The Cory Lidle accident/tragedy/developing story is a poor time to air my grievances with Baseball Think Factory (actually, not Jim Furtado and the fine people who run the site but many of the people who post there). Of course, that never stopped anyone posting on BTF either:
But of course it's not that simple. I make fun of players who scabbed in '95, but most of them were too naive to have really thought about the implications of what they were doing. It's certainly possible that Lidle, and Rick Reed, and Kevin Millar and all the rest are perfectly fine people from a moral standpoint, just a little dim. So there's a difference between being a scab because you don't understand about unions and dignity and human rights and all, just because you want money now instead of later, or you're afraid to pass up an "opportunity," and being a scab after having considered and understood the implications. It's the same difference between killing someone who attacks you and killing someone who's at home in bed at the time.
So maybe Lidle was a jerk who spit in the face of dozens of millions of hard-working American families, or maybe he was a naive, silly kid. I have no idea which, so I certainly feel no concrete emotion but sadness about his untimely death. But I'd be lying to say that I feel as bad about it as I would if it was someone who I knew was a truly good person. People have differing value to history, which is a taboo thing to say, but absolutely true, and something that has to be faced.
This is a very sick man. And the degeneration into politics (of a leftist bent, mostly, since that's the view of most of the posters) on every thread, along with the general jackassery, makes it pretty much intolerable.
Now you might say I'm a hypocrite, that I'm taking advantage of Lidle to complain about BTF. And I suppose you may be right. But I've never seen a better, more timely illustration of why I hate reading that site.
Wednesday, October 11, 2006
Why I Don't Visit BTF Much Anymore
Internet Baseball Awards: Cy Young
Last year, neither pitcher who I chose for the Cy Young won the actual award. In the NL, Chris Carpenter won, while I favored Roger Clemens (although certainly Carpenter had a strong case). In the AL, Bartolo Colon won the Cy on the strength of his wins, while Johan Santana was again the AL’s top starter.
And now the AL mistake grows in magnitude, because Santana should now be polishing his third trophy. There is little doubt that he will win all polls this year, as he won the pitching triple crown. Had Roy Halladay not been felled in September by an injury, he actually would have given him a run for his money. Santana pitched 233 innings with a 3.05 RA (+82 RAR); Halladay 220 with a 3.26 (+72). The peripherals all favor Santana, but Halladay was once again outstanding as well.
Beyond them, it gets a little murky. The next three pitchers on the RAR leaderboard are Chien-Ming Wang (+57), Curt Schilling (+53), and Barry Zito (+53). All of them have shortcomings in the peripheral measures; Wang’s GRA is 4.64, Schilling’s eRA is 4.62 with a 4.04 GRA, and Zito’s eRA is 4.76. Wang fares decently in eRA (4.07)--his poor showing in GRA is likely due to his groundball tendencies. eRA as I am figuring it now incorporates actual doubles and triples allowed, while GRA assumes that balls in play will become hits at an average rate and the type of hit will be average as well. Likely due in large part to his groundball ability, Wang gave up the third-lowest Isolated Power on balls in play in the AL. Schilling managed to be “run lucky” but also “hit unlucky”, allowing a .327 %H, and thus I see his performance as a washout and revert to the run-based measure.
After them is Francisco Liriano (+53), brilliant in a brief campaign. After them is John Lackey and the previously discussed rookies Weaver and Verlander.
I have to whittle the ballot down to five; and I have not mentioned the two relievers who had dominant seasons, BJ Ryan and Jonathon Papelbon. Giving each of them a little extra credit for high leverage justifies moving them past the second tier of starters, and into third and fourth on the ballot. Which spot for each though? Ryan worked more games (65 to 59) and more innings (72 to 68). Papelbon leads in RA and ERA (.91 to 1.33 in ERA and a similar margin in RA). They are very similar in terms of inherited runners per game (.45 for Ryan, .41 for Papelbon). Both were very effective at keeping them from scoring, although more so for Ryan, bringing his RRA to .41 versus .52 for Papelbon. Ryan leads in eRA (1.29 to 1.52) and they are in a dead heat in GRA. I give the edge, ever so slightly, to Ryan. And so my AL Cy ballot falls into place:
AL Cy Young
1) Johan Santana, MIN
2) Roy Halladay, TOR
3) BJ Ryan, TOR
4) Jonathon Papelbon, BOS
5) Chien-Ming Wang, NYA
In the National League, a popular pick among the traditional punditry seems to be Trevor Hoffman. I’ll have some of what they’re having. Hoffman got a lot of (well-deserved) attention for breaking the saves record; he truly is one of the best relievers of all-time. But his 2006 season was hardly Cy worthy, even in rudimentary measures like saves; 46 saves in 51 chances is great, but not historic by any means. Hoffman benefited from a .247 %H, and worked only 63 innings. +25 RAR is nothing compared to the +47 put up by Cy-discussion worthy BJ Ryan in the AL.
In fact, no NL reliever had a good enough campaign to crack by ballot. Takashi Saito was the most impressive (+31 RAR, 2.02 eRA), but giving him as much of a leverage bonus as Ryan or Papelbon is tough to do when he was not the closer for the entire season (he has around a 1.5 LI according to Fangraphs, while Ryan is near 1.7 and Papelbon 1.9).
So it comes down to the starters, not the most impressive crop either. The three I’ll look at for the honor are Roy Oswalt, Chris Carpenter, and Brandon Webb. Carpenter, the defending winner, had another fine season (221 IP, 3.33 RA, 3.38 eRA, +64 RAR). Webb bested him across the board, although not by large margins (235, 3.29, 3.20, +70). Oswalt matched Webb in RAR while pitching 220 innings, but his 3.92 eRA is significantly higher then the other two pitchers.
Looking at wins and losses, they are almost identical; Oswalt and Carpenter 15-8, Webb 16-8. Run support? You guessed it, they’re a match; Oswalt 5.22, Webb and Carpenter 5.40. However, bringing park adjustments into play (Arizona is a strong hitter’s park at 106, HOU moderate at 101, St. Louis 99), Webb’s win-loss mark is 5.8 wins better then a replacement versus 5.1 for Oswalt and 4.5 for Carpenter.
The three are close enough that you can reasonably argue for all of them, but I will go with Webb, who is tied with Oswalt in RAR while besting him in peripherals and posting the best W-L mark. After them, the surprising (and over his head) Bronson Arroyo and John Smoltz round out my ballot. Roger Clemens might be in the mix had he pitched a full or close to it season--he was +41 RAR in 113 innings with a brilliant 2.68 RA and 2.83 eRA.
NL Cy Young
1) Brandon Webb, ARI
2) Chris Carptenter, STL
3) Roy Oswalt, HOU
4) Bronson Arroyo, CIN
5) John Smoltz, ATL
Monday, October 09, 2006
(Rhetorical) Question
Suppose that you read somebody's blog, and they gave predictions on the playoff series. And while this person explicitly told you that he didn't have a lot of confidence in his predictions, and that the playoffs were a crapshoot and anyone could win, etc., he picked the wrong team in each of the four series. The question is, would you care to know who he picked for the next round of the playoffs?
The answer, I believe is yes, you would want to know, because maybe he's always 100% wrong, and that's just as valuable as being 100% right. And in that spirit, I offer (against my better judgement) the A's in the ALCS and the Mets in the NLCS. Why the A's? No particular reason. Call it a hunch (or more likely, a hope).
Of course it is fated that the Tigers will win the World Series. So did the Diamondbacks, Angels, Marlins, and White Sox. What did all these teams have in common? I couldn't stand them.
I need some content for this post, and it will just be some random stuff:
1. There was a little flap this week about some comments Phil Birnbaum posted on his blog (linked under his name on the side of the page). The Cliff's Notes is that he criticized academics for ignoring the work of sabermetricians when they write baseball-related articles. Unfortunately, Phil specifically discussed a paper by JC Bradbury of Sabernomics (also linked on the side of the page). The problem with this is while JC may in this instance (or may not; I don't really have an opinion and don't want to get involved) have been "dismissive" of the work of sabermetricians, he has excellent credentials as someone who engages with the sabermetric community at large, best exemplified by his occasional writing for The Hardball Times. So while I agree completely with Phil's criticisim of academics in general (it is a sentiment that to some degree I expressed here; my target was "statisticians" but of course many of the statisticians I was referring to are academics), it was unfortunate that he coupled this complaint with an article by Bradbury, when there are plenty of other examples out there.
2. Kevin Kennedy criticized Joe Torre and Bruce Bochy for not bringing back their top pitchers to pitch in game four, facing elimination. Now in the case of Bochy, I can see it since Peavy could have pitched game four and maintained his normal rest. And that allows you to come back with Woody Williams in game five, plus maybe use David Wells as a long man on three days rest. So I can understand that. But if you throw Wang on three days rest in game four, you are stuck with Jaret Wright anyway, and you don't get either your "ace" on full rest or skip your bum. So I'm not sure what that is supposed to accomplish. Yes, you can't get to game five if you don't win game four, but your goal has to be to win both games. Having a 60% chance to win game four and a 40% chance to win game five is not more valuable to you then having a 50% chance to win both games (just throwing out numbers there), even if it makes you feel better. It doesn't matter if you lose in four or lose in five.
3. One thing Bochy did that did bother me was bat Josh Bard cleanup. I don't even want to get into the debate about ideal lineup construction and how big of an impact in terms of runs the batting order makes. That's not the point. I just see it as laziness on the part of the manager. And other guys do this too (I remember seeing Mike Redmond batting third for the Twins while Joe Mauer was getting a day off). You slide the replacement into the regular guys spot because that way everybody else stays in their same spot. If the players really do have a psychological problem with batting in different spots (and really, what's the difference between say #6 and #7 anyway?), then if I was a manager I'd see what I could do to shake it up and change their thinking on it.
4. Remedial Math for Torii Hunter. But keeping it simple with run expectancy, and overly simplistic assumptions. Everyone knows that with 2 outs and a runner at first, Hunter dove for a ball and it got by him and was an inside-the-park homer. Given Tango's 1999-2002 RE table, the expectancy is .251 runs. Obviously it is 0 if you make the catch. If it goes for an inside the parker, it is none on and 2 out, which is .117, plus the 2 runs that scored on the play. So 2.117 runs. The breakeven percentage is:
.251 = 0(X)+2.117(1-X)
Where X = 88.9% So if you can make the play 89% of the time, you break even. Of course, this assumes that it always goes for a homer if you don't catch it, which is not true, and makes the play a little better. But the point is, you've gotta be pretty confident that you're going to make the play.
5. Technical Sabermetric stuff that apparently needs to be repeated constantly, and hopefully somebody will remember:
a) Win Shares has a built-in baseline. Putting lipstick (distribution of team wins) on a pig (marginal runs versus a .200 player) leaves you with a pig, not Kate from Lost.
b) OW% is the W% a team would have if they had AVERAGE defense and hit like the player in question. And therefore when you say a replacement player is a .350 player, it doesn't mean that a team filled with replacement players would be .350. They'd be worse. A replacement level offense with an average defense would be .350.
Thursday, October 05, 2006
Internet Baseball Awards: Rookie of the Year
Last year was a bad year in terms of big-impact rookies. Ryan Howard was my choice (and the BBWAA) for the NL, despite having just 345 plate appearances. What else was one to do when the other choices were Jeff Francouer, Willy Taveras, Garrett Atkins, and Gary Majewski?
In the American League, Huston Street was a solid choice, but the crop was not spectacular. That is most decidedly not the case this year. There is a bumper crop, particular among AL pitchers.
I will start my 2006 analysis as always with the superior league that uses modern rules. The top rookie position players are Kenji Johjima (296/323/459, 72 RC, +26 RAR, +10 RAA), Ian Kinsler (278/338/441, 62, +21, +6), and Mike Napoli (231/356/461, 46, +20, +10). Two catchers and a second baseman, with Johjima being the top choice I think. But none of the position players will crack the five-man ballot that can be submitted for the IBAs.
That is because of some remarkable pitchers, two relievers and three starters. Joel Zumaya worked 62 games and 83 innings for the Tigers, with a 2.38 RRA, +35 v. a replacement level pitcher, ranking fifth in the league. But Jonathon Papelbon worked 59 times, 68 IP, with a microscopic RRA of .52, +43 v. replacement. Not only that, but Papelbon pitched in higher leverage situations then Zumaya, making him a clear choice as top rookie reliever.
Then there are the three starters. I expect Justin Verlander to come out as the top choice of the writers, thanks to his 17 wins. And Verlander was good, +48 RAR in 186 IP with a 3.89 RA. But his eRA was significantly higher (4.63), and he got 6.77 runs of support per start. I estimate that his 17-9 record bettered that of a replacement-level pitcher with 6.77 RS by 2.5 games.
Jered Weaver came out with an identical +48 RAR, but in only 123 innings. That’s because he had a 2.72 RA. Weaver’s 3.36 eRA was more in line with his RA, but on the other hand, his %H was just .245, almost certainly an unsustainable performance. His 11-2 with 5.49 runs was 5.1 games better then replacement.
Francisco Liriano pitched even less innings then the other two, just 121. But he bettered them both in RAR at +53, with a 2.31 RA (best among any AL pitcher with 15 or more starts). His 2.57 eRA was also the best in the league, as was his 3.02 GRA. Liriano went 12-3 on 5.36 RS, the lowest of any of the three, good for +5.6 against replacement.
Five runs against replacement is certainly not an insurmountable margin; there might be considerations in somebody’s mind that would tip the scale to Verlander. But I see no reason to rank Verlander against two pitchers who were far more dominant and had they been given an opportunity to be in the rotation on opening day as Verlander was, would have surely bettered him by even more.
The real question is where does Papelbon fit in. WPA advocates will point to his +5.24 WPA and LI of approximately 2, while Liriano was at +3.02. But as I have expressed on this blog before, I believe that WPA is just one way of conceptualizing value, namely in real time. But I do not see its usage, even for value applications, as a given. Furthermore, Rookie of the Year is not a purely value question. I am not sure exactly what the language for the award is, but it seems to me as if it the most outstanding rookie, not the most valuable rookie. So I don’t see any reason to overturn the RAR difference, and I have little doubt that if both stay reasonably healthy, Francisco Liriano will have a more outstanding career then Jonathon Papelbon.
However, since the other two starters have flaws (Verlander’s relatively poor component stats, Weaver’s low hit rate), I will slip Papelbon in front of them. So here’s how I see it:
AL ROY
1) SP Francisco Liriano, MIN
2) RP Jonathon Papelbon, BOS
3) SP Jered Weaver, LAA
4) SP Justin Verlander, DET
5) RP Joel Zumaya, DET
Now the Neanderthal League, where pitchers still carry around clubs and flail them at the birds flying past them. There are three interesting relief pitchers, two of them Dodgers. One is Takashi Saito, with 78 IP, +31 RAR, a 2.39 RRA, a 2.02 eRA, a 2.22 GRA--brilliant all around, and now the LA closer. Jonathon Broxton sets up for him and in 76 IP is +25 RA with a 3.02 RRA and 3.88 eRA. Adam Wainwright, now the Cardinals’ closer, pitched 75 innings, +24 RAR, 3.09 RRA, 3.44 eRA.
There is also quite a bumper crop of NL starters, with guys like Chad Billingsley, Paul Maholm, John Maine, Cole Hamels, and Scott Olsen not getting past the first cut despite solid debut seasons. The four I considered are Florida’s Josh Johnson (157 IP, +38 RAR, 3.76 RA, 4.16 eRA) and Anibal Sanchez (114, +35, 3.20, 3.55, and of course a no-hitter), San Diego’s Clay Hensley (187, +36, 4.20, 4.43) and San Fran’s Matt Cain (190, +32, 4.45, 4.14).
There are a number of position players that deserve a look, but I’ll cut it to three, leaving out notables like Russel Martin, Josh Barfield, Prince Fielder, Chris Duncan, and Josh Willingham. Those three are Nationals third baseman Ryan Zimmerman, and the Marlins double play combo, Dan Uggla and Hanley Ramirez.
Zimmerman hit 290/356/477 in 675 PA, for 102 RC, +40 RAR. Uggla hit 287/340/488 in 659, 98 RC, +42 RAR. Ramirez hit 298/356/489 in 689 PA for 114 RC, +59 RAR.
And that pretty much closes the case. Zimmerman is reputed to have great defense, which is enough for him to jump Uggla, but not make up the twenty run gap to Ramirez. Saito gets a few bonus points for high-leverage, and while Johnson and Sanchez are really too close to call, I’ll go with Johnson, as Sanchez allowed just a .245 %H (Johnson at a more normal .283):
NL ROY
1) SS Hanley Ramirez, FLA
2) 3B Ryan Zimmerman, WAS
3) 2B Dan Uggla, FLA
4) RP Takashi Saito, LA
5) SP Josh Johnson, FLA
Wednesday, October 04, 2006
Internet Baseball Awards: Manager of the Year
In this exciting series, I will share my votes for the Internet Baseball Awards, sponsored by Baseball Prospectus, and the thinking or lack thereof behind my choices.
Manager of the Year has always struck me as a silly award, because you can’t really evaluate a manager statistically. By convention you give it to the manager of the most surprising team, or to the manager of a team that is just really, really good. And of course which team you think are most surprising is biased based on your personal feelings before the season started, which may have been unfounded. Despite all of this, I will reluctantly choose in this manner as well.
In the American League, the big surprise team was the Tigers. The A’s managed to milk a .574 W% out of a .497 PW%, although it was the sub-.500 PW% that surprised me, not their division title. Some will hail the Twins as a surprise, but I picked them to play well. The East has a choke job in the Red Sox, the Blue Jays who did about as well as could be expected, and the DRays and Orioles are non-factors. Out West, the Angels, Rangers, and Mariners did about what I thought they would.
So I will go with the two “surprises” and the guy who managed the best team in the league:
AL MOY
1) Jim Leyland, DET
2) Ken Macha, OAK
3) Joe Torre, NYA
In the National League, the Marlins were pegged by me, foolishly in retrospect, to lose 100 games. There is a good chance that Joe Girardi will win manager of the year after being fired. The Astros and Reds surprised me, but Garner is not likely to garner a lot of support because they were pennant winners last year. The West played out like I thought it would, except I didn’t expect the Dodgers to end up in the postseason. So I have it:
NL MOY
1) Joe Girardi, FLA
2) Jerry Narron, CIN
3) Grady Little, LA
Tuesday, October 03, 2006
Quick Playoff Preview
This is not going to be extensive like those you will find elsewhere, but I just wanted to get on record with my picks (recognizing as always that a short series can easily turn any which way and not to take any prediction too seriously). And yes, my method of setting probabilities is flawed, as I treat PW% as the actual true W%, do not adjust for home/road, do not adjust for pitching matchups, etc. Anyway:
A’s v. Twins
A’s: 574, 529, 497 (W%, EW%, PW%), 4.81 R/G, 4.53 RA/G (park adjusted)
Twins: 593, 578, 553, 4.94, 4.22
Rooting for: A’s
As you can see, the Twins beat the A’s in every category, and the A’s component stats indicate a sub-.500 performance this year. While the A’s are my preference (like many sabermetricians, I have a soft spot for Billy Beane, but more importantly because they have my favorite player in baseball, Nick Swisher). One must remember that the Twins don’t have Liriano, who contributed to much of their success, but it’s still hard to pick against Minnesota.
P(Twins win game; Log5 based on PW%): 55.6%
P(Twins win series; above result, Binomial distribution): 60.4%
Dodgers v. Mets
Dodgers: 543, 546, 549, 5.27, 4.83
Mets: 599, 568, 572, 5.31, 4.65
Rooting for: Mets
Surprisingly, the Dodgers have below average defense, but their offense is strong enough to be equal, park-adjusted, with the vaunted Mets attack. This is a case where I am going to pick the upset: the Mets played poorly down the stretch, and really have a paper thin starting rotation (Glavine +7 RAA, El Duque -7, Trachsel -10, Maine +6).
P(Mets win game) = 52.3%
P(Mets win series) = 54.3%
Tigers v. Yankees
Tigers: 586, 597, 561, 5.23, 4.30
Yankees: 599, 608, 624, 5.86, 4.83
Rooting for: Yankees
The Yankees offense is by far the best in baseball, even more powerful then it was for most of the year with a full crew of Abreu, Sheffield, and Matsui. As Baseball Prospectus pointed out yesterday, the Tigers bench is putrid (Vance Wilson, Alexis Gomez, Ramon Santiago, Neifi Perez, Omar Infante). Their starters aren’t as good as everyone thinks they are; Verlander has a 4.63 eRA, Robertson 4.77, Rogers 4.35, Bonderman 4.51. This is a series the Yankees should win.
P(Yankees win game): 56.5%
P(Yankees win series): 62.1%
Cardinals v. Padres
Cardinals: 516, 513, 494, 4.90, 4.78
Padres: 543, 534, 555, 4.80, 4.46
Rooting for: Padres
This is a far cry from past Cardinal teams, limping down the stretch, with a surprisingly bad rotation (only Chris Carpenter ranks above average). The Padres have solid starters, a better bullpen, and almost as good of an offense. They are my pick.
P(Padres win game): 56.1%
P(Padres win series): 61.3%
I like the Dodgers over the Padres in the NLCS, the Yankees over the Twins, and the Yankees over the Dodgers in a World Series rematch of 1981…and 1978…and 1977…and….and 1941. But I am pulling for a Subway Series because I want to see the two best teams in baseball square off (although all of the NY hype for that is hard to stomach). But in the modern playoff system, that rarely happens.
Monday, October 02, 2006
2006 Park Factors
Herein I will present run and home run park factors for 2006 based on 5-year data (when applicable). I will provide a brief overview of the methodology--if you want a full description you can find it on my website or if you have any questions feel free to leave a comment. Basically, I first calculate a raw park factor (based on runs or home runs per game), accounting for the fact that the “true neutral” context includes a small percentage of games played in their home park as well as the road games (I do not, however, weight by the actual number of games played in each park). Then I take the average of that number and one, to account for the fact that only half of the games are actually played at homes (this means that the final result is meant to be applied to total statistics, not simply home statistics). Then I regress this figure towards one, with less weight given to one as the number of years of data we have for the park increases.
Here, are the PFs for 2006 (listed Run, HR):
ARI: 106, 106
ATL: 99, 98
BAL: 98, 103
BOS: 102, 94
CHA: 102, 113
CHN: 101, 107
CIN: 101, 108
CLE: 97, 93
COL: 112, 112
DET: 97, 94
FLA: 96, 93
HOU: 101, 105
KC: 100, 93
LA: 96, 104
LAA: 97, 94
MIL: 100, 103
MIN: 100, 95
NYA: 98, 101
NYN: 97, 95
OAK: 99, 100
PHI: 103, 108
PIT: 100, 95
SD: 94, 93
SEA: 96, 96
SF: 99, 90
STL: 99, 97
TB: 99, 98
TEX: 107, 108
TOR: 103, 107
WAS: 97, 94
There has been some talk about how Coors Field has played differently this year, often attributed to more aggressive use of the humidor. Coors Field did have its lowest raw home/road run ratio of the last five years--just 1.15, compared to a previous low of 1.24 in 2003. However, this still made it the second most offense-friendly park in the majors this year (Great American in Cincinnati was at 1.15 as well; Royals and Bank One were in that neighborhood as well. Side note: Yes, I know its not called Bank One anymore. Or Royals for that matter. I don’t care, I use the name I’m used to). Some people will argue that this is a fundamental change and that it should be handled differently then other parks, but I’m going to use the five-year average. If somebody could show me that it was a statistically significant difference, that would be one thing, but I haven’t seen evidence to that effect and anecdotes about the ball being moister or whatever the claim is don’t fly with me. However, I will give you the results by a couple alternative ways I could calculate the Coors PF (I certainly do not hold all of these as equally valid)
1) Treat 2006 as one year, don’t regress. Then the PFs would be 107, 108
2) Treat 2006 as one year, regress towards 1 like I would for another park. 104, 105
3) Treat 2006 as one year, but regress towards a historically normal Coors PF like 1.15 instead. 110, 111
If I was going to do something out of the ordinary, it would be option 3. As MGL has pointed out, parks shouldn’t be regressed towards 1. Each park should be regressed towards its own expected value, which we could determine based on a number of factors such as altitude, fence distance, surface type, foul territory, weather, outdoor/dome, etc. However, I have not undertaken the research that would be necessary to produce these custom values, and I don’t believe that anyone else has, at least not published and with a formula that can be used to combine all of the factors (obviously studies have been done into the more specific areas, but you would have to consider all of them in order to properly do what I am talking about).
So since we know that Coors Field has traditionally played as a very strong hitter’s park, and since we know that altitude is one of the largest factors in how a park should play (and that Coors is clearly at a very high altitude, i.e. very conducive to offense), it is silly to expect that the unseen mean we are regressing to for it is 1. That is what I have done in calculating all the parks, of course, but when we have five years of data, the differences are fairly negligible (If I didn’t regress the five year Coors results at all, I would get 114, 113 versus the 112, 112 I show above). For one year, though, it would definitely be an issue, especially for a park like Coors that we expect to be extreme, even if the balls are damp or whatever exactly it is they are supposed to be this year that they weren’t in the past.
Wednesday, September 27, 2006
Evaluating Pitcher W%, Pt. 2
As discussed in the first part of this series, an overlooked aspect of the traditional NW%/WAT approach is that it makes certain assumptions about how a team achieves its winning percentage (namely, all through the efforts of non-pitchers). So why not attempt to improve our methodology by using a more realistic model of W% causation?
I should note at this point that a lot of the ideas I am going to discuss were first published by Rob Wood in the August, 1999 edition of By the Numbers (see link to BTN archives on the right side of the page). While my results may not exactly match his, and my explanation is my own, it would be disingenuous to not acknowledge that he did this stuff first.
For any given team, our best assumption will be that their offense and defense are equally responsible for the team’s deviation from .500. Certainly this assumption will be wrong in some cases, worse then using the Oliver or Deane assumptions. But it will be correct more often and the overall error introduced by this assumption will be less then for others.
To keep things simple, I will assume that all defense is pitching. This is an obviously faulty assumption, but it will keep things workable, and again while this assumption will not always hold, it is better to assume that all pitching is defense then to assume that all deviation from .500 is the product of the offense. If one wanted to get even more precise then we are going to, they could introduce a correction for this.
Suppose we have a pitcher working for a team with Mate .540, who goes 15-10(.600 W%). His NW% and WAT under Oliver are .540 and +1. Under the Deane method, they are .565 and +1.63.
However, we are now going to assume that this team has pulled itself away from .500 through equal efforts by the offense and the pitching (excluding the pitcher in question of course, since we have removed his decisions from the rest of the team’s when calculating Mate). Using the Pythagorean theory, we can write this equation:
Mate = x^2/(x^2+(1/x)^2)
x is the percentage of league average runs the team must score (or inversely allow) in order to achieve a given W%. x can be solved for:
x = (Mate/(1-Mate))^.25
Thus, we expect a .540 team to score runs at 104.1% of the league average and allow runs at 96.1%. Once we know this, we can calculate the W% that we expect the team to have given only the non-average offense (since we are assuming that all defense is pitching and the other pitchers are irrelevant when evaluating our pitcher) to have a W% of x^2/(x^2+1), in this case .520.
We can also solve for the implied runs allowed ratio of our pitcher. He achieved a W% of .600 on a team with an offense that scored at 1.041% of average, so:
W% = x^2/(x^2 + y^2)
.6 = 1.041^2/(1.041^2 + y^2)
y can be solved for as:
y = x*sqrt((1-W%)/W%)
Or in this case, y = .85.
Now we know that by achieving a .600 W% for a team of this caliber, the pitcher’s performance was equivalent to allowing runs at 85% of the league average. To calculate his Neutral W%, we put him on a team with an average offense, and find that 1/(1+.85^2) = .581. That makes his WAT +2.03--significantly different then the Oliver and Deane estimates, because they (largely) assume that only the offense has caused the team to rise above .500, whereas we are assuming that it is a joint and balanced effort between the offense and the other members of the pitching staff.
We can generalize this for non-2 exponents as x = (Mate/(1-Mate))^(1/z), y as x*((1-W%)/W%)^(1/z), and NW% as 1/(1 + y^z), where z is the exponent we are using. But is any of this really necessary?
We found that a balanced .540 team would allow a .500 pitcher to be a .520 pitcher. If we simply calculate NW% for our pitcher as .600-.520+.500, as we did for Oliver, we find .580--pretty much equivalent to our convoluted Pythagorean approach. Not only that, but .520 is also equal to the average of .540 and .500. So can we just use this kind of approximation?
While I am a strong advocate of using methods that are theoretically sound across as many potential contexts as possible, practically we only care about the real range of major league teams, which I’ll just assume for the modern times are bounded between .250 and .750. If we find our expected W% for a .250 team with only the offense, it is .366. The average of .5 and .25 is .375, a difference of less then 3%. So it seems pretty safe to use this simplified assumption, and not screw around with all of the Pythagorean calculations. I should also note that there is a further error introduced when you eschew the calculation of the pitcher’s NW% through the Pythagorean approach as well. So there are two sources of error 1) is estimating the comparison level as the midway point between Mate and .500 and 2) is estimating that the pitcher’s NW% will be the same linear difference from .500 as his W% is from the estimate in part 1).
Again, I should note that this is the conclusion that Rob Wood came to, and also the same as Tango Tiger’s quick and dirty method linked below, in response to the first installment of this series. (As an aside, that is the problem with doing series in installments as I am wont to do, and at the same time having a few smart people read it. They figure out what you are doing, or what you should be doing, before you post it. I should either scrap the installment approach or not allow any readers). So, to summarize the quick approach:
NW% = W% - (Mate + .5)/2 + .5
How does this kind of approach change the standing of the historical pitchers discussed in the first installment? Well, Red Ruffing now has a NW% of .521 and +10.5 WAT. Still not in the league with many other Hall of Fame pitchers, but far from concluding that he was a true sub-.500 pitcher. Steve Carlton in 1972 now has a NW% of .846 and +12.8 WAT. Interestingly, he does better under this approach then the Deane approach, probably because the Deane approach looks at the percentage of possible improvement. But in Carlton’s case, the offense is, at least by our assumptions, so bad, that even an otherworldly performance can only do so much to raise the team’s fortunes.
In a third and final installment, which I promise will be posted by the end of the decade, I will look at replacement level and ask the question “Why even compare to Mate at all”, and perhaps throw in some other odds and ends.
Rob Wood in Aug 99 BTN(pdf)
Tango's blog entry