Carlos Zambrano Reborn in Miami

Through all the tirades and tantrums that marred his eventual exit from Chicago, it can be easy to forget Carlos Zambrano is just 30 years old. This season, Zambrano is showing the world that he just might have something left in the tank. Through 41 innings, Zambrano is the proud owner of a 1.98 ERA. Despite his effectiveness, he wasn’t rewarded with his first victory of the season until Monday night, when he twirled his best start of the season, a complete game, nine-strikeout shutout of the Astros in Houston. In many ways, Zambrano is looking like the pitcher who shined with the Cubs throughout the last decade.

For his entire career, Zambrano has been one of the classic exceptions to DIPS theory. He owns a career .275 BABIP and an ERA a full 0.40 runs lower than his FIP. With 1867.2 career innings and 7957 total batters faced under his belt, Zambrano’s ability to control balls in play is an established phenomenon. That essential piece of his success deserted him in 2011, leading to the 4.82 ERA and the frustration which would eventually serve as the last straw for both Zambrano and Cubs management. His arrival in Miami has seen his peripheral-defying ways return. His 2.3 K/BB is hardly impressive — a fraction below league average, in fact — but he has ridden a .234 BABIP and a 0.66 HR/9 to a 1.98 ERA in the season’s early going.

It should be noted, though, that Zambrano’s peripherals are also at the highest level they’ve been in years. His K/BB ratio is just 0.06 short of his career high of 2.35 in 2005 and 0.03 short of the 2.32 mark he posted in 2004, easily his best seasons as a Cub. In these two years, Zambrano averaged 219 innings, a 69 ERA-, an 82 FIP- and 4.6 WAR. The key is a 10.2% swinging strike rate, which would be his highest in 10 years if he maintains it.

There’s reason to believe this change is real. Zambrano has radically changed his pitch mix, almost completely eschewing the four seam fastball in exchange for more pitches with movement. According to his Brooks Baseball player page, his four-seam usage is down from 27% to 10%, with that 17% going instead to the splitfinger (+5%), the cutter (+7%) and curveball (+3%), among others.

Observe, the eight swinging strikes from Zambrano’s start Monday:

Eight swinging strikes in 99 pitches is a touch below Zambrano’s pace coming into the start, but part of it was the Astros hitting into outs before Zambrano could get to splitter counts. When he did throw the splitter, it dazzled, drawing four swinging strikes out of 17 thrown and going for strikes nine times. Of course, if it were just as easy as throwing fewer fastballs and more moving pitches to generate more whiffs, every pitcher would do it. Zambrano will have to keep throwing those breaking pitches — the splitter in particular — for strikes. He’s thrown the splitfinger for a strike 61% of the time so far, five percent above the MLB average for the pitch.

Six starts in, the Carlos Zambrano reclamation project has been nothing short of a rousing success for the Marlins. The farther Zambrano’s dreadful 2011 gets in the rear-view mirror, the more it looks like he can return to the form that convinced the Cubs to commit $91.5 million back in 2008. If Zambrano can keep using his splitfinger and other breaking pitches with the success he’s shown so far, there’s no reason to believe he can’t remain an effective cog in the Marlins’ rotation.



Print This Post



Jack Moore's work can be seen at VICE Sports and anywhere else you're willing to pay him to write. Buy his e-book.


Sort by:   newest | oldest | most voted
mcbrown
Member
mcbrown
4 years 4 months ago

I remain skeptical that Zambrano has “established” an ability to outperform ERA estimators. Considering the number of pitchers in MLB I am not convinced that we shouldn’t see a number of pitchers exhibit long ERA-outperformance streaks like Zambrano’s simply due to chance – even if you believe in the predictive value of ERA estimators (which I do), in any season a given pitcher has a 50% chance of outperforming and a 50% chance of underperforming. With 300-ish pitchers, 150-ish starters, and a strong survivorship bias in favor of longer outperformance streaks (i.e. pitchers who underperform in a given year are more likely to drop out of the sample the following year than those who outperformed), I will need more evidence than Zambrano has amassed to convince me that this particular performance is more attributable to skill than luck.

mcbrown
Member
mcbrown
4 years 4 months ago

Having said that, Zambrano *is* looking like a better pitcher to me, and his swinging strike rate certainly supports that notion.

mikesavino85
Member
4 years 4 months ago

How many more innings than 1800 do you need to see?

mcbrown
Member
mcbrown
4 years 4 months ago

If we are cherry-picking a pitcher from the entire sample of active MLB pitchers ex-post and making a non-falsifiable claim about his past performance, a lot more than 1800.

If we are making a falsifiable claim about Zambrano’s future and rigorously testing it going forward, a lot less than 1800. If he does it for 3 years starting from now I will be sufficiently confident.

Nick44
Guest
Nick44
4 years 4 months ago

Since when did the burden of proof fall on those who would argue skill.

Perhaps I need more data to convince me that it’s luck.

Why don’t YOU make a falsifiable claim. It’s easy to test claims, disagree with them on the amount of data that is there and say that it will test negative with more data.

It’s certainly more likely now that he exceeds his FIP, from a purely bayesian perspective. That’s a pretty robust prior. I understand that the distribution is there, but it was not generated by a stochastic process. It was generated by the skill of major league baseball players competing against one another.

Jacob Knapp
Guest
Jacob Knapp
4 years 4 months ago

The burden of proof falls on whomever’s argument is harder to accept, in this case, that is you. There could be 30 million pitchers in the major leagues, the odds of Carlos Zambrano outperforming his FIP over 1800 innings only through luck would still be exactly the same – very low. That’s a simple math principle, if you’ve even taken a high school pre-calculus class you should understand your mistake.

Also, I think its a generally accepted fact that some batters and some pitchers can influence their BABIP. Mariano Rivera has a career .215 BABIP. It’s not because he’s lucky, it’s because he breaks a lot of bats.

mcbrown
Member
mcbrown
4 years 4 months ago

@ Nick44: Here is my falsifiable claim: Carlos Zambrano will outperform xFIP by more than 0.05 ER/9 for the rest of 2012 (beginning May 9, to eliminate the influence of his YTD performance), all of 2013 and all of 2014. I estimate the probability of doing so to be about 12.5%, so if he fails to do so we cannot be more than 90% confident that he has a sustainable, identifiable outperformance skill.

@ Jacob: This is simply not true. With a sample of only 300 pitchers I estimate that in any given 10 year sample there is a 25% probability of someone having a 10 year outperformance streak (which, by the way, Zambrano has not even had). The odds would be c. 95% with 3,000 pitchers, and indistinguishable from 100% if there were 30,000,000 pitchers. And by the way, the odds of seeing a 10 year streak by chance alone would be even higher if we are looking for 10 year streaks in a 15 or 20 year period (i.e. we eliminate the constraint that the streak must begin at a specified point in time).

Furthermore, if you think I made a math error, please explain what it is rather than resorting to insults about my alleged lack of education. Here is how I arrive at my estimate, based on very basic probability:

P[see a 10 year streak in a particular 10 year sample of N pitchers]
= 1-P[no streak in N pitchers]
= 1-P[no streak for individual pitcher]^N
= 1-(1-P[individual pitcher DOES have streak])^N
= 1-(1-P[pitcher outperforms in given year]^10)^N
= 1-(1-0.5^10)^N

This does not even consider the survivorship bias issue, which should bias the odds higher as the sample is not random but skewed towards pitchers who have tended to outperform.

Overall, I find the SABR’s community willingness to uncritically accept the “DIPS outperformance” skill extremely puzzling, considering its long history of disbelieving “skills” that could be explained easily by chance, such as clutch performance. I have yet to hear a plausible explanation of why Zambrano in particular should have this skill. I can see an argument for someone like Rivera; DIPS metrics are formulated to explain the fat part of the pitcher performance distribution, not the extreme tails, and Rivera is clearly on the extreme tail. But Zambrano? He is an above-average pitcher for sure, but what about him is so extreme that we should believe he in particular has this skill? Is it his perceived intensity? Roger Clemens was known to be pretty intense too, and he didn’t outperform his career FIP. Is it something about his pitch mix? His skill with a particular pitch, a la Rivera? What is it exactly about Zambrano that we can hang a hat on here?

steve-o
Guest
steve-o
4 years 1 month ago

definitely luck. Rebirth in the bullpen, eh?

Hurtlockertwo
Guest
Hurtlockertwo
4 years 4 months ago

When Zambrano pitched againt the Giants he had a wierd smirk on his face when he was pitching. When the happy juice wears off and he explodes again, so much for his rebirth.

EG hostilE
Guest
EG hostilE
4 years 4 months ago

Why would it wear off, though?

If he keeps pitching well, he’s reunited with his BFF Ozzie, he’s playing in Miami. He gets to watch Giancarlo Stanton take BP.

weird
Guest
weird
4 years 4 months ago

its a rebirth, just wait until he drinks too much coffee and the cramps come.

steve-o
Guest
steve-o
4 years 4 months ago

At first I thought rebirth, pssshaw. He’s faced the Astros twice. Then I looked at the team batting stats and saw that the Astros are fourth in the NL in team wOBA. Why the hell didn’t someone tell me!?! Who’s in charge of these things? Quick honey, get the kids down in the shelter!

MGL
Guest
4 years 4 months ago

It is not an “either/or” for predicting a pitcher’s BABIP going forward. It is a regression based on the number of IP thrown (or BIP, or whatever).

So if Z is .275 for a career 1800 PA, his most likely BABIP in the future is:

3700/(3700+1800) * (.290 – .275) + .275, or .285.

In other words, his BABIP gets regressed 3700/5500 towards the mean.

The argument over whether a pitcher’s sample BABIP in any number of opportunities is “real” is specious. It is always partly “real” (his true talent in the quality of non-HR BIP) and partly “fake” (a statistical aberration), as far as we can estimate.

Like Heisenberg’s Uncertainty Principle, we don’t KNOW what his true BABIP talent is, and if we did, that number would be unlikely to present itself in the future, but the number we get from the above formula is our best estimate of his true talent BABIP and given that estimate is the center of the distribution of likely outcomes in the future. The larger the sample we look at in the future, the narrower that distribution well be. The larger the sample we look at in the past (in this case 1800 IP), the narrow the distribution of possible true talents that our .285 estimate represents.

And you can substitute any mean you think is appropriate for Z for the .290. That mean cannot be generated or influenced in any way by his prior BABIP.

Nick44
Guest
Nick44
4 years 4 months ago

No. It’s not like the uncertainties in values are tied together. Or if the more uncertain you were about some other pitcher attribute made you more certain about BABIP. The Uncertainty Principle is not a one category principle.

I’m thinking Schrodinger’s Cat is a better analogy here.

pgrocard
Member
pgrocard
4 years 4 months ago

It’s terribly confusing when authors mix up HR/9 and HR/FB. I was having trouble figuring if the HR/FB was trully .66 (which would be both remarkable and terrible), or .66%, but it turns out that it’s neither.

Z’s HR/9 is .66. His HR/FB is 9.1% this year.

MrKnowNothing
Guest
MrKnowNothing
4 years 4 months ago

Wait, what’s this about his cat??? Did the cat teach him to pitch?

Bill
Guest
Bill
4 years 4 months ago

Not Zambrano’s cat, Schrodinger’s Cat. You know, the piano player in the Peanuts strip? The comment is like quantum physics, it makes no sense.

Feeding the Abscess
Guest
Feeding the Abscess
4 years 4 months ago

His first pitch strike % is awful, just .3% better than the second worst mark in his career, and 2.1% worse than last season. I can’t imagine his walk rate will stick.

That said, if he can ever figure out how to hammer home first pitch strikes, his splitter could rack up some serious Ks.

EG hostilE
Guest
EG hostilE
4 years 4 months ago

Going from Wrigley to that Cavern in South Beach probably helped, no?

Jack Weiland
Guest
Jack Weiland
4 years 4 months ago

Re: outperforming his peripherals … It’s all relative. Yes, he’s had a solid history of doing so, but never to such an extreme degree. His strand rate right now is 15% higher than his career average. His BABIP is 40 points lower than his career average.

Those were worth mentioning. Certainly possible for a 30-year-old formerly pretty good pitcher to see a spike in performance, but he’s not going to continue stranding 90% of baserunners. He’s just not, new usage rates be damned.

The Real Neal
Guest
The Real Neal
4 years 4 months ago

You mean he’s not going to have a 1.88 ERA for the season?!? Thanks, no one else realized that.

Joe
Guest
Joe
4 years 4 months ago

MGL, I believe you regressed too much – you used his IP, 1800, instead of his BIP, which is 5168, by my count.

3700/(3700+5168) * (.290 – .275) + .275 = .281

I wonder if there are any other active starting pitchers that would have a lower expected future BABIP from this formula.

wpDiscuz