Win expectancy is an approach to assess each player's impact on the game by measuring a team's likelihood of winning before a play and after, and assigning that change as credit or fault to certain players involved. You may have seen this at play in zillions of Hardball Times articles, or in the great graphs that jjjsixsix and Jeff at Lookout Landing put together, or in the mysterious Baby Winexp posts that appear from time to time.

Enough of this banter. A brief technical overview appears at the end, and you can search AN for "winexp" to see a number of past diaries on this topic.

Now for the A's performance (including Sunday) this week.

Top 10 Plays of the Week.

Apr 04, 2006 .  NYY v  OAK. Bot 9.1. R: 0. B:12. Marco Scutaro singled to deep left center, Bradley scored, Swisher to second (1.000, +0.272)
Apr 06, 2006 .  OAK v  SEA. Bot 2.0. R:-1. B:1. Carl Everett homered to right, Beltre scored (0.693, +0.219)
Apr 04, 2006 .  NYY v  OAK.  Top 6.2. R: 0. B:3. Jorge Posada singled to right, Rodriguez scored (0.650, +0.217)
Apr 04, 2006 .  NYY v  OAK. Bot 6.1. R:-1. B:. Eric Chavez homered to right (0.566, +0.214)
Apr 05, 2006 .  NYY v  OAK.  Top 3.1. R:+1. B:13. Gary Sheffield homered to left, Cano and Damon scored (0.849, +0.177)
Apr 03, 2006 .  NYY v  OAK.  Top 2.1. R:+3. B:123. Alex Rodriguez homered to left center, Cano, Damon and Jeter scored (1.000*, +0.177)
Apr 07, 2006 .  OAK v  SEA.  Top 5.2. R: 0. B:1. Mark Kotsay doubled to center, Kendall scored (0.615, +0.163)
Apr 09, 2006 .  OAK v  SEA.  Top 3.1. R: 0. B:12. Mark Ellis singled to right, Payton scored, Kendall to third (0.672, +0.156)
Apr 05, 2006 .  NYY v  OAK. Bot 8.0. R: 0. B:3. Jay Payton safe at first on second baseman Cano's fielding error, Bradley scored (0.905, +0.155)
Apr 09, 2006 .  OAK v  SEA. Bot 9.2. R:-2. B:23. Jose Lopez grounded out to shortstop (0.000, -0.152)

Player Win Expectancy Contributions of the Week

Name            WXC        Plays
Marco Scutaro     0.515    18
Joe Blanton*      0.477    25
Joe Kennedy*      0.28    13
J Duchscherer*    0.264    19
Mark Kotsay       0.251    30
Milton Bradley    0.207    31
Huston Street*    0.174    14
Rich Harden*      0.15    51
Jason Kendall     0.143    24
Eric Chavez       0.078    30
Kiko Calero*      0.048    16
Mark Ellis        0.032    27
Kirk Saarloos*    0.016    24
Antonio Perez     0.005    5
Adam Melhuse      0.003    4
Jay Witasick*     0.002    17
Brad Halsey*      -0.002    13
Nick Swisher      -0.059    21
Jay Payton        -0.078    19
Dan Haren*        -0.111    25
Bobby Crosby      -0.187    10
Barry Zito*       -0.203    35
Frank Thomas      -0.232    24
Esteban Loaiza*    -0.300    27
Dan Johnson       -0.402    22

To no one's surprise, Scooter is the hero of the week, followed closely by Cupcakes.  The goats of the week are DJ (poor guy) and Loaiza.

I wanted to focus on the performance of "RicarJoe RinKen", whom Baby Winexp counted as the third most valuable Athletic this week. By most accounts (including Baby W), Kennedy joined the A's with a flourish mid-season 2005 and then tanked for the rest of the season. Between Urban's note that he wasn't fitting into the clubhouse and the unfair comparison between his starting performance and our memory of the injured Harden, a number of us now have a gnawing sense of doom when they see Kennedy on the mound reminiscent of, well, you know. So why is Baby Winexp so high on JK?

 Joe Kennedy pitching:                
Apr 04, 2006.   Top 8 0  R: 0 B:    .  Jason Giambi walked  ( 0.554 ,  0.075 )
Apr 04, 2006.   Top 8 0  R: 0 B: 1   .  Miguel Cairo ran for Jason Giambi  ( 0.554 ,  0 )
Apr 04, 2006.   Top 8 0  R: 0 B: 1   .  Hideki Matsui reached on fielder's choice to first, Cairo out at second  ( 0.513 ,  -0.041 )
Apr 04, 2006.   Top 8 1  R: 0 B: 1   .  Jorge Posada struck out looking  ( 0.42 ,  -0.093 )
Apr 04, 2006.   Top 8 2  R: 0 B: 1   .  Matsui to second on passed ball  ( 0.469 ,  0.049 )
Apr 04, 2006.   Top 8 2  R: 0 B:  2  .  Bernie Williams flied out to left  ( 0.366 ,  -0.103 )
------------------                
 Joe Kennedy relieved Justin Duchscherer.                
Apr 05, 2006.   Top 8 2  R: 0 B:  2  .  Jason Giambi flied out to right  ( 0.366 ,  -0.103 )
------------------                
 Joe Kennedy pitching:                
Apr 05, 2006.   Top 9 0  R: -5 B:    .  Hideki Matsui flied out to center  ( 0.001 ,  -0.003 )
Apr 05, 2006.   Top 9 1  R: -5 B:    .  Jorge Posada grounded out to shortstop  ( 0 ,  -0.001 )
Apr 05, 2006.   Top 9 2  R: -5 B:    .  Bernie Williams grounded out to shortstop  ( 0 ,  0 )
------------------                
 Joe Kennedy relieved Kiko Calero.                
Apr 08, 2006.  Bot 8 1  R: -3 B: 1   .  Joe Borchard reached on fielder's choice to third, Bloomquist out at second  ( 0.04 ,  -0.031 )
Apr 08, 2006.  Bot 8 2  R: -3 B: 1   .  Ichiro Suzuki lined out to shortstop  ( 0.026 ,  -0.014 )
------------------                
 Joe Kennedy relieved Justin Duchscherer.                
Apr 09, 2006.  Bot 8 2  R: -5 B: 1 2 3 .  Raul Ibanez struck out swinging  ( 0.006 ,  -0.015 )

Well, first and most importantly, JK got the job done each and every time, and he didn't need anyone else to bail him out of the inning. Baby Winexp loves that. Secondly, Baby W doesn't mark off for the balls smashed just foul (see Apr 4 game), or even the leadoff walk or runners in scoring position (as long as they don't score by inning's end). And when assessing the past, that is the right thing to do; the balls went foul, no runs scored, the A's won. But when anticipating the future, it's hard to ignore those events. Therefore, let those of us with Kennedyphobia take a moment to appreciate that JK did get the job done this week. May future weeks be similar.

Finally, for context, a list of the major league leaders this week in Win Expectancy Contributed (WXC) and for kicks, the top plays of the week.

Team    Name            WXC     Plays

COL     Todd Helton     0.881   30
SF      Lance Niekro    0.794   20
NYM     David Wright    0.789   24
ATL     Edgar Renteria  0.755   36
ATL     Andruw Jones    0.751   34
 TB     M Hendrickson*  0.698   30
WAS     Alfonso Soriano 0.662   25
 BOS    Curt Schilling* 0.636   53
CHC     Derrek Lee      0.628   18
WAS     Jose Guillen    0.627   24
HOU     Roy Oswalt*     0.622   57
 CHW    Jim Thome       0.616   27
ATL     Adam A. LaRoche 0.609   29
MIL     Carlos N. Lee   0.601   26
 DET    Chris Shelton   0.583   25

Top 5 Plays of the Week

Apr 08, 2006 .  CHW v  KC. Bot 8.0. R:-1. B:1. Mike Sweeney homered to left, Grudzielanek scored (0.899, +0.509)
Apr 09, 2006 . ATL v SF. Bot 9.0. R:-1. B:. Lance Niekro homered to left center (0.649, +0.460)
Apr 05, 2006 . WAS v NYM.  Top10.1. R: 0. B:1. Jose Guillen homered to left center, Vidro scored (0.928, +0.451)
Apr 03, 2006 .  ANA v  SEA.  Top 9.2. R: 0. B:23. Orlando Cabrera singled to center, Izturis and Figgins scored (0.924, +0.424)
Apr 03, 2006 . ARI v COL. Bot 8.1. R:-1. B:1. Todd Helton doubled to left center, Barmes scored (0.693, +0.398)

---
Some Technical Details

Baby Winexp is a program I wrote to automatically compute win expectancy contributions from play-by-play data. The program assesses a team's winning chances by looking at the historical performance of every team in every game from 1979 to 1990 (data source here) for every combination of inning, score difference, and baserunners. (jjj66 and LL use a different model that is based on a mathematical model based on some historical data, which is why we get different numbers sometimes).

It's called "Baby" because it's immature and not fully debugged. One notable problem is that it cannot assign credit for outstanding fielding (as that information is not in the play by play data); one could manually adjust the data, but I don't have the time. Also, the results frustrate and annoy me at times, but ultimately I learn something from examining the data more closely.

The way, I believe, to use this measure is to see where its results don't match intuition, and then examine the play by play data more closely. It's the smoke in "where there's smoke, there's fire." As far as what the numbers mean, roughly speaking, changes of less than .070 are pretty routine, .100-.200 is a play you'd remember in the play recap, and over .250 are plays a fan would remember all month.

There is evidence (see old diary) that win expectancy correlates well with Win Shares Above Bench.