The Boston Red Sox are simply not a good team right now. So, it is not a surprise that fans are calling for changes, on the field and off of it. One such call for change is to send Joe Kelly to the bullpen, but Damian Dydyn has some good reasons why that may not be the best idea.
There is just no way to sugar coat it, Joe Kelly has not had a good season. At best, he’s been bad. At worst, he’s been abysmal. You can spin it in a number of ways, but at the end of the day, the results are ugly, even if his peripherals are just sort of ugly. His 6.11 ERA hovers well above his 4.28 FIP and 4.02 xFIP so he has not been quite as bad as that first number indicates, but even if his ERA matched his peripherals, he’d still only be a solid back of the rotation starter. For someone with his stuff, that is more than a little disappointing. His troubles have been well documented but no one seems to be able to pin down exactly what is wrong with him. In the midst of all this struggling, some have suggested that Kelly should be moved to the bullpen where he would be more likely to succeed. However, that he would be more likely to succeed in the bullpen is an assumption that might not be safe to make.
Many starters will see an improvement in their results when they are moved to the bullpen because they don’t have to pace themselves and are able to throw at maximum effort. They can also focus on their one or two best pitches instead of needing three or more pitches to try and get through six or seven innings on any given night. For Joe Kelly maintaining maximum effort through his start has never been an issue:
If anything, his velocity goes up after he’s been in the game a while and he did not show an uptick in velocity up through June 18th of 2013 while he was operating exclusively out of the bullpen for the Cardinals:
There is no increase in his velocity, so his stuff won’t play up in the pen like it does for other pitchers. Another argument for moving him to the bullpen is that he can drop his least effective pitches and focus on what he does well. Unfortunately for Kelly, his pitches have been incredibly erratic as far as results are concerned. Depending on the game, any one of them could be leading to line drives:
And here is how opposing batters are hitting for power against his repertoire:
There may be some validity to shrinking his arsenal as his four-seam fastball appears to be the most volatile pitch and he would be able to rely on his sinker more, assuming there is any real difference between the two pitches in the first place. Over his career he has a giant mess of fastballs that heavily overlap in speed and movement. In the charts above there is no significant difference in velocity in either the 2013 or 2015 samples. Looking at his career, we see that holds:
That it is impossible to discern where one pitch ends and the other begins without color coding is an indication that he’s never really had two fully distinct fastballs. In fact, this season he has had very similar results with what PITCHf/x categorizes as two different fastballs. (Ian York touched on this in an article shortly after this was published.) This takes a bit of bite out of the idea that he would be able to focus on his sinker instead of his four-seamer to any great effect since he cannot simply stop using the “bad” fastball and focus on the “good” one. Regardless, we can’t dismiss that possibility just because the pitches overlap heavily on these charts.
By pitch type, Fangraphs lists only his sinker and his changeup as having positive values, so it’s not out of the realm of possibility for him to reduce his repertoire down to a sinker-changeup combination and find some success. There have been some good and even great relievers working primarily with a fastball and a changeup. Trevor Hoffman, Jose Mesa, Aroldis Chapman and Keith Foulke all come to mind, though each of them had an excellent changeup and Joe Kelly’s rates out as a little better than average. Even still, there is hope that this approach could work for Kelly and that brings us to the next benefit of switching to relief.
By moving to the bullpen, Joe Kelly would be able to work exclusively out of the stretch and with simpler mechanics, he should see more consistency and have better command. The problem with this idea is that Kelly has been hit harder out of the stretch this year than he has when pitching from the windup. His OPS against with men on base is .837 versus a .771 with the bases empty. Neither one is good, so there is no indication that he pitches better out of the stretch. Being able to throw exclusively from the stretch is very different than throwing from the stretch when runners get on base, but if we’re looking for positive indicators, we don’t have anything concrete just yet. So far the best we can offer for the pro side of the breakdown is that he would be able to focus on his best two pitches, which hypothetically, would improve his performance. Nothing else we’ve looked at actually suggests he would see an improvement by moving to the bullpen.
So is there any concrete evidence that he would be better as a reliever? There is the fact that he looked good in relief in 2012 and 2013. In 2012 he posted a 2.30 ERA as a reliever with a 2.84 FIP and a 3.79 xFIP. He had a 9.19 K/9 and a 2.30 BB/9 as well and all of which were significant improvements over his innings as a starter. In 2013, his ERA sat at 3.65 as a reliever. His FIP was 4.07 and he had an xFIP of 3.65 with a K/9 of 8.03 and a BB/9 of 2.43. Some of those numbers were better than his innings as a starter, some were worse.
What that should tell you is that pitchers, especially relievers, will see a lot of fluctuations in their stats from season to season. It is a good sign, however, that his strikeout and walk rates remained consistent. More interestingly, they were significantly better than those rates as a starter. In 2012, he had a K/9 of 5.81 as a starter and his BB/9 was 3.15. In 2013, his K/9 was 4.76 and his BB/9 was 3.52. Whatever else is going on, Kelly has struck out significantly more batters and walked far fewer when pitching in relief over the course of his career. This actually flies in the face of the observation made above that his stuff would not play up in the pen. Even though he wouldn’t be throwing harder, something in the translation has lead to better strikeout and walk rates, and that is a big factor.
Unfortunately, the sample sizes we are working with are fairly small. He threw only 15 ⅔ innings in relief in 2012 and only 37 in 2013. Additionally, his K/9 rate as a starter has improved over the last two seasons sitting at 6.17 in 2014 and 7.54 in 2015. The improvements we see from him while pitching in relief could just be random variation.
Where does that leave us? We have a theory that focusing on his best two pitches (his sinker and his changeup) would increase his effectiveness. This makes sense intuitively and very well may be true, but we don’t have a sample of relief pitching where he threw those two pitches exclusively to look at, so we can only guess. It’s a reasonable guess, and when combined with the fact that he strikes out 2.20 more batters and walks 1.11 fewer per nine innings when pitching in relief, it’s easy to see why people are leaping to the conclusion that the Boston Red Sox should move him to the bullpen. But before we anoint him the successor to Koji Uehara, we may want to consider some of the downsides to removing him from the rotation.
We opened by pointing out that Joe Kelly has not been good this year. While true, it is not quite as simple as that. If he had enough innings to qualify, his FIP of 4.28 would tie him for 75th out of 92 pitchers in the majors and 0.44 above the major league average of 3.84. His xFIP of 4.02 (.18 higher than league average) would tie him for 65th with Trevor Bauer of the Indians who has a much more palatable ERA of 3.98. If Joe Kelly had Bauer’s ERA, there is very little chance we are talking about moving him to the bullpen in the first place. So, while there is no metric by which you can argue that Joe Kelly has been a good pitcher this year, there are indications that he’s already good enough to be a back of the rotation starter on a good team if his ERA normalizes.
If you need an example of someone who went through a similarly bad season and bounced back, look no further than teammate Clay Buchholz, who finished the 2014 season with a 5.34 ERA, a 4.01 FIP and a 4.04 xFIP. Bad luck seasons do happen. There’s no guarantee that Kelly will improve the way Buchholz did, but we need to look beyond his ERA if we are hoping to guess what lays in store for him in the future.
Without more concrete evidence that he is likely to be a better reliever, it becomes a bit of a gamble and the potential loss is huge. Pitchers who can throw 96-mph or harder are rare. Those that can maintain that velocity through 7 innings are rarer still. If the Red Sox can harness his potential, or even a significant portion of it, Joe Kelly will be a very good starting pitcher. With his strikeout rate as a starter climbing over the past three seasons, his FIP and xFIP being on the high side of acceptable already, and his ability to maintain elite velocity deep into games, the Red Sox would probably be best served by leaving Joe Kelly right where he is. He needs to be starting and with the season already lost, there is no reason he should not be doing it at the major league level.