The following is not science...

Last season, the Dodgers won 85 games. Let's say next season, you want them to more. How does that happen?

Win Shares, the player evaluation system developed by Bill James and now employed in depth by BaseballGraphs.com, assigns three win shares for each team victory. For example, a team that wins 85 games, like the Dodgers did in 2003, will have 255 win shares. For each additional win, the Dodgers need a net gain of three win shares.

What follows is a list of how the Dodgers got their 255 win shares, divided by roster slot.

If a player was on the active 25-man roster for all or almost all of the season, he gets a roster slot to himself, be he Eric Gagne (25 win shares) or Ron Coomer (0 win shares.).

(Parenthetically, win shares are broken down in three categories - pitching, batting and fielding - with a series of complex calculations performed to determine how much credit each player gets toward these win shares. So a player can actually perform in a significant number of games without getting any credit for a win share.)

When the roster slot was split because a player was disabled or discharged, I did my best to combine the players who shared the given slot. For example, Jeromy Burnitz essentially replaced Brian Jordan, so they share a roster slot.

Los Angeles Dodgers 2003 Win Shares by Roster Slot
25 Eric Gagne
20 Kevin Brown
20 Shawn Green
19 Paul Lo Duca
17 Hideo Nomo
15 Adrian Beltre
14 Guillermo Mota
13 Alex Cora
13 Wilson Alvarez 10, Darren Dreifort 3
11 Cesar Izturis
11 Paul Quantrill
10 Brian Jordan 7, Jeromy Burnitz 3
9 Fred McGriff 8, Robin Ventura 1
9 Dave Roberts 8, Wilkin Ruan 1
9 Jolbert Cabrera
8 Odalis Perez 6, Edwin Jackson 2
7 Kazuhisa Ishii 6, Masao Kida 1
7 Paul Shuey
6 David Ross 4, Todd Hundley 2, Koyie Hill 0
5 Tom Martin
3 Steve Colyer 2, Troy Brohawn 1, Andy Ashby 0
2 Mike Kinkade
1 Larry Barnes 1, Daryle Ward 0
1 Rickey Henderson 1, Jason Romano 0, Chad Hermansen 0, Bubba Crosby 0
0 Ron Coomer

Okay, so how do we generate a net gain of 45 win shares?

First of all, where might we lose win shares? There are six roster slots at significant risk of decline, because of age and/or because their 2003 performances were so extraordinary: Gagne, Brown, Nomo, Mota, Quantrill and Alvarez/Dreifort (mainly thanks to Alvarez).

No surprise that all six are pitchers, huh?

How do we quanitfy how many win shares they might lose? By comparing them to other players. For example, if the Gagne 2004 slot becomes like a Tim Worrell 2003 slot - decent but no longer overpowering - it would fall from 25 win shares to 12, a decline of 13.

After studying the league stats at BaseballGraphs.com, I've determined, ever-so-roughly, that:

• The difference between a bad player and a mediocre player is five win shares.
  
• The difference between a mediocre player and a decent player is five win shares.
  
• The difference between a decent player and a very good player is five win shares.
  
• The difference between a very good player and an All-Star is five win shares.
  
• The difference between an All-Star and Eric Gagne is five win shares.
  
• The difference between Eric Gagne and Barry Bonds is, oh ... call it five win shares. Actually, no - it's more like 15 win shares. But until someone on the Dodgers starts hitting like the greatest hitter of many generations, let's let that go for now.

Brown slot (-10)
Gagne slot (-10)
Nomo slot (-5)
Alvarez/Dreifort slot (-5)
Mota slot (-5)
Quantrill slot (-5)
Total decline: (-40)

It's not that all these declines will happen, but they are declines that I think the Dodgers must prepare to face. If it helps, remember that it's three win shares to a victory. Can't you forsee Kevin Brown losing three wins worth of value, or Eric Gagne, who led National League pitchers in Win Shares this season?

Okay, let's turn to the roster slots with upside potential. If the Dodgers kept their team intact, here are the safe predictions for slots that would improve.

Jordan/Burnitz slot (+5)
Roberts/Ruan slot (+5)
Beltre slot (+5)
McGriff/Ventura slot (+5)
Green slot (+5)
Cora slot (+5)
Perez/Jackson slot (+5)
Izturis slot (+3)
Coomer slot (+3)
Henderson/Romano/etc. slot (+2)
Total gain: (+43)

I'm basing these numbers on these slots being young and still developing, or being able to rebound from disappointing seasons. If you combine the total gain above (43 win shares) with the total decline way above (-40 win shares), you come away with a net gain of three win shares, or one victory - pushing the Dodgers to 86 wins in 2004.

Now of course, this gets a little more complicated, because unless Frank McCourt replaces Dodger general manager Dan Evans with David Blaine, who then locks himself in a box above the Thames, the Dodgers will make changes to their roster.

Here's a sampling of the impact that some potential acquisitions might have. Keep in mind that Win Shares takes park effects into account, so we don't have to mentally adjust for players coming into the Dodger Stadium offensive graveyard.

Again, these are not recommendations - just examples:

• Richie Sexson, first baseman (26 win shares in 2003): If Sexson is acquired for minor leaguers, that's eight or nine wins right there, pushing the Dodgers to 95 wins in 2004. However, if someone like Paul LoDuca went in exchange, that mitigates the gain. You lose LoDuca's 19 win shares, but perhaps get five win shares more out of increased playing time for David Ross...

  Scenario 1: 26 win shares (Sexson) - 9 win shares (McGriff/Ventura) = 17 win shares = 6 wins

  Scenario 2: 26 win shares (Sexson) - 19 win shares (Lo Duca) - 9 win shares (McGriff/Ventura) + 5 win shares (Ross) = 3 win shares = 1 win

• Vladimir Guerrero, outfielder (18 win shares in 2003): Guerrero could potentially come as a free agent - no cost in major league talent - but note that the system puts his value below Sexson's, even though Guerrero plays a more valuable position. Anyway, Guerrero would bring 18 win shares at his 2003 value, but let's bet that he's still ascending and add another five win shares to his total, giving him 23. Guerrero would take the Jordan/Burnitz outfield slot.

  Scenario 3: 23 win shares (Guerrero) - 10 win shares (Jordan/Burnitz) = 13 win shares = 4 wins

• Miguel Tejada, shortstop (25 win shares in 2003): Tejada would presumably take the Alex Cora roster slot, with either Tejada or Cesar Izturis moving to second base. I'm not going to bet on Tejada improving offensively in 2004.

  Scenario 4: 25 win shares (Tejada) - 13 win shares (Cora) = 12 win shares = 4 wins

• Jeffrey Hammonds, outfielder (4 win shares in 2003): Hammonds, the scourge of BaseballPrimer.com, I include for two reasons - his Stanford affiliation, and the fact that while he was overvalued by some teams in his prime, guys like him may now be undervalued. Hammonds is the sample of a guy who could take the Henderson/Romano outfield detritus slot.

  Scenario 5: 4 win shares (Hammonds) - 1 win share (Henderson, etc.) = 3 win shares - 1 win

As I said at the outset, this isn't science. But I think this Win Shares exercise is a useful guide to the offseason. It allows you to quantify the Dodgers' potential improvement and decline, to gauge what kind of moves they need to make.

One trade might bring the Dodgers a gain of 5 wins.
One major free agent signing might bring the Dodgers a gain of 4 wins.
One minor free agent signing might bring the Dodgers a gain of 1 win.
A conservative look at the remaining Dodgers likely performances in 2004 indicates a gain of 1 win.

That's 11 wins. That takes the Dodgers from 85 victories in 2003 to 96 wins in 2004.

It's not science, but it's interesting.