Everyone knows what Jonathon Sanchez did this last season. He finally turned the corner and put all his stuff together for an impressive season:
2008: 5.01 ERA, 158 IP
2009: 4.24 ERA, 163 IP
2010: 3.07 EAR, 193 IP
or did he? Most people who follow the SABR stats pretty closely expect a pretty strong regression. And, in fact FIP and xFIP have shown him to be exactly the same pitcher three years running:
2008: 3.85 FIP, 4.14 xFIP, .327 BABIP
2009: 4.17 FIP, 4.19 xFIP, .290 BABIP
2010: 4.00 FIP, 4.11 xFIP, .262 BABIP
I decided to look at some more detailed data to ferret out the truth.
Joe Pawlikowski sums up the argument against Sanchez nicely:
The reasons for my bearish view on Sanchez’s 2011 don’t center on BABIP, though his .262 mark from 2010 certainly won’t hold up. That will affect his ERA right away. He’ll also probably have worse infield defense behind him, as he’s going from Juan Uribe toMiguel Tejada at shortstop. What really worries me about Sanchez is his high walk rate combined with average-ish home run rate. His HR/FB ratio was right around the 10 percent rate we use for xFIP, and has been around that mark for his career.
Those two factors then further combine with his high strand rate — 79.5 percent, well above his 72.2 percent career rate and fourth highest in the league in 2010 — to create a situation that he probably won’t sustain. If his LOB% falls to his career level next year, those walks are going to hurt more. The home runs will hurt more. I’d expect his ERA to climb into the 3.90 to 4.20 range next year. While that’s not bad by any stretch, it’s considerably worse than his 2010 performance.
First, let's take a simple look at his line drive, ground ball, and fly ball rates:
Sanchez is clearly decreasing the number of line drives he gets in both 2009 and to a smaller extent in 2010. But pitchers have only marginal control over line drive rates. So, is Sanchez just getting lucky or can he really suppress line drives marginally? Looking at his month by daily graphs reveals more.
What I'd like you to notice is the variance from start to start. 2009 and 2010 line drive percents rarely reaches above 20%. His LD% dropping from 21.4% in 2008 to 16.4% in 2009 is a clear change in ability. Balls that used to be line drives were turning into fly balls (which have the highest out% of any ball in play, especially in AT&T). However, whether the additional change to 14.8% in 2010 is a true change is unclear.
So, the 5% drop in LD% from 2008 to 2009 would reflect a true change in his ability to control BABIP. But it's still only 5%, which would convert less than 3% more BIP into outs. That alone would account for the change in his BABIP from 2008 to 2009. Remember, however that Dirty averages over 9 K/9. Out of 27 outs, that leaves only 18 outs that have to come from BIP. His strong K-rate magnifies the effect that suppressing line drives would have on LOB%. And in fact, we see a big jump in LOB% season to season: 67.5% to 72.6% to 79.5%.
But an astute reader would immediately point out that the LOB% jump is almost 7% in 2010 and only 5% in 2009 while the LD% only changed by 1.6% in 2010. At this point, the best conclusion would be that we that most of the 30 points of BABIP and 7 points of LOB% improvement in 2010 was luck.
I decided to dig a bit deeper:
|Total *||- - -||25.7 %||64.1 %||44.4 %||56.4 %||84.1 %||75.9 %||48.7 %||58.5 %||10.4 %|
|2006||Giants||19.4 %||66.9 %||43.8 %||52.2 %||85.7 %||78.4 %||51.4 %||56.2 %||9.4 %|
|2007||Giants||27.2 %||67.5 %||46.3 %||63.1 %||83.8 %||77.4 %||47.4 %||60.1 %||10.1 %|
|2008||Giants||24.3 %||67.0 %||45.8 %||55.4 %||82.9 %||75.7 %||50.4 %||59.1 %||10.9 %|
|2009||Giants||25.1 %||61.9 %||43.0 %||52.9 %||82.7 %||73.8 %||48.5 %||59.3 %||10.9 %|
|2010||Giants||28.2 %||61.4 %||43.8 %||58.4 %||86.2 %||76.8 %||47.0 %||57.4 %||9.8 %|
The first thing that pops up is that he's inducing a lot more batters to swing outside of the zone, and when they do swing outside of the zone, they make contact more. Most of the time contact outside of the zone results in poor contact, and therefore would reduce BABIP. The question is whether it's just luck that batters are taking worse hacks at his pitches.
Let's take a look at some Pitch F/X data
Two things jump out at me: (1) He's using his curve a lot more. Notice the dramatic increase in blue dots. More importantly, (2) His changeup and fastball have much greater vertical separation. The second one is the important point. The changeup is designed to look just like the fastball except drop out of the zone near the batter. The more vertical separation you get, the more likely the batter will misjudge and swing over it.
How is Dirty getting this extra separation? Partially, he has more movement: His fastball rose 9.6 inches in 2009 and 9.7 inches in 2010. His changeup rose 5.6 inches in 2009 and only 4.8 inches in 2010. Thus, the changeup rose 0.9 inches less than the fastball in 2010 than it did in 2009. But that's only less than an inch. Most of the extra distance and separation has come from a change in velocity.
Dirty's changeup now sits at 81.3 mph rather than 83.8 mph. His fastball also dropped 1mph, so the relative difference is only 1.5 mph. That extra 1.5 mph difference means it has more time to drop before it enters the strike zone. Sanchez improved his changeup significantly in 2010. Batters had to adjust, and that allowed all of his other pitches to get better as well. Indeed, his whiff rate was up around 1.5% for his slider, curve, and changeup in 2010.
Reader Vaccaro asks about his fastball, which dropped 1mph from 91.6mph to 90.6mph:
In 2009, 8.9% of his fastballs ended up as whiffs, 17.1% foul, and 15.3% in-play. In 2010, only 7% were whiffs, but 18.7% went foul, and only 14.7% ended up in-play. It seems that the decrease in 1mph made opposing batters swing less overall (40.4% vs. 41.3%), but when they did swing, they ended up fouling the ball off and put fewer balls in play. Could a slower fastball offset the improvement in his other pitches? The decrease in whiff% would indicate that batters made better contact, but the increase in foul% indicates the opposite. Hard to tell.
In going over his splits, I uncovered one last detail about his 2010 performance:
In high leverage situations, 2009 Dirty struckout only 6.75/9, walked 12/9, gave up 2.7 HR/9(normally 1 HR/9), and gave up a 23% LD%. In 2010, he struckout 7.4/9, walked 5.4/9, gave up 0.49 HR/9, and gave up a measly 7.3% LD%. His 2009 performance reeks of panic and immaturity, which any Giants fan could tell you about. His 2010 performance looks like growth and maturity if you put on rose-colored glasses, and looks like some growth (normalized walk rate) coupled with a healthy dose of luck (absurdly low LD% and HR%). The problem with using high leverage situation stats is it suffers from absurdly low sample size issues: 6.2 innings in 2009 and 18.1 innings in 2010. The small number of innings means it's hard to draw conclusions. Even if he was absurdly lucky, 10% more LD% would only result in ~4 more hits in those high leverage situations. Oh, the top pitchers in the game have shown the ability to suppress LD%, HR%, and BABIP during high leverage innings by sacrificing K-rate, exactly what Sanchez is doing. Although none of them have shown it to this extreme degree.
So was Dirty's 2010 a real effect? I've presented evidence that his stuff noticeably improved in 2010. His changeup was a much better pitch, which allowed him to use his curve more, as well. Is this evidence strong enough to conclude that his BABIP and LOB% for 2010 weren't flukes? Hard to say... at least some of the LOB% was luck, but part of it seemed like an improvement in his maturity as a pitcher. As far as BABIP, I'm loathe to go against all the BABIP data we have of pitchers regressing towards the mean. I will say that at least part of his 2010 improvement was due to an improved changeup and other off-speed pitches. How much of what remains is luck? We'll find out in 2011!