---

Tuesday, November 30, 2010

Is Mariano Rivera on the Verge of Collapsing?

I was going to just post this as a comment in the previous thread but I decided it’d work better as a new post.  This is all in rebuttal to sd2528’s understandable concern about Mariano Rivera’s decline in some of his peripheral stats last season and what that may mean going forward.  I’m not picking on you sd2528, I just think it’s an interesting discussion.

The first point is that Rivera’s K rate dropped significantly in 2010.  That is a legitimate concern, because a decline in K rate typically leads to a decline in effectiveness.

Here are Mo’s rates for unintentional BBs plus HBP and strikeouts per batters faced over his career.

There’s no indication of a trend, not that trends are particularly predictive.  Rivera’s rates have fluctuated over his entire career, and it’s almost certainly partly due to the fact that he only faces about 234 batters a year.  That’s the equivalent of one-third of a full season for a batter, and we wouldn’t overreact to a bad 230 PA stretch, would we?  Well, actually we would, but we probably shouldn’t.

I’d like to quote someone who knows a thing or two about this.

Which brings me to something else, and that is the career WAR for pitchers that uses only FIP.  Once you reach a certain number of BIP, say 2000 or 4000, whatever it is, then you NEED to know the pitcher’s BABIP, because that tells you more than not having it at all.  That’s why you need to be careful with coming up with career totals of the advanced metrics which themselves have a defacto regression component.

Tangotiger at the Book blog

By my count, Rivera’s allowed 2337 BIP in his career.

The BABIP discusstion segues to the next thing. 

FIP and xFIP have been underselling Rivera for his entire career.  Rivera has a career RA of 2.41, and a career FIP of 2.79.  Since FIP is scaled to ERA and not RA, you need to multiply that by FIP by about 1.08 to get it on the same scale.  He’s got a career BABIP against of .273.  FIP would regress that to league average (.304)  xFIP would tell you that Rivera should have allowed a HR rate of of something like 10-11%, and that his career rate of 6.3% is due to luck.  That means he’s allowed about 23 fewer home runs than expected, and again that’s due entirely to luck (2002-2010 since batted ball data is not available before that).  And this is while he’s pitched in a a pair of parks that have boosted HRs by about 14.6% on average since 2002.

So the RA scaled FIP would tell you that Rivera has allowed 115 fewer hits and 76 fewer runs than he should have over his 1150 career innings, and that it’s all due to luck.  And he’s done this with a defense behind him that has more often been bad and at times horrendous than good.  RA scaled xFIP would tell you he’s alowed 109 fewer runs than he should have, again all due to luck.

The fact is Rivera does induce weak contact and he does suppress BABIP and it’s mainly due to skill, not luck.  Is some of it possibly due to luck?  Yeah, because the small sample size that says we shouldn’t worry about fluctuations in his rate stats also means there’s a bigger chance he has been lucky in every season.  Not sure what the likelihood of that is, but I’d say probably less than 10%

Taking off my stat nerd goggles and thinking about his physical ability, his average fastball velocity dropped by about 1.4 mph from 2008 to 2009, but it increased by about 0.5 mph in 2010.  His cutter dropped from 92.8 mph in 2008 to 91.3 mph in 2009 and 91.1 mph in 2010, so he’s managed to maintain his effectiveness over the past two seasons despite that.

It’s a good bet he will need to be used less and less frequently because a 40-41 year old body just doesn’t recover like one that’s 30.  That’s a legitimate point in discussing Rivera’s value, because value is not just about rate of performance.  You’re not very valuable if you aren’t pitching.

Just because he’s now turned 40, there’s very little reason in his statistical record to think that he’s about to fall off a cliff.  He certainly could, and he’s got the same risk any pitcher does of hurting his arm and becoming worthless, but consider the following table.

Player	Year	Age	IP	ERA+
Dennis Eckersley	1995	40	50	90
Dennis Eckersley	1996	41	60	129
Dennis Eckersley	1997	42	53	107
Dennis Eckersley	1998	43	40	99
Dennis Eckersley	Avg		51	106
Doug Jones	1997	40	80	232
Doug Jones	1998	41	85	99
Doug Jones	1999	42	104	129
Doug Jones	2000	43	80	142
Doug Jones	Avg		87	150
Trevor Hoffman	2008	40	45	101
Trevor Hoffman	2009	41	54	226
Trevor Hoffman	2010	42	47	67
Trevor Hoffman	Avg		49	131
Mariano Rivera	2010	40	60	238

Dennis Eckersley and Doug Jones pitched through age 43 and were reasonably effective the entire time.  Trevor Hoffman pitched well in 2008 and 2009 before tanking in 2010, and based on his peripherals it doesn’t look like he was unlucky.  Rivera was inarguably better than any of these three pitchers in his age 40 season, so if they were able to pitch effectively at 41 you should feel comfortable Rivera will be able to as well.

Rivera’s CAIRO projection is still top tier for all relief pitchers, and it does include both aging and some component regression to the mean for his FIP and xFIP.  For CAIRO, his projection for runs allowed is based on 35% RA, 30% ERA, 15% FIP, 10% xFIP, and 15% component ERA.  So 40% of his projection includes data that is most likely to regress, and he STILL projects about as well as anyone.

He will eventually reach the point where he’s not an asset.  But all the evidence we have says that’s not going to happen in 2011.

But you never know.