Randy Wells’ struggles from the stretch

Chicago Cubs pitching coach Larry Rothschild has been hanging out in the video room lately. The longest tenured member of the North Side field staff found a flaw in Randy Wells’ delivery.

The problem Rothschild discovered was evident only when Wells was pitching out of the stretch, which manager Lou Piniella tied to the ex-catcher’s struggles in 2010.

Notice that when [Wells] has his troubles, or has had his troubles, there’s a guy on base.

Let’s start by comparing Wells’ performance with runners on or the bases empty in 2009—when everything was presumably hunky-dory—to his less-than-stellar 2010. Using MLBAM Gameday data (including PITCHf/x) we’ll try to spot the difference(s) in Wells’ 2010 performance and ponder the possible ties to his mechanical issues.

I’m biased toward believing in Rothschild (notwithstanding the infamous towel drill). I expected—going in—to find something in Wells 2010 out of the stretch that wasn’t there in Wells 2009. Rothschild has quite a record&,dash;and I’m not just talking about Carlos Silva. An example of his effect on pitchers can be found in this article at Another Cubs Blog. In a nutshell, time spent with Rothschild usually means more strikeouts and fewer walks.

Back to Randy Wells.

Performance metrics

Let’s start with the bottom line. Or lines. The rvera tables below show linear weights converted to an ERA-like number. rveraA is based on pitch-by-pitch outcomes (ball, strike, error, out, single, double, triple, home run) and adjusted for the count. rveraE is the same but, instead of actual batted ball outcomes, uses league average outcomes for line drives, ground balls and infield flies.

rveraA Bases Empty Runner(s) On
2009 3.80 2.66
2010 3.39 6.03

rveraE Bases Empty Runner(s) On
2009 4.23 3.86
2010 4.34 4.94

While Wells toughened up with runners on in 2009, he’s now prone to getting pounded out of the stretch. Piniella was right.

Where does the problem manifest itself? Balls and strikes?

A Hardball Times Update
Goodbye for now.

Whiff Bases Empty Runner(s) On
2009 0.203 0.190
2010 0.244 0.232

B:CS Bases Empty Runner(s) On
2009 2.0 2.7
2010 2.0 2.7

Interesting splits, but nothing “new” for 2010—beyond more whiffs overall. Let’s try batted balls.

Ground ball % Bases Empty Runner(s) On
2009 43% 55%
2010 45% 45%

Call me crazy, but maybe he’s leaving the ball up and/or not throwing his ground ball stuff as often.

Pitch selection

Setting aside his rarely used cutter, Wells is a four-pitch righty: fastball, sinker, change-up and slider. As with many pitchers, Wells gets most of his ground balls from one of the sinker/slider combo. His change-up is ground ball neutral and his fastball is like most four-seamers. That is, prone to be put up in the air.

Fastball % Bases Empty Runner(s) On
2009 32% 21%
2010 37% 20%

Despite a tendency to throw a few more fastballs with the bases empty this season, Wells is still using it at the 2009 rate with runners on.

Sinker % Bases Empty Runner(s) On
2009 23% 30%
2010 25% 24%

Instead of increasing sinker usage with men aboard as he did in 2009, Wells shows no split in sinker usage in 2010.

Slider % Bases Empty Runner(s) On
2009 26% 30%
2010 20% 32%

Wells 2010 offers fewer sliders with the bases empty and a smidge more with runners on compared to Wells 2009. He’s turned a small-ish split into a large one.

Change-up % Bases Empty Runner(s) On
2009 17% 16%
2010 18% 23%

Wells is throwing more change-ups in 2010 with runners on base while he had no split in change-up use in 2009.

Putting it together, Wells is leaning less on his best ground ball pitches when runners reach base. Seems backwards to me, but doesn’t seem to be the big a-ha. Pitch selection is part of the story, but the story deepens when we look at the performance metrics by pitch type. Specifically, ground ball rates.

The last row of this table is the same data shown above in a slightly different layout. BE is bases empty, MOB is men on base. Obviously, the first four rows of data here are by pitch type.

Pitch 2009 GB% BE 2009 GB% MOB 2010 GB% BE 2010 GB% MOB
Change-up 46 48 43 36
Fastball 27 56 33 30
Slider 51 61 56 48
Sinker 44 53 50 58
  43 55 45 45

I can imagine a situation where a mechanical flaw takes a pitcher’s stuff further up in the zone, when he may have actually found it easier to get the ball down before the problem started. Wells is not having a problem getting ground balls out of the windup, so we seemed to have found a sign of agreement with Rothschild’s diagnosis—even though we don’t know exactly what he found or what he expects it to fix.

What we just saw was, in short, a failure to improve ground ball rates out of the stretch. Wells could do it in 2009, can’t do it in 2010—except with his sinker, which not only retained its split but improved overall. He’s throwing it less often with runners on base, though.

Going up

To tighten up the link between our observations and Rothschild’s, let’s dive into the all-important issue of pitch location. My analysis is incomplete, but here’s what I have found so far, a look at the average height of pitch by base state. There is no adjustment for batter height, which may make a difference.

2009: Nothing more than two-thirds of an inch different on average, except slider which was a little more than an inch higher with bases empty.

2010: Sliders have a larger split (nearly four inches) and sinkers are still within an inch, but on the right side of our splits (lower with men on base). More importantly, change-ups (3.8) and fastballs (2.7) are now higher when men are on base. That’s the “wrong” way, and fits the theory.

As I said these results are incomplete, but they are also inconsistent. The slider’s improved height split does not correspond with the drop in ground balls with men on base. The sinker is better in both (trivially so in height), and the all important fastballs and change-ups line-up with our expectation—being left up with men on base, resulting in fewer ground balls.

But it can’t be all about the raw height of the pitch. If it were, the amazing split in fastball ground ball rates from 2009 would be expected to show up in average pitch height. Perhaps a batter height adjustment, and analysis of the full distribution, are in order.

So, no firm conclusion, but I do have a strong feeling that Rothschild’s adjustment will get Wells back on the ground ball track with men on base.

References & Resources
PITCHf/x data from MLBAM and Sportvision. Pitch classifications by the author. Batted ball data from MLBAM.

Newest Most Voted
Inline Feedbacks
View all comments
Brad at Cubs Stats
13 years ago

As always, Harry, great work.

13 years ago

You need to correct the headings on the last chart.  All four columns are labeled, “2009 GB%”.  I presume two of them were supposed to be for 2009 and the other two for 2010.

13 years ago

I wonder if men on/bases empty ground ball splits are sustainable.  That is, since Wells allowed 12% more GB with men on in 2009, would we expect his “true talent” GB rate with men on to equal his overall GB rate or to be higher?  To what extent would we regress his GB rate with men on?  If pitchers can outperform their total GB rate with men on, then they would continually outperform component stats such as FIP.

Harry Pavlidis
13 years ago

Thanks @JoePepitone, we’ll fix that. You’re exactly right.

Harry Pavlidis
13 years ago

@Alex – on a slightly related note, I found Trevor Cahill to be unusually successful on ground balls. Something to chew on: http://www.cubsfx.com/2010/06/trevor-cahills-magical-sinker.html

13 years ago

Harry, quick question: can you break down his pitch selection one step further, and see if there’s a difference between when Hill and Soto are catching? I wonder if either who is calling the game, or perhaps the relative amounts of confidence in the two catchers affects the number of sinkers they call for.

Harry Pavlidis
13 years ago

Good question, Shawn. This is based on starting catcher, so the rare game where Soto/Hill were subbed out before Wells was done are mixed in.

sinkers 25%
fastballs 27%
sliders 26%
change-ups 20%
cutters 2%

sinkers 27%
fastballs 29%
sliders 27%
change-ups 15%
cutters 2%

13 years ago

Not a huge split there, except for the changeups.