Defensive Win Shares redux

Back in August, I suggested that the Lo Duca's defense might be overstated by defensive Win Shares. One test would be to see if his defense remained stable from the Dodgers to the Marlins. As it happens, he had 5.3 defensive Win Shares in LA and 1.8 in Florida. That's over 779 2/3 innings verus 413, which works out to 0.00680 WS/inn. vs. 0.00436. So it would appear that Lo Duca benefited from the Dodgers good defense.

Unfortunately, this is a really small sample. It's easy to imagine a catcher slowing down over the season and becoming less effective, which is just the reputation Lo Duca has. A better study would include lots of trades with catchers moving from bad defensive teams to good and vice versa. This still might not be enough.

The other way to look at the effect of the trade would be to compare team defense before and afterwards. The data isn't really available, however. (At least not without doing a lot of work.)

Open source risk for Oracle?

One of the things I worried about when I bought Oracle was "What happens when someone makes an open source version of their products?" I mean, why would anyone pay $149 a user for Oracle Database Standard Edition One, when they could get PostgreSQL free? Assuming they have the same functionality (which they don't), the answer is the cost of changing database software. Joel Spolsky has an excellent article explaining why companies might support open source software even to the point of paying developers to work on it. You should go read that article if you promise to come right back.

More informed? Good. Ok, Oracle is happy to have, for instance, open source file systems that are optimized for running their database. And it probably doesn't bother them too much to have open source "Application Servers", like JBoss, even though they offer a competing product. That's because they want complements to be cheap so they can charge more for their core product: the database. So why aren't they more worried about open source databases such as MySQL and PostgreSQL?

It's clear that Oracle has lowered their price on the very basic versions of the database in response to not just open-source products, but Microsoft and other low-cost providers. At the high-end, there has been price pressure on Enterprise Edition as well, but that doesn't have much to do with open source. Obviously Oracle would love to charge more for low-end products, but they aren't making the big money on small customers. Selling to small customers is a bonus (especially if they eventually become big).

From a strictly technical point of view, PostgreSQL is the only real open source competitor to Oracle. Unlike MySQL (and most commercial database systems), Oracle uses multiple versions of the data stored in the database, which means different applications can change and access the database at the same time without bring it to a standstill. On top of this base, Oracle has designed a system to spread the database across a cluster of computers. PostgreSQL doesn't yet have this ability, but it is only a matter of time before someone will make it happen.

In the meantime, Oracle is trying to diversify away from reliance on the database product, which is the purpose of buying PeopleSoft/JD Edwards. If you read SEC filings of many different companies, you've probably noticed charges labeled "SAP" that occur over the course of several quarters. Huge software/process upgrades like this are the market that Oracle is trying to tap into, and open source software is a compliment for these systems. This is doubly so if Oracle manages to build up it's outsourcing business.

As was brought out in the anti-trust trial, most of the applications companies buy are not all that difficult to duplicate. Basic accounting and human resources functions can be found in any number of products. More difficult (especially if not built on Oracle's database), is moving all of a company's information into one place. Vastly more difficult than that is creating a structure to hold all the data of any particular organization.

Your 2005 LA Dodgers

I recently finished Win Shares, so I got curious how the new Dodgers compare to the old Dodgers. First, the additions:

Player         Team   POS    Bat  Pitch Field  ExpWS    WSP   WSAA Total
J Drew         ATL    OF    29.4    0.0   4.8     17   .986     17    34
J Kent         HOU    2B    15.9    0.0   7.4     16   .711      7    23
J Valentin     CHW    SS     8.4    0.0   6.1     13   .551      1    14
R Ledee        PHI    OF     4.7    0.0   1.1      3   .819      2     6
D Lowe         BOS    P      0.2    5.4   0.0     10   .272     -5     6
P Bako         CHC    C     -0.2    0.0   1.6      4   .166     -3     1

That works out to be roughly 28 wins. Now, here's what they lost:

A Beltre       LAD    3B    31.2    0.0   6.1     18  1.031     19    37
A Cora         LAD    2B    11.9    0.0   5.1     14   .622      3    17
S Green        LAD    1B    13.7    0.0   2.4     18   .438     -2    16
J Hernandez    LAD    2B     7.5    0.0   2.3      7   .741      3    10
S Finley       LAD    OF     7.1    0.0   2.0      7   .673      2     9
S Finley       ARI    OF     6.7    0.0   2.9     12   .399     -2     9
J Lima         LAD    P     -1.1   10.6   0.0      8   .583      1     9
R Ventura      LAD    1B     5.3    0.2   0.7      5   .681      2     6
B Mayne        ARI    C      0.4    0.0   1.0      3   .242     -2     1
B Mayne        LAD    C     -1.4    0.0   1.4      3   .000     -3     0
H Nomo         LAD    P     -1.1   -5.4   0.0      4  -.808    -10    -6

That's 38 wins (or 36 if you give the Dodgers credit for losing Nomo). I included Finley and Mayne's stats with the Diamondbacks because if they had stayed Dodgers and contributed the same in 2005 as they did in 2004, the Dodgers would be beneficiaries of the entire season's statistics.

WSAA is a Hardball Times stat (Win Shares Above Average) that tries to correct for players who get a lot of playing time and some Win Shares, but hurt their team overall. The new additions increase wins above average by about 6 and the guys who are leaving add about 4. The discrepancy is largely due to guys like Shawn Green, who added 16 win shares but was expected to contribute 18. In other words, this off-season has been largely about replacing players who contribute less then they should for players who contribute more. The two biggest exceptions are Beltre and Lowe.

To quantify that a bit more, the new players were expected to get 63 win shares and actually got 84. The old players were expected to get 99 win shares and got 108. That works out to 0.667 Win Share rate versus 0.545. Assuming players do about the same they did last year, the new Dodgers should be a better than the old.

Here are the players who haven't moved (as far as I can tell):

C Izturis      LAD    SS    18.2    0.0   7.1     20   .642      6    25
M Bradley      LAD    OF    13.1    0.0   4.4     16   .537      1    17
E Gagne        LAD    P     -0.1   16.1   0.0      8   .979      8    16
J Werth        LAD    OF    10.5    0.0   1.6      9   .683      3    12
O Perez        LAD    P     -2.3   14.6   0.0      9   .659      3    12
J Weaver       LAD    P     -0.8   12.6   0.0     10   .573      2    12
G Carrara      LAD    P     -0.1    6.7   0.0      3  1.025      3     7
D Sanchez      LAD    P      0.2    5.7   0.0      4   .671      1     6
W Alvarez      LAD    P     -1.3    7.2   0.0      6   .491      0     6
K Ishii        LAD    P     -1.3    7.1   0.0      8   .356     -2     6
Y Brazoban     LAD    P      0.0    4.5   0.0      2  1.187      3     4
O Saenz        LAD    1B     3.8    0.0   0.3      3   .660      1     4
J Grabowski    LAD    OF     3.3    0.0   0.5      5   .398     -1     4
E Dessens      LAD    P     -0.1    1.7   0.0      1   .775      1     2
D Dreifort     LAD    P      0.0    2.2   0.0      3   .321     -1     2
M Kida         LAD    P      0.0    0.7   0.0      0  1.482      0     1
B Penny        FLO    P     -2.4   10.6   0.0      6   .662      2     8
B Penny        LAD    P     -0.1    0.8   0.0      1   .639      0     1
D Ross         LAD    C     -0.5    0.0   1.6      5   .103     -4     1
R Myers        LAD    P      0.0    0.3   0.0      0  1.482      0     0
J Flores       LAD    3B    -0.1    0.0   0.0      0  -.276      0     0
M Venafro      LAD    P      0.0    0.3   0.0      1   .295      0     0
A Perez        LAD    2B     0.1    0.0   0.1      0   .161      0     0
T Wilson       LAD    C     -0.2    0.0   0.1      0  -.356      0     0
S Stewart      LAD    P      0.0    0.3   0.0      1   .191      0     0
J Thurston     LAD    2B    -0.1    0.0   0.0      0  -.062      0     0
C Chen         LAD    OF    -0.6    0.0   0.2      0  -.879     -1     0
B Falkenborg   LAD    P     -0.1   -0.7   0.0      1  -.481     -1     0
E Jackson      LAD    P      0.2   -0.6   0.0      1  -.143     -2     0
H Choi         LAD    1B     0.1    0.0   0.3      2   .090     -2     0

The still-Dodgers had 146 Win Shares and were expected to get 125. When you add in the new players 84, you could estimate that the Dodgers will get about 77 wins. Ultimately, the Dodgers need to distribute more playing time to more efficient players in order to improve in 2005.

And for completeness, here are the guys traded during the 2004 season:

P Lo Duca      LAD    C     10.1    0.0   5.3     10   .735      5    15
P Lo Duca      FLO    C      4.6    0.0   1.8      6   .557      1     6
D Roberts      LAD    OF     9.1    0.0   1.4      7   .722      3    11
D Roberts      BOS    OF     1.4    0.0   1.0      3   .452      0     2
G Mota         LAD    P     -0.5    8.6   0.0      4  1.057      4     8
G Mota         FLO    P      0.6    2.3   0.0      3   .546      0     3
J Encarnacio   LAD    OF     6.1    0.0   1.6     10   .394     -2     8
J Encarnacio   FLO    OF     4.1    0.0   1.2      5   .522      0     5
T Martin       LAD    P      0.0    1.2   0.0      2   .352     -1     1
T Martin       ATL    P      0.0    1.2   0.0      1   .460      0     1