The other day I was thinking about Joe Biagini and his role going forward. While he showed flashes and the raw stuff to be a solid rotation stalwart in the future, the obvious need was for much better consistency. A thought occurred to me that there were parallels to J.A. Happ in his first go-around with the Blue Jays.
There has been a lot written of the changes in Happ from that first stint in 2012-14 to his second stint over the past two years (mainly being more aggressive and relying on his stuff), but in terms of results it seems to me the biggest one is that he’s more consistent. After all, I can recall numerous starts back then that were quite dominant...but many more numerous starts of 100 pitches over five innings.
So I wanted to see whether the data supported. From 2012-14, Happ made 50 starts for the Jays (and eight relief outings, which will be ignored) and over the past two seasons he’s made 57, so conveniently the two samples are similarly sized.
There’s no perfect metric to measure individual starts, but I used the updated version of Game Score from FanGraphs. As a reminder, a Game Score of 50 is considered a quality start, and the higher the number the better. For both periods, I divided games into quartiles by Game Score and took the average:
In his first stint, Happ averaged a 72 Game Score over his top quartile (“best starts”). That’s actually really, really good. So good, that his best starts over the last couple years are as a group only marginal better at an average Game Score of 74.
Looking at the bottom quartile (“worst starts”), there is some improvement from averaging 28 to 36. But we’re still talking pretty bad starts overall, and while there is some value in having one’s worst starts be less worse, generally it’s not going to make a huge difference.
But the biggest difference is in the middle. The second quartile results improve from 54 to 63, or from decent to very good. The third quartile results improve from 40 to 53, or basically from poor starts on average to good starts.
This is really where the difference in Happ manifests itself. Instead of ranging from bad-to-okay starts in his median outcomes, he’s ranging from decent-to very-good starts. And that makes a huge difference in the team’s likelihood of winning.
Here’s what it looks like visually plotting Happ’s starts by Game Score in ascending order:
That big bulge from roughly the 20th to the 75th percentile? That’s the more consistent J.A. Happ. He was always capable of dominant, and he still gets shelled the odd time. But he’s much better in between.
To put it another way, the old Happ posted a quality start (50+ Game Score) in only 24 of 50 starts, but averaged a Game Score of 63.8 when he did. The new Happ has posted a quality start in 42 of 57 starts, with a very similar average of 63.3. When he’s on, Happ is no better on average than he used to be. Just he’s on more like 75% of the time compared to 50% of the time.
And that’s the difference between a good mid-rotation starter, and a fringe backend starter. Take notes, Unique Joe.