps retain accuracy.

Please keep in mind that 75 plate appearances is still a small number,
especially when they are divided into left/right splits.  A player
with 75 PA might get only 50 PA versus righties and 25 versus lefties.
If a player is projected to get 7.53 hits in those 25 lefty atbats, we
round it up to 8.  The difference between 7.53 hits and 8 hits is
.47 / 25 = .019, or 19 batting average points.  So don't be alarmed if
you look at the left/right stats for one of these players and they seem
a little high or low.  It's probably just the effect of rounding the
underlying projections.

The manager profiles are set up to allow for a normal amount of rest and
injury.  If you play the season many times, most starting position
players will play about 155 games and your rotation starters will make
28-33 starts.  In any one season, they may play a little more (if they
avoid injuries altogether) or a lot less (if they are seriously hurt
one or more times).


Real-life injuries
------------------

Certain players have injuries that will cause them to miss all or part of
the 2004 season.  There were two ways we could have represented this
knowledge on the Projection Disk.

One option was to enter a set of DL/Activate transactions that would have
placed these players on the DL before opening day and restored them when
they were due to return.  This would be the best option except that it's 
impossible to turn on DMB's injury system if your league is in real-life
transaction mode.

Fortunately, there is another way to limit playing time for these players,
and that is to give lots of spot starts to other players at their positions.
This isn't a perfect solution because the projected starter will end up 
starting some games in April and sitting out games later in the season, 
but it's a good way to make sure he doesn't play too much or too little 
overall.

This solution applies when you want to autoplay the entire season.  If
you're managing the games yourself, you can promote and farm the players
at any time in order to keep the rosters set up the way you want.

If a prominent player is expected to miss the entire season, we usually
leave him on the disk so you could play some what-if games that show 
how their clubs might have fared if he was available to play.


Settings
--------

We've set the injury rule for the Inter-league organization to Random,
since we believe it's impossible to project future injury rates.  You
can, of course, change this to None if you don't want injuries or to
"Injury rating" if you prefer to use the injury ratings.  Keep in mind,
however, that all players currently have the same (Normal) injury rating,
so using the "Injury rating" setting won't distinguish among the players
unless you choose to modify their injury ratings first.

The use of real-life transactions and saved lineups has been turned off,
since we cannot know those things until the real-life season is played.

The "Limit bench playing time" option has been turned off.  That option is
designed to ensure that the computer manager doesn't overuse batters who
put up big numbers in a limited amount of real-life playing time.  Because
players on this projection disk are rated based on three years of playing
time, there are no players with unrealistically high statistics.  So there's
no need to restrict how the computer manager uses the players.

The compilation of game-by-game statistics is turned on.  You can turn this
off if you don't plan to use the game-by-game reports or generate reports
based on time intervals.  If you turn this off, your autoplayed seasons will
run faster and you'll conserve a few megabytes of hard disk space.

Generation of game accounts is turned off, so if you're the commissioner of
a league based on this projection disk, don't forget to turn that on so your
managers can send their game results back to you.

The pitching charts in the manager profiles have been created with Strict
5-man rotations.  Time mode is inappropriate for a projection disk because
that mode is designed to match real-life playing time, and the real-life
season hasn't yet been played.

Because of inter-league play, we have prepared saved lineups and depth
charts for both DH and non-DH games.  In all cases, they are in 
Game-by-game mode.  Track starts mode is inappropriate for a projection 
disk because it is designed to match real-life playing time, and the 
real-life season hasn't yet been played.

All manager tendencies have been set to Neutral, since we don't yet know
how the managers will behave in the coming season, and Neutral settings
tell the computer manager to play the percentages and make decisions based
on the skills of the available players.

We have generated playing time limits for all players based on their
projected playing time, but you can ignore those limits.  They are used
only when your pitching chart is in Time mode or when you are using the
"Limit bench playing time" option.  You shouldn't be using either of those
features with a projection disk.


A few technicalities
--------------------

Most stats are dependent on the league and park in which they are compiled,
and we've already discussed how we take this into account.  But there are 
quite a few stats that are dependent on other things as well, and we'd like
to discuss some of them here:

- projected runs and RBI for batters are based on historical averages.  This
isn't a bad start, since they are adjusted for league and park effects.  But 
you're better off looking at the computer-league R and RBI totals, because 
they will take everything into account -- the player's projected performance 
(getting on base, hitting for power), batting order position, and the
quality of the other hitters in the lineup.

- pitcher runs and earned runs allowed are projected by applying the Bill 
James runs created formula to the projected rates of hits and walks 
allowed.  This enables us to base our ERA projections on the league- and 
park-adjusted stats and to cope with the volatility of real-life ERAs, 
especially for relief pitchers (for whom ERAs can be quite out of synch with
the rest of the pitching stats because they work so few innings per season).

- pitcher wins and losses are projected based on their ERA and the
assumption that all pitchers will get average run support (for their team)
and average help from their bullpens.

- pitcher saves are also projected based on ERA.  Closers with similar roles
are assumed to get the same number of save opportunities, with the quality 
of the pitcher determining how many are converted into saves.

- most players have been assigned the same ratings they had in 2003, with 
a few exceptions.  Error rates, passed ball rates, wild pitch ratings, balk
ratings, and ground ball percentages are based on three-year averages.
We've added defensive ratings for players who are being moved to new
positions or who will resume playing a position they once played.  We
adjusted the ratings for some players who played hurt last year and are
expected to be back to normal this year and for some older players whose
skills are declining.  The better Clutch and Jam ratings were given only
to players who earned them more often than not in the past three years.

- all players have been assigned the Normal injury rating, in the belief 
that past injuries are often not a good indicator of future injuries.  Yes,
there are players who seem to get hurt over and over, but there are others
who have a history of injuries and suddenly go several years without 
getting hurt (Bagwell, for example).  More importantly, there are large
numbers of players who have no history of injury who will get seriously
hurt this year anyway, and we don't know of any way to predict who they
are.  So we've rated everyone the same, and we believe that's realistic.

- we have not tried to project fielding stats, mainly because it's not 
necessary.  The range ratings, error rates, throwing ratings, and 
passed ball ratings control all of the fielding aspects of the 
simulation.  The fielding stats are there for reporting purposes only.


The April update
----------------

As promised, we have released this updated Projection Disk about ten
days after the regular season began.  

If you have started a league season using the first edition of the 
disk, there isn't an automatic way to merge the updates into the 
season that is underway.  This means you have three choices:  

(a) ignore these changes and continue your season, 
(b) start the season over using the updated disk, or 
(c) import the new players, manually add the new ratings that 
      are listed below, and continue your season.

Here is a summary of the changes:

1.  We updated the rosters to reflect trades, retirements and other moves,
and we updated the manager profiles to reflect our growing understanding
of the roles each player is expected to play this year.  These changes
are too numerous to list here.

2.  If a player made the opening day roster but was not part of the first
disk, we added him.  Here are the eight players who were added:

  Ari -- Casey Daigle
  Bal -- Jose Bautista
  ChN -- Mike Wuertz
  KC  -- Shawn Camp
  Mon -- Chad Bentz
  NYA -- Donovan Osborne
  Pit -- Mike Johnston
  SF  -- David Aardsma

3.  We made the following schedule corrections:

  Boston @ Baltimore moved from 4/5 to 4/4
  Detroit @ Toronto moved from 4/8 to 4/5
  Minnesota @ Detroit moved from 4/9 to 4/8
  Anaheim @ Seattle moved from 4/5 to 9/13

4.  We do not change player ratings or event tables between the first disk
and the update except to correct any mistakes or to add ratings for
players who have been moved to a new position or pitching role.  (By the
way, there is no need for us to update the event tables for players who
were traded because our event tables are stored in a form that is league-
neutral and park-neutral.)

The following pitchers were given a relief durability rating:

  Darren Dreifort (LA), Wayne Franklin (SF), Kevin Jarvis (Sea),
  Gary Knotts (Det), Jose Lima (LA), Rodrigo Lopez (Bal),
  Damian Moss (Tam), Shane Reynolds (Ari), Nate Robertson (Det),
  Aaron Sele (Ana), Amaury Telemaco (Phi), Justin Wayne (Flo)

The relief durability ratings were changed for the following pitchers
because they are expected to play a different role in the bullpen:

  Jung Bong (Cin), Steve Colyer (Det), Brian Meadows (Pit),
  C.J. Nitkowski (Atl), Matt Perisho (Flo), Andy Pratt (ChN)

The following pitchers were given a starter durability rating:

  Erik Bedard (Bal), Steve Sparks (Ari), Tyler Yates (NYN)

The following position players were assigned ratings at a new position:

  Bos -- Cesar Crespo rated as a shortstop
  Cle -- Alex Escobar rated as a center fielder
  LA  -- Jason Grabowski rated as a first baseman and left fielder,
         Dave Roberts rated as a left fielder,
         Jayson Werth rated as a left fielder
  Mil -- Keith Ginter rated as a right fielder
  Min -- Michael Cuddyer rated as a second baseman
  StL -- Roger Cedeno rated as a left fielder,
         Hector Luna rated as a second baseman
  Tex -- Kevin Mench rated as a right fielder,
         Eric Young rated as a left fielder

Jose Valentin has announced that he is giving up switch hitting.  We 
chose to leave him as a switch hitter on this disk because (a) his 
event tables were generated with that assumption, (b) he says he will
continue to switch hit against certain left-handed pitchers, and 
(c) it's quite possible he will abandon this experiment if he 
doesn't get off to a good start.

Now that the April update has been released, we will not be issuing any 
further updates as the season progresses.




