Not unlike the rest of us, the Red Sox are just tossing in the surf waiting for the inevitable to happen. Inevitable demise = fun! Their starting pitching is a huge reason why they're in this unenviable position, and as a starting pitcher for Boston, Joe Kelly is an individual part of that collective failure. Boston was depending on him to pitch like a major-league starter, to be serviceable, but mostly Kelly was just bad. But now, all of the sudden, he's pitching well! Kelly has always had potential if not results to spare. Does the past month cancel out the rest of the season before it? Is Kelly a good pitcher now and, if so, should he be considered a part of the Red Sox's rotation next season? Some people are saying yes. I'm saying slow down, some people!
On Aug. 1, Joe Kelly gave up five runs in five innings to Tampa Bay, the lowest-scoring team in the American League. That brought Kelly's ERA up to 6.11, the highest it had been all season. It's hard to call that the low point of Kelly's season because he previously pitched so badly he was sent down to Triple-A, so perhaps we can stipulate it to be one of multiple low points. What's worse: Slipping and falling into a cake, or the fact it was your boss' wedding? Tough choice! In Kelly's case we can ignore which is the lowest point, call it a bad season and move on to the rest of this article.
The lousiness of Kelly and his rotation-mates was not an insignificant point in now-ex-GM Ben Cherington getting replaced as head of baseball operations in Boston. Now, with new team president Dave Dombrowski in charge, the team is using the last few weeks of the season to assess players already on the roster in order to determine what must be done this offseason in preparation for next year. You might think the whole 6.11 ERA thing would have sealed Kelly's fate, but over his past five starts Kelly is 5-0 with a 1.69 ERA. So, hooray, right? This is the Joe Kelly the team traded for last season! The one with great stuff, the one who can be a contributing member of a major-league rotation on the cheap, the one who can grow a mustache that you might not laugh at immediately upon seeing it (but you should)!
People are already writing Kelly into next year's rotation because when we believe a player is capable of something and he goes out on the field and conforms to our beliefs, we tend to not look past those beliefs to any greater truths. "Joe Kelly: part of the failure of 2015" is now "Joe Kelly: part of the solution for 2016." But is this Joe Kelly really any different from the one we saw earlier this season?
I mentioned Kelly's 5-0 record and that's nice and sure looks pretty on the page. But as we know, pitcher wins and losses aren't indicative of how a pitcher pitched, let alone how he is going to pitch going forward. The same could be said of ERA, though in a less strident manner. Advanced metrics are better indicators, but for our purposes right now, let's stick with walks, strikeouts and ERA.
We know good pitchers strike out hitters, they don't walk many hitters, and they give up fewer runs. Now, here are two segments of Kelly's season.
| Time Frame | K% | BB% | HR% |
| Before August 6: | 18.5% | 8.2% | 2.7% |
| After August 6: | 17.6% | 8.0% | 2.4% |
HR% is the number of home runs allowed over the number of batters faced. All the above numbers are if not identical then very similar. Before Aug. 6, Kelly struck out almost a batter more per every 100 faced, but beyond that he was almost exactly the same pitcher. Now let's look at the same data again, but add one more category.
| Time Frame | K% | BB% | HR% | ERA |
| Before August 6: | 18.5% | 8.2% | 2.7% | 6.11 |
| After August 6: | 17.6% | 8.0% | 2.4% | 1.69 |
So that's weird, huh? Kelly is getting very similar results in the categories where we know pitchers have the strongest effect on the game, but the outcome is vastly different. The first pitcher is a candidate to get cut or sent down (Kelly did!) while the second is a Cy Young candidate, and yet they're both fundamentally the same guy. In fact, forget fundamentally, they are literally the same guy!
Now let's introduce one other stat: FIP, Fielding Independent Pitching. The FanGraphs glossary says FIP, "strips out the role of defense, luck, and sequencing, making it a more stable indicator of how a pitcher actually performed over a given period of time ..." compared to ERA. Here's Kelly's FIP for this season broken down by month alongside his ERA.
| Month | ERA | FIP |
| April | 4.94 | 3.55 |
| May | 6.53 | 5.04 |
| June | 5.23 | 3.71 |
| July | 8.31 | 6.82 |
| August | 2.68 | 3.64 |
Kelly's ERA has been all over the map, going from bad to really bad, to pretty bad, to horrendous, and then to pretty good. His FIP has been a bit more stable. The league-average starter has a 4.01 FIP this season, so you can see that by FIP Kelly has actually been an above-average starting pitcher for three of the five months. He was so bad the other two months, however, that he's still a below-average starter by FIP on the season.
So by FIP, Kelly's August isn't much different than his April or his June, two months that had been on the bad side of the ledger. In fact, it was June that was the straw that broke the "don't send Kelly to Triple-A" camel's back. In retrospect it was kind of a weak camel.
Kelly's August isn't much different than some of his previous months by FIP, or by more standard pitching stats like strikeouts, walks and home runs. But that's not to say Kelly isn't doing things differently. He is, and we can't discount that entirely. In late May, Eno Sarris did a piece for FanGraphs where he argued that Kelly needed to use his curveball more. He based that on the different pitch speeds hitters would be required to adjust to, the groundball tendencies of the pitch, and Kelly's ability to throw his curveball for strikes.
So has Kelly been throwing the curve more? Maybe that's the reason for the five-run difference in his ERA that isn't captured in his strikeout, walk and home-run numbers. Maybe he has been keeping the ball on the ground with his curve, which has led to weaker contact and better overall results.
| Pitch Type | Before Aug 6 | After Aug 6 |
| 4 Seam | 30.5% | 27.2% |
| Sinker | 39.4% | 29.0% |
| Change | 9.1% | 17.8% |
| Slider | 11.3% | 19.9% |
| Curveball | 9.7% | 6.1% |
Nope. Kelly has actually used his curveball less. In fact, it's now his least-thrown pitch. But he hasn't gone away from what Sarris recommended entirely because he has thrown more off-speed pitches and less hard stuff. Before he'd been throwing his fastball and sinker, both of which he throws at 96-97 mph, about 70 percent of the time. This is the crux of the Joe Kelly problem. He's throwing hard, really hard and you see that and you think, heck, no way I could hit that, no way that hitter could hit that. But those are major-league hitters, so of course they can hit that. They're major-league hitters. Major-league hitters need to be fooled by movement and location. Velocity is a tool, not an endgame in itself. If it were, this article would be entitled "Joe Kelly: Awesomer or Awesomest?" But it's not titled that. Kelly has ceded his pitch mix to that notion. Now he's employing more off-speed pitches. He still throws 56 percent hard stuff, but now he's throwing 43 percent movement pitches, 10 percent higher than he had previously.
The strange part about this is that Kelly has altered 10 percent of his arsenal and it hasn't materially affected his strikeout numbers, walk numbers or home-run numbers. He's the same guy throwing a slightly different mix of pitches. His ERA is vastly different, but his FIP isn't fooled. He's got a mustache now, so you can't ignore that even if you want to, but beyond that and beyond the specifics of his arsenal of pitches, Joe Kelly is still Joe Kelly. The only difference is the results, and there might not be anything Kelly can do about those. That's something the Red Sox might want to keep in mind when planning for next season.
