Cliff Lee Turning the Corner?

On Saturday, Cliff Lee tossed seven dominant innings of shut-out baseball against the Atlanta Braves, continuing a torrid two weeks for the lefty. In four starts, Lee struck out 31 and walked two with a 1.91 ERA in 28.1 innings. It was also his third consecutive start without allowing a home run. Considering how up and down his season had been, Lee’s recent hot streak looks Kaufaxian by comparison. Why the sudden success?

In doing the research to answer that question, I was sure I’d find something as the root cause. I had a couple hypotheses in my head:

  • Lee was placing his cutter better. 12 of the 19 home runs he had allowed prior to August 16 were on fastballs or cutters.
  • Lee was inducing more ground balls. More ground balls means fewer home runs.

No matter how I divvied up the stats, no matter which heat maps I looked at, I just could not find a meaningful difference between the inconsistent Lee of April through August 11 and the amazing Lee since August 16. Sure, Lee’s 15.5-to-one strikeout-to-walk ratio is pretty good, but it’s not significant in 28.1 innings, and Lee’s K/BB ratio was already an MLB-best 6.2-to-one anyway. And, of course, we’re only two years separated from 2010, when Lee led baseball with a 10.3-to-one ratio. Lee, in general, is good at missing bats and being stingy with the free passes.

As it pertains to the above hypotheses, none of them explain Lee’s success. If anything, Lee is arguably placing his cutter more poorly recently, hitting the middle of the strike zone more often.

Hitters, predominantly right-handed by a ratio of three or four to one depending on your sample, are swinging at them at the same rate.

The pitches themselves are very similar, using pitch data from Brooks Baseball. I compared his worst month (June) to his best month (August).

Pitch Count Frequency H. Mvt V. Mvt Mph H. Rel V. Rel Spin RPM
Cut (June) 423 0.15 -4.89 -23.00 86.52 1.6 6.50 208 1099
Cut (August) 514 0.18 -4.46 -22.49 86.78 1.6 6.56 205 1111

H.Mvt = horizontal movement; V.Mvt = Vertical movement; H.Rel = horizontal release point; V.Rel = vertical release point; RPM = rotations per minute.

Overall, there hasn’t been a big shift in Lee’s pitch usage. He is using his curve five percent more often and his cutter five percent less often, but between the small sample size and pitch classification errors, it is not that meaningful. (The database I’m using has 0.2 percent sliders for Lee through the 11th, and 2 percent since the 16th. Lee doesn’t throw sliders; rather, they were likely incorrectly-labeled cutters.)

In terms of balls put in play, Lee has induced eight percent fewer ground balls and eight percent more fly balls while keeping his line drive rate steady at 17 percent. With fewer home runs allowed, we would have expected more ground balls, but it has been the exact opposite — 45 percent fly balls, 38 percent ground balls. However, Lee’s BABIP on ground balls went from .261 through the 11th to .172 since the 16th. Similarly, fly balls have gone from .198 to .161. So some combination of better luck and better defense has contributed to Lee’s recent resurgence.

When I looked at Lee’s performance back in June and July, I concluded much of the same — not much has changed. He was always due to regress back to his mean, his 3.11 SIERA. After going winless through his first 13 starts, he has won four of his last 11 and lowered his ERA by more than 60 points, from 4.13 to 3.52. What we’ve seen lately is not a new and improved Lee, but the same pitcher through the lens of more time and a larger sample size.