Quantifying Yu Darvish's Injury

There are a lot of poorly constructed narratives and talking points when it comes to Yu Darvish, and I have no interest in deconstructing or rebutting any of them. However, from watching nearly every single Darvish start since he came to the United States (mainly because of geography), I found it interesting, though perhaps not surprising, that Darvish was struggling with an injury, in this case, a nerve problem causing him to lose feeling in his leg.

For this post, I wanted to see if there was something in the Pitch F/X data that showed if/how the injury was hampering Darvish, and to look at his season as a whole. I decided, since I am looking at each outing individually, that I would only to look at his 2013 home outings. This way we won't be subject to any biases in Pitch F/X data from park to park (this is especially important when looking at release point and movement data).

Darvish made 19 home starts in 2013, and just to use a random sample size, his last 7 starts had a higher game score on average than his first 7 starts. He also averaged over a strikeout more and about a half a walk less per outing in those last 7 starts. His ERAs and FIPs of those starts were better in the last 7 than the first 7, so it is impossible to make the argument that Darvish faded down the stretch when it comes to effectiveness. However, it does seem that Darvish wasn't able to go as deep into games, as he averaged 2 less batters and 6 less pitches in his last 7 outings. Whether this was quicker hooks, general fatigue, or the nerve issue (or the back or neck issues which bothered him in Japan and off and on in Texas) isn't worth speculating about unless you've stolen Darvish's medical records or given him a physical. So instead, let's look at the data and see if there is any evidence that his injury was affecting how he was pitching.

First let's see where he has located the ball, on average, per start:

You probably wouldn't think it considering that Darvish was a high strikeout, high home run pitcher in 2013, but he is a pretty extreme low ball pitcher. He throws a lot of sliders, so perhaps it isn't too surprising, but he does stay arm side on average (he does change his slider location based on platoon split a little more than average). In September, you will notice that he was throwing the ball lower than he usually does, and two of those starts were more glove side than usual. His April 7th was his most glove side start. August, perhaps his best month (by wOBA against it was), was the month he consistently threw higher. He wasn't as effective in September (May was still probably worse), and it is easy to see the differences between the two months. Here we can see how Darvish's release point changed from start to start at home:

August shows up as the highest release point, something that makes sense with the highest average locations. Higher release points are usually better, and the statistics don't suggest anything different here. Darvish went on the DL for a short time in early July, and we saw that one of his late June starts had an average location stand out for being pretty heavily glove side. That particular start (the 18th) has a pretty normal release point compared to his others, but June 30th stands out for being closer to the center of the rubber, as does an early July start. Late April and early May is when his release point got the furthest from the center of the rubber and the lowest. His September starts don't stand out here, but his August starts do again.

The article linked to above does show that his fastball lost a little velocity at the end of the year. How about fastball movement?

The June 13th start stands out on the graph, but I believe this is due to an error. One of the fastballs had negative vertical movement, which doesn't make a lot of sense (though it was 91.5 MPH so I don't think it was a classification error, just a Pitch F/X error). Other than that, it appears that to be hard to find real trends. In September, he didn't have the lowest vertical movement starts, but he lacked the elite ones he got throughout the season. However, his September 14th start had the 2nd most horizontal movement of the home part of the season. I don't think this is particularly helpful.

It seems that the data doesn't explicitly show an injury late in the season for Darvish, as we don't see in either the release point or location data a clear difference in September from the rest of the months. That doesn't mean that the injury was not there, it just didn't show up in an obvious way in the Pitch F/X data. The biggest difference was actually August, a month where he was dominant. One can easily see the difference between August and September, but August seems to be the outlier, and would be the red flag month if you didn't know that he was struggling to finish games in September.