In Part 1 two weeks ago, I looked at how Casey Janssen emerged into an effective high leverage reliever over the past two seasons, after having previously being more of a serviceable middle reliever (apologies for the way longer-than-intended delay; unfortunately, I'm just not a very proficient writer and quality prose come neither quickly nor easily to me). As hinted at towards the end, I saw some commonalities between him and Scott Downs, and what their careers could possibly tell us about Brett Cecil's future as he too in my view shares some of those similarities and appears destined for a future in the bullpen. The conventional wisdom is the Blue Jays blogosphere is that Cecil could make a good LOOGY, but won't contribute much more than that. Consequently, combined with his out-of-options status and other lefty options such as Aaron Loup, this would make him extraneous to the Jays plans going forward. I do not agree with this view, and what follows is my explanation why.
PROFILING JANSSEN
Previously, I focussed by design only on Janssen's performance (and platoon splits) as a reliever, in order to look at his improvement on an apples-to-apples basis. In addition to that, Janssen has also seen time in the starting rotation, consisting of 22 starts in two stints between his rookie season of 2006, and 2009. We can therefore examine his performance and splits as a starter in the manner as was done in Part 1, and compare against his performance as a reliever to see the progression upon conversion:
Brief refresher on the numbers: I measure pitching performance according to both FIP (measuring what the pitcher has most control over) and wOBA (measuring all results on a linear weights basis), and use an average to determine the platoon split. A positive number indicates a typical platoon split given the reliever's handedness, a negative number is a reverse split.
Before digging into the numbers, one quick note. In 2006, Janssen made 17 starts and two relief appearances. Since those relief outings represent an immaterial part of the overall 2006 numbers, and would not affect the relief totals, I have not broken them out and they are counted as part of the starting sample.
The numbers paint a straightforward picture: as a starting pitcher, Janssen was adequate against righties. His FIP of 4.55 and wOBA allowed of .328 are both basically around league average. Against LHB he was very poor, to the point of approaching replacement level, posting a FIP of 5.27 and allowing opposing batters to rack up a .371 wOBA. Overall, his platoon split amounted to around 14%.
In the pen, he was markedly better, even looking only at the 2007-10 period immediately following his first stint as a starter in 2006 (coincident with his second stint as a starter) and prior to his 2011-12 improvement. Against RHB, his FIP improved by 0.89 runs/9, though his wOBA only came down 17 points, compared to a larger reduction against LHB of 1.19 runs in FIP and 52 points of wOBA. Granted, those aren't great numbers against lefties, but whereas he was essentially replacement level as a starter facing them, as a reliever he was competent and closer to average.
Unfortunately, we don't have any Pitchf/x data for 2006, and so it's not possible to look at it to get a better idea of exactly how Janssen's stuff ticked up in the pen, in terms of both raw velocity and ability to generate swings and misses and other positive outcomes (ground balls, more strikes, etc). There is some Pitchf/x data from the five 2009 starts, but it came after missing all of 2008 and the beginning of 2009 to a torn labrum and recovery setback in addition to a DL stint immediately afterwards. Between that and it being a small sample, it would be a poor basis for comparison.
With that said, let's step back and take a broader look at Janssen's career track. Drafted in the 4th round of the 2004 out of UCLA, he signed the same month and was assigned to short season Auburn, making 10 starts. He posted solid peripherals, which are summarized below alongside the other levels. That earned a promotion to low-A Lansing for 2005, and further success earned two further promotions such that he finished the season at AA New Hampshire. Overall, the results were quite positive at this juncture:
Note: BB* = BB + HBP
Underlying these strong peripherals, however, were some yellow flags. In Lansing, he did not allow a HR in 46 IP and yielded only 5.3 H/9. At Dunedin those marks rose to 0.3 HR/9 and 6.9 H/9, and in New Hampshire they crested at 0.6 HR/9 and 10.3 H/9. This pattern indicates increased hitability against better hitters. This is to be expected to some degree, but in my view (with hindsight) the magnitude of such increases is an indication of the quality of the pitcher's stuff and its ability to play at higher levels. One mitigating factor in this case is that Janssen's K/9 was maintained (indeed, increased) from level to level, indicating a continued ability to miss bats.
In 2006, Janssen started at AAA Syracuse and made four starts before being called up to the big league rotation when A.J. Burnett had to miss a start. Across 20 IP in AAA, he tallied similar peripherals as he did the previous season at New Hampshire: 18K (8.1 K/9), 1 BB, 2 HR (0.9 HR/9), and 21 hits (10.5 H/9).
Initially, Janssen had a run of success for the Jays. In his first nine starts across 55.2 IP, he compiled a 5-3 record with solid peripherals: 27 K (4.4 K/9), 12 BB* (1.9 BB*/9), 4 HR (0.65 HR/9), 43 H (7 H/9). Compared to the minor league numbers, Janssen's K rate fell dramatically (due a swinging strike rate of only 7%), while his walk and HR rates were mostly in line. Interestingly, he gave up a lot fewer hits, though this was mostly due to a BABIP of .222, and if that number if normalized the hit rate would increase significantly. Nonetheless, he was quite an effective starter over this time.
Unfortunately, this state of affairs was not to last. From June 12 until being optioned back to AAA at the end of July, the results and most peripherals declined markedly: 4.0 K/9, 3.8 BB*/9, 1.9 HR/9, and 14.1 H/9 (.359 BABIP). Interestingly, the strikeout rate did not change significantly between the splits, as it had already cratered from the minor league levels. The chart below shows the progression of several of Janssen's peripherals from level to level:
Note that all numbers are only through 2006. Overall, Janssen maintained a solid K rate until he hit the major leagues, when it cratered. His BB* rate similarly stayed very good until increasing in the majors. However, in terms of hitability (yielding hits and HR), the rates were consistently increasing from level to level, including at the major league level.
PROFILING DOWNS
Similar to what was done above with Janssen, we can look at Downs' career splits as a reliever and as a starter
I split out the relief innings in 2005-06 from those subsequently, as during this time Downs was essentially a swingman going back and forth between the bullpen and rotation and making multi-inning relief appearances rather than being a one inning reliever. I will not consider those 2005-06 relief innings further, as in my view they consequently do not represent a good basis for comparison. The story is broadly similar to that of Janssen. As a starter, Downs was somewhat better than Janssen against both same handed and opposite handed batters (much better on a FIP basis, slightly worse on a wOBA basis), but with a similar overall platoon split. As a reliever, Downs improved against same handed batters, pulling in his FIP by 0.60 runs and wOBA by 70 points. Again, however, like Janssen the improvement against opposite handed batters was much larger, with Downs reducing his FIP by 1.50 runs and pulling his wOBA in by 100 points. As a result, he eliminated the platoon splits he had as a starter.
Once again, there is no Pitchf/x data for Downs as a starter, so there's no ability to compare the stuff. I won't go into the same level of detail, but once again, let's step back and look at Downs' profile. He was drafted in the 3rd round of the 1997 draft, and also moved fairly quickly through the minors, earning a rotation spot for the Cubs out of spring training in 2000 (despite not having pitched above AA). The following chart lays out his peripherals from level to level:
MLB* denotes Downs' numbers in his initial MLB stint (2000), whereas the MLB column denotes all MLB inning as a starting pitcher (to show any changes from initial performance). As with the MLB splits, a similar progression is evident for Downs as for Janssen. In general, the strike out rate was quite good, averaging just under one strikeout per inning though with a significant dip at the high-A level so not as consistent as Janssen. His walk rate ticked up from level to level in a more pronounced manner. Significantly for me, the rate of yielding hits and home runs increased in a very consistent and similar manner. Overall, the similarities in statistical similarities between the two are quite striking to me, even moreso than what I initially expected.
WHAT ARE THE COMMONALITIES?
Stitching some of these similarities to move towards a profile of the failed-starter-to-valuable-reliever convertee:
1) College pitcher drafted fairly early, in the 75th to 125th overall range
2) SP in minors, moved quickly to the major leagues
3) Generally solid or good peripherals in the minors, however negative trends apparent from level to level in terms of hits and HR allowed.
4) Significant platoon splits in the major leagues: competent or better against same handed batters, significantly below average to replacement level against same handed batters.
5) Gopher ball problem in MLB: high HR rate. Additionally, a strikeout rate that plunged at the big league level. Combined, this suggests inferior stuff for a big league starter.
HOW IT APPLIES TO CECIL
1) Cecil was also a college pitcher, drafted in 2007 out of the University of Maryland; however, he was a much higher pick than either Janssen or Downs, chosen 38th overall in the 1st round supplemental round. Interestingly, Cecil was used primarily (and almost exclusively) as a reliever in college, though my understanding is the Jays always intended for him to be a starter. On balance, his pedigree as a prospect coming out of the draft was higher than that of Janssen or Downs. Unlike Janssen and Downs (to my knowledge anyway; he did have elbow issues and ultimately Tommy John surgery), Cecil suffered from a well-documented velocity loss towards the end of 2010 and into 2011 and beyond. The resulting lesser stuff would have reduced his draft status had it occurred prior to turning pro, so overall I think the pedigree/profile is quite similar at all between the three.
2) Until the latter months of 2012, Cecil had been exclusively a starting pitcher as a professional. He moved very quickly through the minors, reaching MLB in 2009, less than 2 years after being drafted and having covered four minor league levels.
3) The following chart shows Cecil's peripherals from level to level
The general story is very similar to Janssen and Downs. Cecil had a consistently high strikeout rate in the minors, but it fell significantly immediately on hitting the majors and stayed at the lower level. His minor league walk rate was higher in the upper minors, however it settled in at a comparable (and acceptable) level in the majors. His home run rate was a little higher overall, however he too experienced a large increase at the big league level and it settled in at a higher level. This is somewhat attributable to Cecil being more of a fly ball pitcher, especially compared to Janssen and Downs who have consistently been ground ball pitchers. Additionally, he was increasingly hittable as he moved up, similar to the others. Overall, Cecil's peripheral progression is very, very similar to Janssen and Downs.
4) Below is a chart detailing Cecil's MLB platoon splits
I have broken out 2009-10 from 2011-12 to try to isolate the effect of the velocity loss, which appears to be a reality going forward. Overall, Cecil has had larger splits as a starter than Janssen and Downs, but the picture is broadly similar: solid against same handed batters, awful against opposite handed batters. But I think it's more instructive to look pre- and post-velocity loss splits. In 2009-10, Cecil had more modest splits that were more in line with Janssen and Downs as starters. He was not a world beater against RHB, but he was playable with an average-ish FIP though decidely below average wOBA. As would be expected, he was much better against lefties. Overall, this version of Cecil was susceptible to line-ups stacked with righties, but would have have had a decent shot at sticking in the backend of a rotation.
In 2011-12 post velocity loss, Cecil's performance was drastically different and in a very interesting way. Against RHB, the results were a trainwreck with a FIP 1.20 runs higher and a wOBA 27 points higher, both numbers well below average. The average RHB hit like 2012 Josh Hamilton against Cecil, that's how bad it was. Conversely and quite surprisingly, against LHB Cecil was much better against LHB, improving his FIP by 0.69 runs and 73 points of wOBA. How to explain this? The next section will look at some Pitchf/x data, but the end result was that his platoon splits blew out to 67%. The final point I want to make is referring back to Cecil's velocity chart: as would be expected, Cecil's velocity ticked up at the end of the year coming out of the bullpen to the 90-92 range from the 87-89 range he sat in as a starter for most of 2011-12. This was on par with where he sat in 2009-10, when he was far more effective against righties. This alone gives me a fair degree of confidence that even barring any other improvements, as a reliever Cecil can be far more effective against RHB than he has the last two seasons when he was not playable against them.
I intentionally did not include in those above numbers his splits as reliever at the end of 2012, since they represent a very small sample size and Cecil was bounced around. In 12 relief appearances, three times he faced just one batter, and three time pitched two innings (facing 7 to 12 batters) and the rest in between. I do want to present it for completeness:
Again, I want to caution against giving this much inferential value going forward, but Cecil handled RHB effectively coming out of the pen. He still exhibited a large split, but that's largely because he continued being very tough on lefties.
5) Anyone who has watched Cecil pitch over the last couple seasons is familiar with his propensity to give up home runs as a starter. The less said about that the better. As detailed above, the K rate fell. In terms of stuff, this is best addressed looking at some PitchF/X data.
WHAT DOES PITCH F/X SAY ABOUT CECIL?
Unlike for Janssen and Downs, Cecil's entire career is covered by the Pitchf/x era. We can't really make any comparative inferences, but I was hoping Pitchf/x could supplement some of the statistical profiling of Cecil, particularly as it relates to things like his 2011-12 improvement against lefties.
Using data from Brooks Baseball and their fantastic player cards, let's first look at pitch selection:
Like Janssen, Cecil has at various points made use of six different pitches. Initially, he used a quite conventional four pitch mix consisting of the two seamer, four seamer, slider, change-up. Against LHB, the slider was Cecil's main secondary pitch, mixing in the change-up a little bit when ahead in the count or with two strikes. Against RHB, it was the opposite, using the change-up as the main secondary pitch and using sliders mainly to put away batters. Once in a while, a cutter or curveball was mixed in. One thing to note is that even in 2009, Cecil only used a pure fastball about 55% of the time, which less than most pitchers.
In 2010, Cecil cut back on the use of his two seam fastball, particularly against LHB. Instead, he used four seamers and sliders more often, with a smattering of change-ups but even its usage was dialed back. Against RHB, he also shifted to more four seamers, though not as drastically, and shifted from sliders to more change-ups.
Then came the velocity loss in 2011. It appears that nothing was radically altered at this point: increased slider usage against RHB at the expense of change-ups, but that's about it. It's a little curious, since sliders tends to work better against same handed batters and change-ups against opposite handed batters, but it may have been a case of just going falling back on what works when other stuff isn't (ineffective fastball not properly setting up the change-up).
2012, however, brought huge changes. I broke out his mix by starting and relieving and there are some differences, but the important changes are from prior years. Cecil basically completely reinvented himself in terms of pitch mix. For starters, the slider was completely junked against hitters from both sides. As can be seen below by looking at whiff rates and ground ball rates, this was one of his most consistent and effective pitches previously:
In 2011, the ground ball rate declined on the slider, though that may have been due to using it more against opposite handed hitters who got better looks at it (the 0% in 2012 is not meaningful, only nine sliders were put in play). It strikes me as odd that the slider usage was completely eliminated, but I'm thinking it may be related to the velocity loss. In his first couple years, Cecil threw the slider at around 85 MPH, which was a 6-7 MPH difference from his fastball. However, when his velocity fell more into the 88 range, though the slider velocity fell also (perhaps by design), the differential was only 5 MPH.
In its place, Cecil turned to his curveball, which was previously a show-me pitch for all intents and purposes. My thinking is that this was driven by the fastball velocity decline, since curveballs are slower than sliders and therefore the fastball/breaking ball differential would be larger. However, out of the pen with better velocity he did not go back to the slider. Perhaps this was due to it being late in the year, or maybe it represents a more fundamental shift. Cecil had a major uptick in the whiff rate on his curveball in 2011, which was maintained in 2012. This may suggest an improvement in the pitch overall. Additionally, the velocity increased to over 79 MPH, which makes it a little bit more of a power curveball.
Another main difference is that Cecil turned to a cutter, particularly against LHB. He used it some against RHB out of the rotation, but pared it back out of the bullpen (small sample size caveat). Again, this too was preceded by an increase in his cutter whiff rate in 2011. Cecil completely junked his use of change-up against LHP, and out of the pen barely used it. This is somewhat surprising to me, as it had a good whiff rate and generated a strong amount of ground balls (unlike Janssen, who also junked it as a RP, but for whom it was a marginal pitch). Finally, coming out of the pen, he also started using his fastball more against RHB, around two-thirds of the time. This is more in line with most pitchers, and makes sense given that the velocity ticked up.
All in all, I'm not sure exactly what to make of all this. Between the velocity loss from 2010 to 2011 and the complete reinvention in 2012, there's a lot of moving pieces. I don't see anything that jumps out to explain how Cecil cracked down on LHB in 2011-12. Moreover, I'm not sure how much we can say about the approach going forward. As a full-time reliever with a fastball at 90-92, the velocity differential with the slider would be where it previously was in 2009-10. On the other hand, the curveball seemed to work. It will be interesting to watch.
CONCLUSIONS
In light of the above, I believe that Brett Cecil can be a quality late inning reliever, not just limited to situation matchups (LOOGY). I am not sure that he will get to the point of being equally effective against opposite handed batters (or better) as Janssen and Downs did, but I don't believe that's a requirement for success. This is in large part because he was been so incredibly tough against LHB over the past couple seasons, that even with a typical platoon split or worse, he can still be effective against RHB. At worst, I think he'll be playable against RHB, though it may require using him selectively to avoid too much exposure against righties (more than a strict LOOGY, though not automatic one inning guy).
I assume the Jays will be going with a 7-man pen to start the year, likely with two lefties, but the situation remains unsettled without knowing exactly what Oliver intends to do. If Oliver elects not to retire, he obviously slots in as the main late inning lefty. I would slot Cecil in as the second lefty, and use him in a hybrid manner. In close games late, use him fairly strictly as a one or two out guy against lefties. Additionally, get him some full innings and exposure to righties in lower leverage situations to see how he fares, and perhaps some full innings in the 6th/7th innings if there's a run of lefties when managers would be less likely to empty their benches and play match-ups. That leaves Aaron Loup on the outside, but he has three options left. In my view, he's more likely to be a strict LOOGY anyway, meaning Cecil has the higher ceiling.
If Oliver retires, it's more complicated assuming that no acquisition is made to replace Oliver. My inclination would be to slot Cecil in as the late inning lefty, but with high expectations for 2012 that could be difficult. But there's no obviously better option, and if it doesn't work out, AA can always look to upgrade in-season. What I really don't want to see is the Jays cut bait on Cecil without giving him a fair opportunity to thrive in the bullpen and become a solid option at the back of the pen. Relievers lacking electric stuff like Janssen and Downs (and Cecil) may not be the most glamourous players and are usually criminally under-rated, but their stabilizing influence and versitility is very valuable.
Questions, comments, thoughts are appreciated below. Again, apologies for the delay to those who have been waiting.