Part 1 here.

"Barry Zito is going to start sucking any day now. He's in decline. He's been lucky. He's losing it."

I've been hearing words to that effect from the numbercrunchers for about four years now. I didn't believe it then, and I don't really believe it now, either.

Rich Lederer described the issue well:

To Zito's credit, his actual ERA has consistently defied his FIP and DIPS calculations, as well as his PECOTA projections. He is obviously doing something well that isn't being captured in these systems.

Over at Baseball Prospectus, Nate Silver found that if you use Zito's past ERA as a predictor of his future ERA, his new contract is actually worth the money. If you use his peripheral numbers, the Giants are paying double what he's actually worth. Silver writes:

But while ERA is a very useful backward-looking metric — it's helpful in settling Cy Young Award debates, for example — it's not such a good forward-looking metric. A pitcher's peripheral statistics predict ERA much better than past ERA itself.

I'd agree with that 99% of the time, but I can think of several reasons why Zito may be an exception to the rule. Consider:

• These systems make predictions based on the fact that similar players perform similarly. But really, who is similar to Barry Zito? Sinker-slider pitchers are a dime-a-dozen, new flamethrowers appear every year, but a lefty with a big, tight curveball like Zito's are quite rare. Knuckleballers are probably more common sight.

  The curveball sets up a core pitching pattern that is unique, I believe, to Zito. More on this in part 3.
  

• DIPS and FIP are supposed to be better predictors of future ERA than ERA itself, but in Zito's case, his ERA has been less than his DIPS ERA every single year, missing by an average of 0.77.

  Year   ERA    DIPS   Diff
  -------------------------
  2000   2.72   3.88   1.16
  2001   3.49   3.67   0.18
  2002   2.75   4.07   1.32
  2003   3.30   4.24   0.96
  2004   4.48   4.53   0.05
  2005   3.86   4.51   0.65
  2006   3.83   4.87   1.04
  

• 2004 is the one year where his ERA and DIPS agree that Zito is a mediocre pitcher. But 2004 is also unique in that Zito tried a new delivery out of the stretch, standing more upright instead of hunched over. He did this, he said, to try to take some pressure off his left knee, in which he was apparently having some tendonitis.

  Again from Lederer:

  The most obvious abnormality is Zito's outstanding career Batting Average on Balls in Play (BABIP) of .269 (vs. a more normal league-wide rate of about .300). It should be noted that the one year (2004) in which Barry had a BABIP of .300, his ERA was 4.48.
  

  Whether it was the tendonitis or the altered delivery, Zito's usual ability to reduce BABIP vanished in early 2004. His BABIP in the first half of the season skyrocketed to .317. Zito abandoned this stretch delivery after the all-star break, and went back to his old one. His BABIP returned to a more Zito-like .278 in the second half.
  

• Feeling that perhaps the league had gotten used to his core pitching pattern, Zito reinvented himself in 2005. Rather than change his delivery as he did in 2004, he changed both his pitching repertoire (adding a slurve and a two-seamer) and his approach at the beginning of that season.

  The changes during 2004 and 2005 make me suspect of any trend analyses of Zito. To me, there are basically three Zitos: the three-pitch Zito from 2000-03, the failed experiment of 2004, and the five-pitch Zito from 2005-06. If you include 2004 in the analysis, you're including an anomaly. And if you just look at the last two seasons, is two a big enough sample size to identify a trend?
  

• Another thing that makes me skeptical of the "Zito is losing it" arguments is that in recent years, Zito has gotten off to slow starts. He's had horrible Aprils, and then afterwards, has settled down and had typical Zito performances throughout the summer months. Here are his career ERAs in April and August:

  Year   April   August
  ---------------------
  2000    --      2.72
  2001   4.58     1.02
  2002   4.81     2.16
  2003   2.63     3.93
  2004   6.83     3.48
  2005   6.60     2.13
  2006   5.93     3.40
  

  Since 2002, only once has Zito had an ERA over 4.06 in any May-August month, and that was in the anomalous 2004. For the bulk of the season, Zito has been as steady as they come.

  If there's one thing to be concerned about, it's that he has had ERAs over 5.00 each of the last two Septembers, which may indicate some fatigue.
  

• Many argue that Zito is a fly-ball pitcher who has been greatly helped by the large foul territory and the damp air of the Oakland Coliseum. And yet there's this: his career ERA is better on the road (3.44) than in Oakland (3.66).

  Last year, he had a road ERA of 2.97, but an ERA at home of 4.71. Obviously, the big foul territory isn't really a big part of his success.
  

• I've also heard many arguments that Zito has been helped by the A's great outfield defense. Funny, I seem to remember that when Zito came up, the A's outfield defense consisted of Ben Grieve, Terrence Long, and Matt Stairs.

  I've probably watched at least half of Zito's career starts, and my memories aren't exactly overflowing with images of Mark Kotsay running down a bunch of balls in the alleys. I have more memories of worrying that Long or Eric Byrnes would flub another easy catch.

  In other words, I'm not buying the idea that there is much correlation between the quality of the A's outfield and Zito's ERA.
  

• In fact, I'd guess that Zito has such a low BABIP because he makes batters hit easy-to-catch fly balls. He keeps his BABIP low by inducing batters to hit weak fly balls. Zito is consistently among the MLB leaders in popup percentage. Adam Morris broke it down at Lone Star Ball:

  Zito is getting almost twice as many pop ups from righthanded hitters than the general pitching population is. So while Zito is inducing fewer ground balls than the average pitcher, he also isn't giving up a ton of fly balls...instead, he's just getting more popouts.
  

  For most pitchers, a ball in play is a ball in play, and a ball hit in the air is a ball hit in the air. DIPS theory holds that once the ball is hit in play, what happens to it is almost entirely up to the batter. But Zito seems different. He gets batters to hit the ball up, but not far. He gets lots popups on the infield, and just as many shallow fly balls that have no chance of becoming doubles, triples, or home runs. Much of the time, these lazy balls in play are easily catchable outs.

  In other words, perhaps it's Zito who is making the outfielders look good, not the other way around. Is this possible? Can Zito control where batters hit the ball, when other pitchers seemingly can't? How does a fly ball pitcher with an 86-mph fastball even make it out of A-ball, let alone flourish in the major leagues?
  

It seems to me that Zito presents a one-man sample-size problem. His style and approach are so clearly unique, his numbers so consistently non-conformist, with a big shift in style in recent times, that you'd be wise in this case to take what the numbers are predicting with a big grain of salt. Or better yet, a big grain of scout.

In my next entry, I'll try to scout Zito, and figure out what he does that helps him defy conventional statistics. Then we'll combine the stats and the scouting, and decide if the Giants are crazy or not.