Add new comment

.

Q.  Hisashi Iwakmua had an actual ERA of 2.66, but a hypothetical ERA (that is, an xFIP) of 3.28.  What does the Detect-O-Scope mainframe say about this?

A.  A thing that is very, very similar to what Father Frankenstein (Bill James) would say.  And similar to the what Geoff Baker would say, and similar in approach to the one that Jeff Sullivan took.

.

Q.  Does a hypothetical ERA show us what a pitcher "should have" done?

A.  It does not.  Thanks for asking.

For pitchers as a group, hypothetical ERA's tend to predict the future better than actual ERA.  Mat Olkin discovered this about 1993, with his "Olkin ERA" of SLG x OBP x 31.  He realized that it predicted future ERA better than ERA predicts itself.

That's pretty cool, when you find component stats (in any segment of life) that predict surface stats better than they predict themselves.  It's like body language.  It's a huge advantage to know that when you're talking, and somebody puts their hand near their mouth, they dislike what you're saying.

.

Q.  WHEN might luck --- > NOT be luck?

A.  You can't say that all pitchers "should have" the same BABIP -- nor the same HR-per-fly.  Joe Saunders' fly balls weren't tatoo'ed any harder than Kuma's?  Some pitchers give up a higher MPH on their batted balls, you know.

Pitchers don't pitch to the park, to their fielders, etc?

...............

Sully pointed out, on USSM, that Iwakuma had an excellent result with men on base.  Being the flexible, open-minded man that he is, he was friendly to the possibility that some of these runners failed to score because Iwakuma pitched good.

Greg Maddux had a ballet-like motion and he used to do his bullpens from the stretch "because your most important pitches in a game occur from the stretch."  Hisashi Iwakuma throws very, very good pitches with men on base.  Tom Wilhelmsen, not so much.

Let's suppose that's true.  In that case, WAR and FIP are making assumptions about Iwakuma that are not accurate.

.

Q.  Are you saying that you expect Iwakuma to pitch better-than-average in 2014 with men on base?

A.  I do, yes.

.

Q.  What if you used xFIP and WAR to dictate the Cy Young?  Why not just hand the Cy to whichever pitcher has the most WAR?

A.  Because at some point you've got to stop caring what somebody coulda done, or woulda done, as you see it ... and ask what they did done.  It's actually more objective to do that!  We could argue about FIP as a base assumption behind WAR.  There is no arguing about Max Scherzer's actual 21 wins, nor Detroit's upcoming game in the playoffs.

CC Sabathia ran an ERA of 4.78 this year.  His hypothetical ERA -- normal defense, HR-per-fly rate, etc -- was 3.76.   It's true that Sabathia threw the ball better than his ERA indicates.  

But guess what:  the Yankees in fact lost 77 games.  The Yankees in fact lost 15 games that Sabathia started -- including Opening Day, by an 8-2 score to the Red Sox.  

Are you going to "normalize" those actual losses out of existence?  Are you going to say the Yankees didn't actually score 650 runs and give up 671?  Should we use OBP and SLG to declare the pennant winner -- in fact, why 9 innings and a binary outcome to each evening's AB's?  

(Our own Joe Saunders also had a real ERA that was more than 1.00 runs worse than his hypothetical ERA.  Did those runs sting you any less because they hypothetically wouldn't score next time?)

.

Q.  Hm.  When do you use W's, L's, and runs actually scored to give out player awards?  And when do you use fine-grain analysis of underlying skills?

A.  One of the first things Bill James wrote that I remember -- the late 1980's -- he advised his fans, "Be very careful, when you are accounting for Wade Boggs' doubles, that you don't normalize the Red Sox' wins out of existence.  If Boggs has a real ability to dump pitches off the Green Monster, then real wins and losses will result."

I don't mean it in a snide manner; I mean it as a critical philosophical question.  A lot of sabermetricians -- including every sabermetrician who doesn't consider Iwakuma a bona fide Cy Young candidate -- need to ask themselves a philosophical question.  Who cares about the score being 5-2 after nine innings?  Why does that even matter?  Would (say) Joe Sheehan prefer a game in which the 5-2 score did not decide anything?

Boggs hit real doubles off the Green Monster, and Hisashi Iwakuma locked down real lineups that he was facing.

.

Q.  Sure, he locked down lineups, but Scherzer locked them down better ...

A.  Check out a little table here:

Pitcher	Team	IP	ERA	FIP	WAR
Iwakuma	Sea	219	2.66	3.44	4.2
Scherzer	Det	214	2.90	2.74	6.4
Sanchez	Det	182	2.57	2.39	6.2
Verlander	Det	218	3.46	3.28	6.0
Felix	Sea	204	3.04	2.61	6.0
Sale	Chi	214	3.07	3.17	5.2
Fister	Det	208	3.67	3.26	4.5

If a fan were to do nothing more than scan down the WAR column, he would think that Hisashi Iwakuma should be nowhere near the Cy Young.  But how many fans even realize that pitching WAR does not reflect actual results on the field?

Chris Davis hit 53 home runs this year, like Hisashi Iwakuma had a 2.66 ERA in 200+ innings.  What are you going to do about his Oriole home run record?  Rule it irrelevant and focus on his ground ball rate?  Whoops, dumb question.  I'm sure the day is coming when references to HR are considered to be in bad taste, as RBI are right now.

Will the day come when W's and L's are considered to be in bad taste?  Don't laugh.  We've broached the subject.  The World Series is a random outcome, don'cha know.  Even attempting to win it -- as opposed to attempting to amass WAR -- is considered naive by some sabes.

...........

Scan down the IP and ERA column and Iwakuma would be #1 in the Cy.  Actual scoreboard changing, Iwakuma was the best.  But WAR rules him out of the discussion.

Which do you consider misleading here?  IP and ERA, or WAR?  ... so why does it "break code" to even mention ERA?

.

Q.  Does SSI like WAR?

A.  Loves it.  

Loves Win Shares.  Loves VORP.   Loves knowing --- > which couple of stars were wayyyyy lucky, and being aware of that for next year.

It's just that Dr. D is aware of the fact that sometimes, "luck stats" should not be "normalized" completely out of existence.  And that Dr. D does not value "should have beens" as much as --- > he values "what did beens."

.

NEXT