Sox Watch

Sunday, April 30, 2006

Red Sox WPA through 4/30/06

And with this, the road trip mercifully comes to a close. Not a moment too soon.

Since it's the biggest positive WPA factor in today's game, this is a good time to provide a little more explanation on the "Opponents' Errors" category. In each game, the WPA for each team must sum to either 0.5 or -0.5 (depending on whether the team wins or loses). In most at-bats, the batter gets credit (either positive or negative) for the change in the win probability during his at-bat, and the pitcher gets equivalent credit in the other direction. Each at-bat is therefore a zero-sum transaction - if the batter gains, the pitcher loses, and vice versa. However, in the case of an error during a Sox at-bat, the batter does not usually deserve credit for reaching base. The error means that he should have been out, but reached only because of the error.

In a situation like this, when calculating the WPA, I treat the at-bat according to what would have happened had it not been for the error (which usually results in a negative WPA for the batter, and a positive WPA for the pitcher). Then I enter another transaction that accounts for the difference between what should have happened and what actually happened. The positive WPA for this transaction is credited to a catch-all category called "Oppenents' Errors", and the negative WPA is credited to the fielder on the opposing team who made the error.

The result of all this bookkeeping is that the category called "Opponents' Errors" ends up measuring the extent to which the Sox have benefited from errors made by the opposing team. Today, this was a particulary large category. In the seventh, Varitek reached on Burroughs' throwing error. If he had been out, Varitek would have received -0.035 WPA. Instead, reaching first in that situation (down by 2 in the 7th inning) results in a WPA increase of 0.056. The difference between the result that would have happened had the error not occurred and the actual result, which is 0.091, gets put into the "Opponents' Errors" category. Similarly, in the ninth, Harris stole second and advanced to third on the bad throw from Hall. Getting the tying run to second in this situation is quite important, and Harris gets 0.052 WPA for the steal itself. But the advance to third is even more important (there was only one out at that point), and worth an additional 0.097 WPA, and this 0.097 gets credited to the "Opponents' Errors" category.

Confused yet? It's really just a bookkeeping transaction to make sure all the numbers add up, but it's interesting (and depressing) to see that at this point in the season, the team has received substantially more benefit from their opponents' errors than they have from the combined performance of the whole offense.

Individual Player WPA Contributions
Sunday, 4/30/06
Red Sox 4, Devil Rays 5



Full-Season Player WPA Contributions
Through Sunday, 4/30/06



Full-Season Category WPA Contributions
Through Sunday, 4/30/06

Red Sox WPA through 4/29/06

Individual Player WPA Contributions
Saturday, 4/29/06
Red Sox 9, Devil Rays 6

Full-Season Player WPA Contributions
Through Saturday, 4/29/06

Full-Season Category WPA Contributions
Through Saturday, 4/29/06

Saturday, April 29, 2006

Red Sox RCI (Roster Change Index)

In the wake of the fairly extensive Red Sox personnel changes in the offseason, I've been wondering about historical levels of team continuity from year to year. Is this level of personnel shake-up higher than in most years? And if so, by how much? Like many fans, I've also bemoaned the increased mobility of players in recent years. How do current levels compare to the past?

I've come up with a rough system to try to quantify these measurements by examining historical statistics. It works like this:

For the offense, add up the total of all the plate appearances for a given season. The Red Sox had 6,476 plate appearances in 2005. Then look at how many of these plate appearances came from players that were gone by the following season. For the Sox in 2005, this was 3,826 (from Damon, Renteria, Mueller, Millar, etc.)

The idea here is to try to weight the loss of a player by his playing time - losing an everyday starter like Damon, who was up 700 times, is more significant than losing Roberto Petagine and his 36 ups.

For the pitching, perform a similar calculation, using batters faced by each pitcher. This parallels the offensive situation - you are essentially looking at how much you saw each pitcher over the course of the season. The Sox had 6,277 batters faced from their pitching staff in 2005, and of these, 2,437 were from pitchers who are now gone (mostly from Arroyo and Miller, but also Halama, Embree, and a host of others).

Totalling these numbers up, the Red Sox lost 49.3% of their 2005 players. I'm calling this value the Roster Change Index, or RCI. The Red Sox RCI of 49.3% for 2005 means that pretty close to half the players we saw in 2005 are gone in 2006. This seems high, but for historical perspective, the question to ask is how it compares to previous seasons. Using data from Retrosheet, I looked at Red Sox rosters going back to 1901 and performed this same set of calculations for each year. The results are shown below (note that you can click on the graph for a larger view).


Looking at the 2005 figure in historical perspective, we can see that it is well above the values in recent years, and in fact is the fourth-highest value in team history. The only three years that had higher levels of player change were the following:

  • 1945 (RCI = 58.4%) The conclusion of WWII saw the return of many prominent players who had been serving in the military, such as Ted Williams, Dom DiMaggio, and Joe Dobson. This displaced many of the players who had been playing during the war years, and so the 1945-1946 transition was the largest change in team history.
  • 1994 (RCI = 56.8%) The strike year. The strike began in August 1994, cancelled the second half of the 1994 season, the 1994 post-season, and the beginning of the 1995 season. This resulted in substantial shakeups to the lineups of many teams, and the Red Sox were no exception.
  • 1922 (RCI = 50.0%) The late teens and early twenties were a period of fairly high changes from year-to-year, with 1922 being the peak. Almost the whole pitching staff turned over in 1922, including Jack Quinn and Herb Pennock.

It's also interesting to note that although the RCI has been high in recent years, the levels aren't that different from other historical periods. The first decade of the team, along with the period from the late teens through WWII, saw rosters changing roughly at the same rate that they have been over the last ten years. The biggest long-term shift appears to be right around the 1994 strike, when we went from levels of roster change averaging around 20%, up to the current levels averaging around 35%.

A couple notes on the methodology: I'm not attempting to distinguish between players lost in the offseason and players lost during the season. So Damon (traded in the offseason to the Yankees) and Bellhorn (cut in the middle of the 2005 season) essentially fall into the same category - they appeared in 2005, but are gone in 2006. This makes sense, because the question we're trying to answer here is "How different is the 2006 team from the 2005 team?". Also, the 2006 season is just getting underway, so there is still the possibility that some players who were active in 2005 might still appear in 2006. For example, if Cla Meredith, currently with the PawSox, were to be brought back up, or if Bellhorn (perish the thought) were to be reacquired, then the total change number from 2005 to 2006 would go down, because these guys are currently being counted as "appeared in 2005 and gone in 2006".

Red Sox WPA through 4/28

An odd game indeed tonight. Clement was shaky from the start, and never really settled down. His control was abysmal, with six walks and lots of balls in the dirt, but he still managed to strike out four, and his four earned runs might have stood up if the Sox had not gone 2-for-19 with RISP. Varitek's "triple" will make the season highlight reel - Jonny Gomes might still be waiting for that ball to come down, with his arms in the air and a blank look on his face.

In WPA-land, Clement takes most of the heat tonight, while Lowell's GIDP as the potential tying run in the ninth further cements his overall position in the basement for the season. The "Greek God of Walks" was the top contributor tonight, going 1-for-2 with 3 walks.


Individual Player WPA Contributions
Friday, 4/28/06
Red Sox 2, Devil Rays 5

Full-Season Player WPA Contributions
Through Friday, 4/28/06

The starting pitching slide over the last week has now put the bullpen back on top of the heap as the biggest overall contributor this season.

Full-Season Category WPA Contributions
Through Friday, 4/28/06

Friday, April 28, 2006

Red Sox WPA through 4/27

This one was over before it started. Beckett gave up -0.267 in WPA in the first inning, and that was all the Indians really needed, since the Sox offense has apparently decided not to come along for this road trip. Manny's eighth-inning homer on Tuesday was the only thing preventing the first sweep of the season.

Individual Player WPA Contributions
Thursday, 4/27/06
Red Sox 3, Indians 15

Full-Season Player WPA Contributions
Through Thursday, 4/27/06

Full-Season Category WPA Contributions
Through Thursday, 4/27/06

Red Sox WPA through 4/26

Individual Player WPA Contributions
Wednesday, 4/26/06
Red Sox 1, Indians 7


Full-Season Player WPA Contributions
Through Wednesday, 4/26/06


Full-Season Category WPA Contributions
Through Wednesday, 4/26/06

Red Sox WPA through 4/25

Individual Player WPA Contributions
Tuesday, 4/25/06
Red Sox 8, Indians 6

Full-Season Player WPA Contributions
Through Tuesday, 4/25/06

Full-Season Category WPA Contributions
Through Tuesday, 4/25/06

Sunday, April 23, 2006

Red Sox WPA through 4/23

Finally today, an end to the three-day losing streak. Clement had a decent start, but the big story today was in the bullpen. Foulke's 1.2 perfect innings made him the WPA star of the game, while Timlin and Papelbon were also strong. Youkilis halted his mini-skid at the plate, and Lowell finally had a decent game.

Individual Player WPA Contributions
Sunday, 4/23/06
Red Sox 6, Blue Jays 3


Full-Season Player WPA Contributions
Through Sunday, 4/23/06


Full-Season Category WPA Contributions
Through Sunday, 4/23/06

Saturday, April 22, 2006

Red Sox WPA through 4/22

This was an all-around miserable game. DiNardo started out poorly, and things went downhill from there. And did I mention that the Yankees won, too? The only bright spot was Josh Bard's performance, going 2-for-2 with a walk.

Individual Player WPA Contributions
Saturday, 4/22/06
Red Sox 1, Blue Jays 8

Full-Season Player WPA Contributions
Through Saturday, 4/22/06

Full-Season Category WPA Contributions
Through Saturday, 4/22/06

Red Sox WPA through 4/21

A tough loss last night to the Blue Jays, the first Sox extra-inning game of the season, and their first one-run loss. Beckett was reasonably strong through seven and a third innings, earning a positive WPA despite giving up five earned runs. Papelbon and Foulke earned the most credit, both pitching multiple strong innings. Foulke took the loss, and gets a a bit of a WPA hit for walking Glaus in the twelfth, but Seanez of course bears the primary responsibility for the loss.

Individual player contributions from Friday night's game:

For the season, this is Papelbon's strongest WPA game yet (pitching 2.1 one-hit innings in a tie game will do that for you), and moves him into the top spot overall for the season.

Full-season player contributions through 4/21/06:


For the category breakdown, Friday's game continues the disturbing trend of divergence between the starters and relievers. Over the last five games, we continue to get strong starts, but the bullpen has not been as strong as earlier in the season.

Full-season category breakdowns through 4/21/06:

Thursday, April 20, 2006

Player WPA through 4/20

Player WPA totals for the season through Thursday's loss to the Devil Rays. Tonight was a bad night for just about everyone. Wakefield came away on the positive side, even though he took the loss, while Mohr's HR made him the biggest offensive contributor. Player totals for tonight's game:

Wakefield 0.156
Mohr 0.069
Error 0.036
Harris 0.013
Foulke 0.007
Loretta -0.008
Ramirez -0.037
Gonzalez -0.040
Varitek -0.044
Youkilis -0.052
Pena -0.064
Tavarez -0.072
Nixon -0.083
Bard -0.120
Ortiz -0.127
Lowell -0.135

Category WPA through 4/20

Current category WPA totals through the April 20 5-1 loss to the Devil Rays. Can't this team score some runs for Wakefield? Offense was clearly a problem tonight, while Wake gets some credit for eight strong innings allowing only two earned runs.

Player WPA through 4/19

Here are the current player WPA breakdowns, through Wednesday's Devil Rays game. The biggest contributors last night of course were Schilling and Youkilis, but just about everyone gained some points in this blowout. Loretta, Lowell, and Nixon were other big winners.

Category WPA through 4/19


Here are the current Red Sox WPA breakdowns, through Wednesday's blowout of the Devil Rays. This chart shows the contribution of Red Sox starting pitching, relief pitching, and position players. The data for position players includes their contributions on both offense and defense. My WPA tracking assigns most of the defensive WPA value to the pitcher, but I assign a portion to other fielders in the case of exceptionally good or bad plays (based on my subjective evaluation).


The current chart highlights the dominance of the starting pitching this year, with quality from the bullpen as well (the overall staff ERA of 3.76 currently leads the AL).

The offense has started to kick in over the last three games, finally crossing into positive territory last night, but clearly it has been the pitching that has been winning games so far this year.

Red Sox WPA Tracking

This year's hot statistic is WPA, or Win Probability Added. This statistic, as described at The Hardball Times by Dave Studeman, uses statistical analysis to determine the probability of each team winning at each point in the game, and gives players credit for the extent to which their actions raise or lower their team's chances of winning the game.

2006 looks like the breakout year for WPA, with hordes of people tracking WPA for individual teams. I began tracking Red Sox WPA statistics at the beginning of the season, with a goal of producing WPA graphs for each game. This effort was made obsolete this week, however, by the great work of the folks at Fangraphs, who are now producing automated WPA graphs for each game.

I'm now focusing instead on tracking aggregate WPA statistics for Red Sox players. Sox Watch will provide updated player total WPA statistics, along with summary-level stats, such as Offense vs. Defense and Starting vs. Relief Pitching.