Cliff Lee's renaissance revisited

In the 2009 THT Annual Mike Fast wrote up a complete PITCHf/x analysis of Cliff Lee with the goal of identifying some of the main factors leading to his improvement. Throughout this post, I'll be using Mike's article as a baseline for understanding. Since Mike has since been employed by the Houston Astros as an analyst, I think it's pretty safe to say that he knows his stuff, especially when it comes to PITCHf/x.

Back in 2010, Justin Bopp created an infographic here on Beyond the Box Score that highlighted the change in Cliff Lee from before and after his renaissance. His final point gets at the first important point about Lee's sudden dominance:

2002-2007 GB% Average: 34.65%

2008-present: 43.03% -- a difference of 8.38% from the previous 5-6 years. What a change a groundball can make, huh?

Justin Bopp - Beyond the Box Score

I've included the infographic below so you can see what Bopp was getting at.

The original post that I linked above was picked up by Tom Tango, at which point a commenter noted another huge aspect of Lee's turnaround. From 2002 through 2007, Cliff Lee struck out on average nearly two batters for every one he walked (actual K/BB was 1.96) over the 741.2 innings he pitched during those seasons. Since then, he's averaged a K/BB ratio of 6.58 over 1,333.2 innings pitched.

There are a lot of possible reasons for Lee's renaissance, many of which Mike Fast mentions in his analysis. The first one to look at is his ability to get ground balls, something Justin noted in his infographic. Mike Fast's take on Lee's sudden ability to get ground balls is where we'll start:

We already observed that Lee added a two-seam fastball for 2008 and used it frequently. Were the extra sinking and tailing action on the two-seamer responsible for the increase in ground balls? Maybe to some degree, but it actually appears that the change-up, curveball, cutter and slider were responsible for more grounders than the two-seam fastball.

...

As also mentioned previously, the curveball to right-handed hitters generated a lot of weak ground balls, most of them toward the left side of the infield, and very few fly balls. Nobody could hit Lee’s curveball with authority in 2008. Only two of 25 ground balls hit by right-handed hitters off the curveball made it through to the outfield.

I don’t really have a good answer for why Lee became a groundball machine in 2008. The introduction of the two-seam fastball made some difference, but it was less of a factor than the new groundball tendencies on his off-speed stuff. There are some hints in the data that better location on his off-speed stuff may have played a role, but I don’t feel confident in claiming that as the cause.

Mike Fast - The Cliff Lee Turnaround - 2009 THT Annual

Ultimately, Lee was able to all of a sudden get a ton of ground balls, and there doesn't seem to be a good explanation for the drastic change. A bunch of smaller factors could have added up to make this big impact, but it's difficult to know for sure without reliable PITCHf/x data from 2007. The interesting thing is that this wasn't a temporary jump in ground ball rate, as Lee has sustained his post-2007 level of performance through the 2013 season.

What we can do is take a look at Lee's approach for his various pitch types. As Mike mentioned, Lee's breaking balls and off-speed pitches are generally the reason that he was able to generate ground balls at such a good rate in 2008 and beyond. Looking at PITCHf/x classifications from Brooks Baseball, we can see how Lee used his two types of pitches (note: generally speaking, we'd have three pitch type classifications: hard pitches, breaking balls, off-speed. For the sake of this ground ball argument though, I'm lumping off-speed pitches in with breaking balls).

In 2008 Lee threw 2,468 hard pitches according to PITCHf/x*, 50.49% of which were located in the strike zone. Only 34.48% of those pitches were in the bottom half of the zone. When it comes to breaking balls and off-speed pitches however, there seemed to be a distinctly different approach. In 2008 Lee threw 737 breaking balls or off-speed pitches captured by PITCHf/x, 41.93% of which were in the strike zone. 55.90% of Lee's curveball, slider, or changeups were in the bottom of the zone. The bottom of the zone, as you can see in the image below, I'm identifying as the bottom 10 zones on the chart.

*Only includes pitches captured and publicly available through PITCHf/x.

This chart also happens to show something Mike Fast hinted at when he discussed Lee's command in his article for the 2009 THT Annual. Fast mentioned that we can get an idea of command by taking a look at Lee's release angle, which can be deciphered by looking at the diagonal line from bottom left to top right where Lee's breaking balls generally fall. The zones in the bottom right are red/purple as well as a result of Lee's changeup which will fade to that part of the zone naturally.

Mike Fast noted Lee's consistency in his approach in his analysis:

Lee threw each pitch type to only one location, almost without fail. Certainly, he made some small variations in where he aimed the ball, and his execution sometimes matched the plan better than other times, but the catcher’s target was always near the same place for a given pitch type—in the middle for the curveball, to Lee’s right for the four-seamer, the cutter and the slider, and to Lee’s left for the two-seamer and the change-up.

In 201 pitches across two games, I found only three pitches that were exceptions to this rule: two two-seam fastballs targeted inside to right-handed hitters, and a 0-2 curveball targeted outside to left-handed hitter Jim Thome.

In 2007, Lee did not follow this pattern with his fastball. He didn’t use a two-seam fastball with any regularity, so to vary his fastball location; he needed to spot his four-seam fastball to both sides of the plate. However, his change-up, curveball and cut fastball target locations appeared to follow the same pattern in 2007 as they did in 2008.

Lee's command is likely one of the largest factors in his ability to get hitters out more easily than he ever was before in his career. Both Kelly Shoppach and Carl Willis, Lee's catcher and pitching coach, respectively, noted Lee's superb control throughout the 2008 season, as Lee made a run at an improbable Cy Young. This can also be seen through his plummeting walk rates, something that's incredibly rare.

Comparable Pitchers

John Choiniere helped me pull some data looking at pitchers who saw a drop in their BB/9 and BB% from one season to the next by 45% or more, as well as identifying if they were able to keep that level of production up for the following two seasons after that. In total, four pitchers have been able to reduce their BB/9 and BB% from Year One to Year Two, and sustain that drop (increase of no more than 25%) over the subsequent two seasons since 1900 according to the data. Note that Cliff Lee isn't included on the list because he only threw 97.1 IP in 2007 due to injuries; his BB/9 fell from 3.33 in 2007 to just 1.37 the following season.

From 1902 to 1903, Addie Joss of the Cleveland Bronchos and Cleveland Naps was able to lower his walk rate from 2.51 to 1.17. A few decades later, Carl Hubbell saw his BB/9 drop from 2.43 to 1.27. That was way back in 1931-1932, when Hubbell pitched for the New York Giants. Finally, Brandon Webb was able to do this as well in more recent seasons, as he saw his BB/9 drop from 5.15 in 2004 to just 2.32 in 2005. This coincided with Webb's rise to becoming one of the most dominant pitchers in baseball, a run that was unfortunately cut short due to injury. It's possible other pitchers could qualify for this list as well, but the lack of data means we can't be sure. The main point here is that Lee is in elite company.

Two current pitchers may be able to join this list in the near future, and looking at their performances may give us some insight into how Lee did it. Those two pitchers are Colby Lewis and David Price.

Colby Lewis

Colby Lewis saw his BB/9 drop from 2.52 in 2011 to just 1.20 in 2012. There are a couple of reasons to be suspicious here though, as Lewis only threw 105 innings in 2012 barely qualifying for the list. That's one reason for suspicion, but the reason for the limited innings is the bigger concern: Lewis had surgery to repair a torn flexor tendon, ending his 2012 campaign*.

*Updated to indicate that Lewis did not have TJ Surgery, but surgery to repair a torn flexor tendon. Thanks to @clayw78 for the heads up.

Still, Lewis made some significant changes from 2011 to 2012 that could have played a role in his ability to drastically limit walks. In 2011 Lewis was essentially a two-pitch pitcher, throwing his fastball and slider more than 75% of the time. His curveball, sinker, and change up combined to make up the remaining 23% or so of his offerings. In 2012, however, Lewis used his other pitches much more frequently, throwing them a combined 37% of the time. This included him throwing his sinker more than twice as frequently as he had the year before, and getting his curveball into double digits for usage percentage.

Despite his increased usage of this sinker and curveball, Lewis' best two groundball-inducing offerings, he did not see an improvement in groundball rate like Lee did from 2007 to 2008. Lewis also began attacking hitters early in counts, resulting in a 3.6% increase in first pitch strike percentage, which may or may not have been intentional. His overall zone% increased by just 1.1% from 2011 to 2012, so it's possible that Lewis was making a concerted effort to get ahead of hitters.

David Price

David Price famously saw his walk rate plummet in 2013, going from 2.52 to just 1.30. Price's jump in strikeout to walk ratio of 2.12 (from 3.47 K/BB in 2012 to 5.59 in 2013) is reminiscent of Lee's improvement in 2008. The biggest difference between Lee and Price is that Price's reinvention comes as a result of trying to salvage a solid season that would have otherwise been ruined by increased fly ball rates and decreased ground ball rates.

Taking a look at plate discipline data, we can see that Price isn't attacking the zone more than previously, similar to Lee in 2008, but he was throwing first pitch strikes roughly 5% more often than he had in previous seasons. This increase in first pitch strike percentage is very similar to the increase Cliff Lee saw in 2008.

This approach—getting ahead of batters—isn't breaking news to anyone. However, Lee and Price both seem to make an effort to get ahead of batters allowing them to work from ahead rather than behind. This largely opens up the options to them as far as what pitches in their repertoire that they want to use.

Back to Cliff Lee

Trying to decipher what was behind Cliff Lee's dramatic improvement has led us down a long rabbit hole of possible causes, interesting insights, and appreciation for how significant his improvement truly was. In the end, it's unclear whether we're any closer to determining what exactly caused Lee's dramatic improvement. Mike Fast characterized his analysis by saying:

Perhaps throwing the same pitch to the same area has allowed Lee to improve the consistency of his delivery and command of his pitches. I don’t know any way to measure things like confidence or deception, but I believe we can start to get a handle on command.

...

It’s clear that Lee did a lot of things right in 2008, and we’ve examined a number of them here. It’s obvious that Lee is a much different and much better pitcher than he was in 2007, having added an effective two-seam fastball, greatly improved his command, and significantly increased his ground ball rate.

The only statistics that clearly appear to be very different for Lee are his first pitch strike percentage which increased a great deal from 2007 to 2008 and never fell back to pre-2008 levels, and his groundball rate which was largely a result of improved secondaries.

My best guess at the cause behind these improvements is two-fold. First, Lee undeniably improved his command, allowing him to attack hitters with more confidence, something Fast alludes to above. He threw strikes early in counts to get ahead, allowing him to minimize walks and increase his opportunities to face hitters with two strikes. Second, Lee mixed up his pitches better using multiple types of fastball to keep hitters guessing. In 2008, it was primarily a combination of a four-seam fastball and a two-seam fastball (or sinker depending on classification). More recently though Lee has all but ditched his four-seamer in favor of a two-seam fastball/cutter combination reminiscent of former Phillies teammate and future Hall of Famer Roy Halladay.

We may never know everything that went into Cliff Lee's dramatic turnaround from fringe fifth starter to Cy Young winner. By trying to dig into some of the numbers, look at comparable pitchers in recent history, and examining Lee's evolution since 2008 via PITCHf/x we can get glimpses into what the answers might be. The goal is simply to start a discussion, inspire greater analysis, and delve even further into the world of Cliff Lee. Previously, I explored how Lee was so good at not walking batters in 2013, which is truly just the tip of the iceberg when it comes to analyzing Lee as a pitcher.

All we know for certain is that since 2008, Cliff Lee has arguably been the best pitcher in Major League Baseball.

All statistics courtesy of Fangraphs, Brooks Baseball, and Baseball-Reference. Thanks to Mike Fast (@fastballs) and John Choiniere (@johnchoiniere) for inspiration and data research help respectively.

Jeff Long is a writer at Beyond The Box Score and Baltimore Sports and Life. You can follow him on Twitter at @BSLJeffLong.