Note: Statistics and PITCHf/x data exclude Moore's start yesterday vs. Boston.
Since Matt Moore has returned from the disabled list, he has been largely ineffective. In his first five starts, he posted a 7.61 ERA and a 13.3% strikeout percentage, which, if he had enough innings to qualify, would give him the sixth lowest strikeout rate in the league. His walk rate has climbed back up to 4.5 per nine innings, limiting his effectiveness and capping his potential upside, even if he could regain the strikeouts. He clearly hadn't been at his best, and was eventually optioned to Durham. Here, he can work out his problems in a low-leverage environment. After another poor start yesterday, he was optioned to triple-A Durham.
While trying to create projections for him before he returned would have been challenging, and likely futile, we can look at what he has done since his return to try to understand what has gone wrong.
The importance of consistency
While analyzing pitches based on their shape can lead to some meaningful conclusions, it is also important to take the consistency of a pitch into account. Meaning: Is the pitcher going to hang a slider, for example, half of the time, and then throw a devastating slider for the other half? Looking purely at overall shape, the pitch in this example might average out to a decent shape, but it doesn't tell the whole story. By looking at pitches on a granular, pitch-by-pitch level, we can begin to understand the consistency of a pitcher's different pitches.
We can't expect the same shape each time, because of both the human aspect of pitching and errors in PITCHf/x cameras, but if we set a historical baseline variance for a pitch's shape, we can look at more current data to see if the pitch is continuing to have the same amount of expected variance, or if the variance is more pronounced.
I was interested in applying this type of consistency analysis to Matt Moore, in an attempt to determine if his pitch shapes post-Tommy John surgery have the same level of consistency as they did before. To do this, I compiled Moore's raw PITCHf/x data for the 2013, 2014 and 2015 seasons, and then grouped the pitches by pitch type. After looking at the sample sizes and classifications, I determined I would be able to evaluate his four-seam fastball and changeup.
Ideally, I would have liked to analyze all his pitches. However, the sample size was not large enough for his sinker or his cutter, and there were some classification issues with his curveball that forced me to exclude it for now.
I then broke down the groups even further, as I put each pitch into a "2015 season", or a "2013-2014 season" bucket. Now, the fun begins.
As the image shows, Moore's velocity has been trending downwards throughout his career. This isn't too surprising, as aging curves show that average fastball velocity decreases almost immediately after the pitcher reaches the majors. Since his return, his fastball has had the lowest velocity of his career.
While the lower velocity will impact the results on his fastball, the movement contributes greatly as well. Here is a graph showing the horizontal movement of his fastball.
Moore's average horizontal movement this season appears to be in line with his career trends, at 7.54 inches. But, if we look at a histogram of the horizontal movement of his pitches, there is a change in the distribution of measurements.
While the general shape of the histogram is the same, and there is a large portion of fastballs with between 9 and 11 inches of horizontal movement, there is a large group of pitches in 2015 that have had around 5 inches of horizontal movement. I originally thought that this was a classification error, but looking at his arsenal, he doesn't have a pitch that is close to having five inches of vertical movement that could be confused with his fastball.
A break in the trend that is this large looks to be an example of an inconsistency in his pitch shape. This is further supported by a larger standard deviation for Moore's 2015 horizontal movement measurements - in 2015, the standard deviation is 3.21, while in 2013 and 2014, the standard deviation was 1.76. It is unclear if this is a common effect of Tommy John surgery or something specific to Moore, but more of his fastballs have been "flat" this season.
Here is a graph of the vertical movement of Moore's fastballs this season.
While his vertical movement has been inconsistent, it seemed to settle in during the 2013 season, and posted above average vertical movement recordings. A lot of vertical movement on a fastball can cause a batter to swing "under" it, as it drops less than others, and can contribute to higher rates of pop-ups and fly balls as well. It also can cause off-speed options to look like they drop more, which could increase the effectiveness of his secondary offerings.
In 2015, Moore has lost nearly a full inch of vertical movement on his fastball. This, combined with his lower velocity, makes his fastball look much more pedestrian. To see if this change in average vertical movement is legitimate, we can check the distribution of vertical movement measurements.
Here, the histograms look similar. Moore has thrown more pitches with less vertical movement in 2015, but there isn't the extreme shape disparity like with the horizontal movement histograms. The standard deviation of his fastball vertical movement measurements is slightly larger in 2015, moving from 1.50 to 1.87, but a difference this small is essentially negligible. Based on the distribution, it looks like the inconsistencies of his fastball shape come from the horizontal movement, and not the vertical movement.
While having consistency with pitches is good, the lower vertical movement will likely hurt Moore when considered with the inconsistent horizontal movement. Looking at the results this season, he has allowed three home runs so far, and all have come on fastball. Opposing hitter are also slugging .630 on this pitch this year. While it may not be disastrous in the long run, his fastball isn't as sharp as it once was.
Moore's changeup velocity has been all over the place during his career, but if we look at it on a year-by-year basis, we see that his changeup velocity is at a career low.
A lower changeup velocity, combined with his lower fastball velocity, maintains the velocity gap between the two pitches. Since velocity difference is positively correlated with generating whiffs on changeups, counteracting the fastball velocity decrease and keeping the velocity difference prevents Moore from losing whiffs on his changeup. However, velocity was positively correlated for groundball rate, so the slower changeup may cause him to generate fewer groundballs.
His horizontal movement, on the other hand, has increased this season.
Horizontal movement was insignificant in generating groundballs on changeups, so the increase may not have too much of an effect on the pitch. But, looking at histograms can indicate whether Moore has been able to get a consistent shape with it.
These histograms are a little harder to read. On one hand, the 2015 distribution is different because it is skewed, and the 2013-2014 distribution is nearly normally distributed. But, the range is similar and the sample size for the 2015 histogram is small, meaning that few pitches with a slightly different horizontal movement could drastically change the shape of the histogram. Considering the standard deviation of the horizontal movement measurements is similar in both 2015 (2.38) and 2013-2014 (2.77), I'm inclined to say that this isn't a case of inconsistency. However, the situation is certainly worth watching going forward.
While Moore's horizontal movement increased, his vertical movement has decreased.
Less vertical movement is positively correlated with both groundballs and whiffs, so this would help Moore's changeup be more effective. Looking at the histograms, the pitch has had consistent movement, as they are nearly identical.
The sample size is still small, so the lower vertical movement may not continue, but Moore's changeup looks to have a consistent shape in both its horizontal and vertical movement. This is encouraging, as he will need to rely more on his secondary offerings if the fastball continues to look flat.
So what does this all mean?
Moore has been bad, and his fastball has been a big part of his poor performance. Looking at its shape on a pitch-by-pitch basis, it looks inconsistent and has had little horizontal movement on a concerning number of his pitches. This has likely contributed to the success that opposing hitters have had on this pitch. His changeup, however, has had a consistent shape, and despite having a lower velocity this season should continue to be an effective pitch. It'll be important to continue to check on Moore's pitch consistency as the season progresses to see if he can solve the problem.
In the end, we had to expect some issues with Matt Moore when he returned from the disabled list. While some pitchers have had immediate success, like Jose Fernandez, most see some sort of decline in performance, and 20% never make another major league start. The Rays have plenty of adequate options in the rotation, so optioning Moore to Durham is for the best, although disappointing. While the Rays' playoff odds are shrinking, a strong second half for Moore would be encouraging as we start to look toward 2016.
PITCHf/x data is from BaseballHeatMaps.com and BrooksBaseball.net. Other statistics are from FanGraphs. Histograms made in R.