Archive for WPA

Earlier this week I posted about the best bunts of 2011. Taking some of the comments to that post into consideration, the obvious follow-up is the worst bunts of the 2011 season according to Win Probability Added (WPA).

Read the rest of this entry »

…since 1974, when our play-by-play database begins.

On Monday, I posted about the three worst double plays ever according to Win Probability Added (WPA). From 1974 through yesterday’s games, there have been 89072 double plays caused by groundouts (I’m leaving out other sorts of double plays as, from the hitter’s perspective at least, they have more to do with dumb luck). Of those, only 51 have a positive WPA. However, it does happen from time to time, and while the shifts aren’t as dramatic in terms of WPA, the circumstances make them more interesting (at least to me) than the negative WPA occasions.

Read the rest of this entry »

In the course of a discussion at The Book Blog about the Angels’ (of late) recent outperformance of (some) projections, I was reminded of a related yet quite different issue I’d thought about looking into a while back (and then promptly forgot about). The Angels are one of the teams in baseball that are praised for “playing the right way” and “doing the little things.” Whatever people mean by that, one thing we can say is that recently, the Angels have consistently outperformed their Pythagorean Win Expectation. Looking (somewhat arbitrarily) at the last three seasons in which the Angels have won the American League West and comparing their actual record with what we’d expect given their run differential based on PythagenPat.

2007: Actual 94-68, Expected 90-72, difference +4
2008: Actual 100-62, Expected 88-74, difference +12
2009: Actual 97-65, Expected 93-69, difference +4

I should say right now that this post is not saying that I am not claiming either a) that the Angels “just got lucky” and weren’t as good as their record, or b) that they have some “intangible” ability (perhaps from their manager) that has enabled them to outperform their run differential the last three seasons. Both of those are copouts, at least at this point. For now, I’m only going to look at this issue with reference to their offense.

One might say that they’ve been “good in the clutch.” And that is, in fact, true. FanGraphs’ clutch score, which measures whether players outperform not only their peers, but themselves in high leverage situations, has the following win values for the Angels’ hitter from 2007-2009:

2007: 5.19
2008: 7.34
2009: 3.22

These numbers are impressive, but they sort of beg the question. Unlike relievers, hitters don’t “earn” their high leverage playing time — unless you think most of those scores were put up by Angels pinch-hitters picked for their “clutchness.” This seems to say what we already knew — the Angels won more game than their runs scored indicate that they “should have”. Undoubtedly, there are “clutch hits,” but this doesn’t tell us how they did it — just that they did.

One thing that “right way” teams are praised for is situational hitting. FanGraphs has a stat for that: RE24. While FanGraphs’ primary “runs created above average” stat, wRAA, uses the average change in run expectancy given an event irrespective of the base/out situation, RE24 does incorporate base/out state. For wRAA, a home run is a home run whether the bases are empty with none out or loaded with 2 out, while RE24 takes into account the different base/out run expectation. As I discuss here, if we subtract the average linear weight runs (wRAA) from the RE24, we can see how much better the Angels performed in terms of “situational hitting.”

2007: wRAA +7, RE24 30.5, situational +23.5
2008: wRAA -18, RE24 18.7, situational +36.7
2009: wRAA 88, RE24 92.8, situational +4.8

Impressive. However, it actually doesn’t tell us what we want to know. This tells us that we would expect the Angels to have scored more runs than traditional linear weights (wRAA) would suggest, but the Pythagorean expectation is already using their actual runs scored. We want to know why they outperformed their run differential (for now, from the offensive perspective) — not why they scored more than their linear weights suggest, but why they won more than their actual runs suggest.

Enter WPA/LI. While RE24 takes base/out context into account, WPA/LI goes one step further, by taking base/out/inning into account. You can follow the link to read up, but basically, it’s “unleveraged” Win Probability. It sounds like Clutch, but it’s actually WPA without the Clutch/Leverage element. To use an example to differentiate WPA/LI: with two outs in the bottom of the ninth with the bases loaded, for WPA/LI a walk and a home run have the same linear weight, whereas those events would be different for both wRAA and RE24, since they don’t take game state into account. So, if any stat could take into account a player or team adjusting their play to a situtation, this would be it. As I did in my earlier Little Things post for individuals, we can do for teams: convert wRAA to wins (I crudely divide by 10), then subtract that from WPA/LI to get the situational wins above average linear weights.

2007: wWAA +0.7, WPA/LI -1.32, -2.02 Little Things
2008: wWAA -1.8, WPA/LI -1.21, +0.59 Little Things
2009: wWAA +8.8, WPA/LI +6.37, -2.43 Little Things

Now that is just bizarre. With RE24, we saw that the Angels the last three seasons have been very good at maximizing their situational hitting in certain base/out states. But “Little Things” shows the exact opposite in 2007 and 2009. They’re about “even” in 2008, although far short of what RE24 shows, and they’re 2 wins below their traditional linear weights in 2007 and 2009. It’s not just that the Angels’ hittesr aren’t living up to their reputation (according to this measure) of “doing the little things,” it’s the contrast between RE24 and WPA/LI based “little things” that is striking. It’s as if the Angels do a great job of hitting with runners in scoring position when they’re playing in blowouts, but make terrible situational plays (relative to the average run expectancy) in close games. And then if you look at their hitter’s “Clutch” scores from those years… It’s really hard to know what the big picture is.

This post has no conclusion other than to note that the title is ironic. It would be foolhardy to dismiss this all as luck. The Angels have been a very good team no matter how you slice it. And just because we don’t understand “how they do it” at the moment doesn’t mean we can never know. But at the moment, I’m simply struck by the oddity.

Much has been discussed over the last few days regarding the lists I have brought forth involving single-game WPAs exceeding 1.0, as well as the top performances of the year via the same win probability metric. I wanted to clear something up: just because Jon Lester‘s and Carlos Zambrano’s no-hitters do not show up in the top ten, does not mean that these games were inferior to those on the list. To truly determine which games were the best, something as simple as the Game Score would be just fine. The point of the previous posts, though, was to show interesting occurrences of WPA and which performances in 2008 added the most wins to a team.

Some great points evolved in the comments thread of the two-part series evaluating the top ten pitching WPA performances of the season. It is very difficult for a pitcher to post a single-game WPA above 1.0, counting only pitching, unless he manages to log 12+ innings. As we saw last night, of the 13 pitchers to accrue a WPA above 1.0 since 1974, only two featured pitchers logging under 8.2 innings. In fact, one threw 10.2, two threw 11, and five threw 13!

Mark Mulder’s 2005 game was mentioned in the thread as well; in that game, Mulder tossed a 10-inning complete game shutout. He only managed to amass a 0.832 WPA, leading us to wonder which pitchers have come closest in recent history. Was Mulder the closest? Below are the top ten WPA performances from 2000-2008:

Roy Halladay    9/6/03   10 IP, 3 H, 0 ER, 1 BB,  5 K   0.936 WPA
AJ Burnett      6/26/05   9 IP, 2 H, 0 ER, 2 BB,  7 K   0.904 WPA
Jason Schmidt   5/18/04   9 IP, 1 H, 0 ER, 1 BB, 13 K   0.856 WPA
Brad Penny      4/4/02    9 IP, 4 H, 0 ER, 3 BB,  3 K   0.834 WPA
Al Leiter       4/18/02   9 IP, 2 H, 0 ER, 2 BB,  8 K   0.834 WPA
Johan Santana   8/12/05   9 IP, 3 H, 0 ER, 1 BB,  9 K   0.834 WPA
Mark Mulder     4/23/05  10 IP, 5 H, 0 ER, 0 BB,  5 K   0.832 WPA
Steve Trachsel  5/6/00    9 IP, 3 H, 0 ER, 3 BB, 11 K   0.818 WPA
Brad Radke      9/17/00   9 IP, 8 H, 0 ER, 1 BB,  4 K   0.815 WPA
Curt Schilling  6/7/07    9 IP, 1 H, 0 ER, 1 BB,  4 K   0.811 WPA

All of these were complete-game shutouts, with Schilling and Schmidt coming within pitches of a no-hitter. Schilling’s game was the one during which he shook off Jason Varitek before surrendering his lone hit at the end of the game. Mulder’s 2005 game may be the most recent wherein a pitcher logged ten or more innings while shutting the opponent out, but Halladay did the same thing two years earlier, earning about a full tenth of a win more than Mulder.

What happens when we incorporate the Game Score? Looking at the top game scores from 2000-2008, Schmidt’s performance in 2004 ranked ninth, with a 97 GSC, and Halladay’s 2003 performance ranked 97th. None of the others in the top ten via WPA finished amongst the top 100 game scores this decade. When evaluating single-game performances, the game score is likely the best way to do so, or at least the WPA/LI, but it is nevertheless interesting to see which games brought with them the most added wins to a team.

In case anybody out there has failed to notice, I have been particularly obsessed this month with WPA and interesting situations involving the metric. We have explored the ten best offensive plays of the season via shifts in win expectancy, the ten best pitching performances via single-game WPA, as well as instances when hitters and pitchers either exceed +1.0 WPA, or fall below -1.0, in a game. When discussing hitters who have been worth one or more wins in a single game, our focus was on Kurt Suzuki and Cody Ross, both of whom accomplished the rare feat in 2008.

There are, however, 35 other players who have done so at one point in their career, and it just felt natural to share these players and their great games. From 1990-2008, 19 players were so great in a single game that they actually contributed more than one win to their team. First, here are the players from the Y2K era:

6/20/08   Kurt Suzuki      1.002 WPA   4-5, 1B, 2 2B, HR, 5 RBI
6/7/08    Cody Ross        1.133 WPA   2-4, 1B, HR, BB, SB, 3 RBI
6/29/07   Mark Loretta     1.002 WPA   2-3, 1B, HR, BB, 2 RBI
9/7/05    Ryan Langerhans  1.115 WPA   3-4, 2 1B, 2B, BB, 3 RBI
6/11/04   Todd Helton      1.071 WPA   4-5, 1B, 2 2B, HR, 5 RBI
8/24/03   Brandon Inge     1.032 WPA   3-5, 2 1B, HR, SB, 3 RBI
8/21/00   Brian Daubach    1.273 WPA   3-5, 2 1B, HR, 4 RBI
5/10/00   Midre Cummings   1.023 WPA   1-2, HR, 3 RBI

Not exactly your standard list of all stars, eh? Sure, Cody Ross and Brandon Inge have power, and Mark Loretta has been solid offensively his whole career, but outside of Todd Helton this list is not all that impressive. And yet, these eight hitters put together arguably the top single-game performances of the whole decade. To top things off, Daubach’s game in late August of 2000 was over one-tenth of a win better than the next closest player. Did Brian Daubach really have the best game of the decade?

Now, let’s take a trip back to the 1990s, when Alanis Morisette and Hootie and the Blowfish hogged the radio, South Park began its first season, pogs were popular, and people knew who Jon Nunnally was:

4/8/99    Raul Mondesi     1.055 WPA   4-5, 2 1B, 2 HR, BB, 6 RBI
6/13/98   Travis Lee       1.036 WPA   3-5, 1B, 2 HR, 5 RBI
6/10/98   Dante Bichette   1.074 WPA   4-6, 1B, 2B, 3B, HR, 5 RBI
9/10/96   Steve Finley     1.063 WPA   4-5, 2 1B, 2B, HR, SB, 3 RBI
9/2/96    Mike Greenwell   1.029 WPA   4-5, 1B, 2B, 2 HR, 9 RBI
8/24/96   Fred McGriff     1.093 WPA   5-5, 2 1B, 2B, 2 HR, 4 RBI
6/11/94   James Mouton     1.005 WPA   2-4, 2 1B, BB, 2 RBI, 3 SB
8/12/91   Barry Bonds      1.103 WPA   2-4, 2 HR, BB, SB, 4 RBI
5/10/91   Roberto Alomar   1.042 WPA   3-4, 1B, 2 HR, 2 BB, 2 RBI
4/16/91   Dave Henderson   1.082 WPA   5-6, 2 1B, 2 2B, HR, 5 RBI
6/23/90   Dwight Evans     1.147 WPA   3-5, 1B, 2 HR, 3 RBI

Okay, so yes, Dante Bichette’s game was a cycle, let’s clear that up first. And, yes, Mike Greenwell knocked in nine runs on that fateful September 1996 day. Fred McGriff is the only player not to be retired. Also, Dwight Evans, whose game actually led all 1990-1999ers, appears to be the least interesting, at least relative to the stats posted next to the WPA. Which brings me to the next point: Evans’ WPA of 1.147 is 0.126 wins, behind Daubach’s 1.273. Did Brian Daubach really have the best game offensively from 1990-2008!?

Via WPA, which counts certain plate appearances as worth more than others, due to the clutchiness factor built in, it appears so: Brian Daubach had the best offensive game relative to shifts in win expectancy over the last nineteen seasons.

Over the last two weeks, I have used WPA, win probability added, to discuss the top ten offensive plays of the 2008 baseball season. The merits of this list could be debated to death, I am sure, but what cannot be denied is that these ten plays brought with them the biggest shifts in win expectancy. Hitting a three-run home run in the sixth inning, when down by a run may seem like a monumental play while watching the game in question, but a two-run walkoff home run with two outs in the bottom of the ninth inning is likely going to vault a team from a 9-11% win expectancy to the 100% victory.

With that in mind, here is the top ten in its entirety:

10) Jason Michaels hit a two-run home run in the bottom of the tenth inning against the Cardinals on July 12. The Pirates trailed 10-6 entering the bottom of the ninth inning, and won 12-11 in ten innings. His play was worth a shift of 78.6%.

9) Nick Swisher, who didn’t make his first plate appearance until the 11th inning, hit a walkoff two-run homer in the bottom of the fourteenth against the Tigers on August 5. The home run was worth a 78.8% shift in expectancy.

8) Nate McLouth delivered a three-run home run in the top of the ninth, on April 14, off of Takashi Saito and the Dodgers. Unlike the other plays on this list, McLouth’s 79.6% shift in win expectancy was not a walk-off hit.

7) Nicknamed “the first half highlight” by, well, me, Josh Hamilton hit a walk-off dinger off of Francisco Rodriguez prior to the all star break. The home run, which occurred on July 9, saw a shift of 82.9% in win expectancy.

6) This was probably the oddest finish to any game this season. The Giants led the DBacks on September 10, 3-2, as the top of the ninth began. Chris Young hit a two-run triple to give the Snakes the 4-3 lead, before Eugenio Velez hit a two-run triple of his own to win the game for the Giants in the bottom half. Velez’s triple brought with it an expectancy shift of 83%.

5) On June 5, Jason Giambi and his moustache provided an 89.6% shift in win expectancy when he blasted a two-run walkoff home run off of BJ Ryan and the Blue Jays.

4) Two days later, on June 7, Cody Ross of the Marlins provided the fourth biggest play of the season, worth 89.8%, when he lined a two-run homer with a 1-1 count off of Francisco Cordero.

3) On May 2, Pat Burrell hit a two-run walkoff home run off of Giants closer Brian Wilson, an expectancy shift of 89.95%.

2) June 29, Ronnie Belliard delivered a two-run walkoff home run off of George Sherrill in an interleague battle for Maryland supremacy. The dinger brought with it an expectancy shift of 90.05%.

1) Finally, on July 12, David DeJesus hit, you guessed it, a walkoff two-run homer, off of Brandon Morrow. His home run, 90.36% worth of WPA, was the biggest play of the 2008 season.

And there you have it, the top ten offensive plays of the season. Brian Wilson appeared twice on this list: he gave up Young’s two-run triple in the top of the ninth in #6, before being bailed out by Velez, and gave up the walkoff to Burrell in #3. #1 and #10 both occurred on the same day, July 12. Three days prior, Hamilton hit his much shown home run off of K-Rod, meaning three of the top plays of the entire season occurred within four days of each other.

The Fireman of the Year Award is a rather meaningless award designed to promote some sponsor of major league baseball while simultaneously honoring the game’s best closer or relief pitcher. While the Academy Awards have the Razzies, this Fireman award does not have a reverse award designated to honor the league’s worst relievers. Yesterday, we took a look at the five worst starting pitchers of 2008, and today we will look to those in the bullpen. While WPA/LI was used to evaluate the starters, I decided to use both the leveraged and unleveraged WPA to check out the relievers. Here are the bottom five relievers via WPA/LI:

Brad Hennessey, -1.36: 13 g, 17.1 IP, 40 H, 8 BB, 12 K, 2.77 WHIP, 12.46 ERA
Jamie Walker, -1.36: 59 g, 12 HR in 38 IP, 1.68 WHIP, 6.87 ERA
Joel Peralta, -1.23: 40 g, 1.33 WHIP, 5.98 ERA, 15 HR in 52.2 IP, 14 BB, 38 K
Bob Howry, -1.23: 72 g, 1.46 WHIP, 5.35 ERA, 13 HR in 70.2 IP, 13 BB, 59 K
Gary Majewski, -1.19: 37 g, 1.90 WHIP, 6.53 ERA, 15 BB, 27 K in 40 IP

From these five, it is pretty remarkable that Hennessey accrued such negative impact in just 13 games, which portends that he might be able to dethrone Brandon Backe as the worst pitcher in baseball. Walker has given up plenty of home runs relative to his innings pitched, and Majewski does not really have anything positive working for him either. Peralta’s walk and strikeout numbers look solid, as he is right around the corner from a 3.00 K/BB, but he has given up a ton of home runs in just 52.2 innings. Howry’s K/BB is over 4.00, but he gives up a lot of hits and home runs. Next up, the bottom five by WPA:

Jason Isringhausen, -2.96: 42 g, 1.64 WHIP, 5.70 ERA
Luis Ayala, -2.85: 81 g, 1.45 WHIP, 5.71 ERA
Jamie Walker, -2.41: SEE ABOVE
Mark Lowe, -2.24: 57 g, 1.76 WHIP, 5.37 ERA
Masa Kobayashi, -2.19: 1.42 WHIP, 4.53 ERA, 2.50 K/BB

Two things should stand out from these five pitchers. First, Jamie Walker is the only reliever to bottom out in both WPA and WPA/LI, and secondly, Kobayashi does not look that bad. He has 14 walks and 35 strikeouts in his 57 appearances, but his hits allowed have raised the WHIP. He is actually the only reliever of the nine listed here with a sub 5.00 ERA. ERA might not be the best evaluator for relievers, but his controllable skills look better than most of the rest.

So, I will leave the votes in your hands for the Anti-Fireman of the Year Award. Does it go to Jamie Walker, who showed poor skills whether the plate appearances counted as one or more plate appearances? To Hennessey, who stunk it up so badly in just 13 games that he actually posted the worst WPA? Or maybe Majewski, who had literally nothing positive working for him or to fall back on?

Over at the Inside the Book: Blog a week or two ago, TangoTiger discussed a method in which a pitcher’s “effective win-loss record” could be calculated using primarily the +WPA and -WPA components found here at Fangraphs. For all of the gory details, click here. For more of a calculation summation, this modified record can be produced by doing the following:

1) Double the +WPA and -WPA, otherwise known as Win Advancement and Loss Advancement
2) Regardless of the + or -, add the WA and LA (if a pitcher has a +8.50 +WPA and -6.50 -WPA, he would have 17 WA and 13 LA)
3) This new number gives us the GA, Game Advancements
4) Multiply the GA by .308
5) Subtract the product from #4 from the WA and LA to get the ‘Effective W-L’

With the exporting feature on the leaderboards I was able to enter this into a spreadsheet to calculate the effective w-l record of all those who qualify. Via winning percentage, the top five pitchers in this category are:

1) Ryan Dempster, .823 (12.3-2.6)
2) Cliff Lee, .813 (14.1-3.2)
3) Tim Lincecum, .724 (13.2-5.0)
4) Justin Duchsherer, .705 (9.6-4.0)
5) CC Sabathia, .691 (14.0-6.3)

Ian Snell clocked in with the lowest winning percentage, a .328 spawned from his 5.8-12.0 effective record. To view the spreadsheet I used and see everyone’s winning percentage, click here.

Whether or not this metric will find itself ingrained in the general saberstream or not is yet to be determined but it definitely serves the purpose of leveling out a pitcher’s record. By adjusting it to be indicative of the positive and negative shifts in win expectancy we are really looking at how much a pitcher helped his team win or lose.

Back at the end of April, Dave posted his First Month Heroes, a position by position look at the highest WPA/LIs through one month of play. In the comments section I had joked that it would be even more interesting to see the inverse, or the worst context-neutral performers in the same span. My wish became Dave’s command and the following players surfaced: Josh Bard, Adam LaRoche, Robinson Cano, Ryan Zimmerman, Troy Tulowitzki, Garret Anderson, Andruw Jones, Jose Guillen.

Two full months later I thought it again to be prudent to check on these low-level offensive performers. So, as of July 8, the worst WPA/LIs by position are:

C: Kurt Suzuki, OAK, -0.84
1B: Daric Barton, OAK, -1.33
2B: Freddy Sanchez, PIT, -2.56
3B: Melvin Mora, BAL, -1.15
SS: Jason Bartlett, TB, -1.71
LF: Emil Brown, OAK -1.20
CF: Michael Bourn, HOU, -1.67
RF: Jeff Francoeur, ATL, -1.31
SP: Bronson Arroyo, CIN, -1.94
RP: Brad Hennessey, SF, -1.26

Yes, Frenchy was demoted to the minors, and yes, it took me three attempts at spelling his name correctly, but he has been the least productive rightfielder on offense. After him it gets a little hazy since certain guys who may be classified as rightfielders played different positions (SEE: Gary Matthews, Jr, and Mark Teahen).

One interesting part of this group is that their WPAs, which could benefit from some game state bias, are not all the lowest at their position. In fact, the only players listed above with both the lowest WPA/LI and WPA are Suzuki, Barton, and Frenchy. Sanchez, Bartlett, and Brown have the 4th worst WPA at their respective positions; Bourn has the second worst; and Mora actually ranks at the halfway point in WPA terms.

The other aspect of this group that piqued my interest is that three Oakland Athletics position players are, as of this moment, the worst win-contributors at their respective positions. In terms of WPA, not WPA/LI, the most productive offensive total is the 0.80 of Eric Chavez. Overall, though, their offense has cost them six wins; thankfully their starting rotation and bullpen have combined to the tune of ten added wins. Perhaps this is why some As fans were calling for offense in return for Harden. Regardless, I’ll be very curious to revisit this closer to the end of the season not only to see how these players improve or digress, but whether or not any of them remain.

One of the best parts of this site is the accessibility of David Appelman and his willingness to improve and/or update the site to feature more statistics and new parameters for those numbers. The newest addition to the Fangraphs statistical team is calendar year rankings. By going to the leaders page you can now sort not only by month or last 7/14/30 days, but also by the last 1, 2, or 3 calendar years.

For instance, did you know that Ryan Howard, with 145 home runs, has the most in the last three calendar years? Or that Alex Rodriguez ranks second, with 131, fourteen less than Howard?

How about the best and worst WPA counts for hitters in this same span?

BEST
1) Albert Pujols, 18.68
2) Lance Berkman, 17.49
3) David Ortiz, 17.29
4) Vladimir Guerrero, 12.57
5) Ryan Howard, 11.88

WORST
5) Jose Lopez, -3.59
4) Yuniesky Betancourt, -3.73
3) Jack Wilson, -3.86
2) Brandon Inge, -4.24
1) Ivan Rodriguez, -4.93

Hmm. Of the worst five contributors over the last three calendar, two are from the Tigers and two are from the Mariners. In terms of context-neutral wins (WPA/LI), Pujols and Berkman switch places; Berkman’s 17.34 comes in ahead of Pujols’s 16.69.

How about starting pitchers and WPA?

BEST
1) C.C. Sabathia, 9.06
2) Johan Santana, 8.92
3) Roy Halladay, 8.89
4) Brandon Webb, 8.67
5) John Smoltz, 7.91

WORST
5) Dave Bush, -1.94
4) Carlos Silva, -2.79
3) Livan Hernandez, -3.46
2) Jason Marquis, -3.64
1) Matt Morris, -7.83

Wow. Numbers 2 and 3 combine for -7.10 and Morris still comes in 7/10 of a win worse than them. In terms of WPA/LI, Johan reclaims his spot atop the throne with a 10.77, a full 1.60 wins ahead of second-place Brandon Webb’s 9.17. Santana also has the best K/BB (4.45) in this span, as well as the highest LOB% at 78.7%.

It has been reiterated recently that instead of using current seasonal statistics to evaluate players it would be much more accurate to use a rolling projection. While these calendar statistics do not necessarily weight the past any differently they do allow us to see which players have been good enough recently so as to trounce atypical poor early performance.