clock menu more-arrow no yes

Filed under:

Fastball/changeup: An unhappy marriage

New, 10 comments

The relationship between a pitcher's fastball and changeup might not actually be as harmonious as we think.

Kris Medlen (now a member of the Kansas City Royals) has one of the best changeups while simultaneously throwing a terrible fastball.
Kris Medlen (now a member of the Kansas City Royals) has one of the best changeups while simultaneously throwing a terrible fastball.
Dale Zanine-USA TODAY Sports

Quick question: most good changeups start with the pitcher having a good fastball, right?

This seems to be, more often than not, common sense. Both pitches don't need to be great, but both need to be good and it doesn't hurt if one of them is great. A fastball declares a pitcher's mound presence, while the changeup does exactly as advertised: changes pace on the batter.

Beyond the Box Score's Justin Schultz recently posted an article about the Philadelphia Phillies. More specifically, he wrote about Cole Hamels and how his changeup is one of the best in the league. Justin brought up a very interesting point though: Hamels' fastball isn't that great. In fact, over the past two seasons (424.1 IP), Hamels' fastball has been worth 0.46 runs below average per 100 fastballs thrown (-0.46 wFA/C).

Is it possible that we have been thinking about the relationship between fastball and changeup all wrong? There are some assumptions we make about the relationship between these two pitches: 1) Fastballs get thrown more often than changeups (this one really is almost always true); 2) If a batter shows signs that he's catching up to the fastball, throw the changeup; 3) In fastball counts, sometimes it's best to throw a changeup, expecting the batter to swing through before the ball actually arrives. Obviously these are not steadfast, but they are general in explaining to a casual baseball fan how the two pitches work together.

But maybe they don't work together this way. I mean, if Hamels is any indication, maybe his superior changeup is a product of having a bad fastball. The more I thought about it, the more I realized this actually seems intuitive; that one of these pitches be below average, or at least noticeably worse than the other. For instance, if you were known for throwing a meaty fastball, wouldn't that help your changeup? More explicitly, if batters came up already geared for your fastball, wouldn't your changeup benefit? Were there other pitchers like Hamels? My guess is, implicitly, yes. How about vice versa? Could a pitcher's fastball positively benefit from a pitcher's changeup being worse?

To look into this, I gathered the top 30 fastballs and changeups by wFA/C and wCH/C respectively over the past five seasons. I then looked at the relation between a good fastball and the same pitcher's changeup, and vice versa. To be included in this analysis, the pitcher must have pitched for 400 innings or more and, during that time, thrown at least 35% fastballs (FB%) and 5% changeups (CH%). Let's start with the pitchers who had the top 30 changeups over the last five years. Here is how their changeups look in relation to their fastballs:

<script type='text/javascript' src='https://public.tableausoftware.com/javascripts/api/viz_v1.js'></script><div class='tableauPlaceholder' style='width: 1370px; height: 543px;'><noscript><a href='#'><img alt='Sheet 1 ' src='https:&#47;&#47;publicrevizit.tableausoftware.com&#47;static&#47;images&#47;wC&#47;wCH&#47;Sheet1&#47;1_rss.png' style='border: none' /></a></noscript><object class='tableauViz' width='1370' height='543' style='display:none;'><param name='host_url' value='https%3A%2F%2Fpublic.tableausoftware.com%2F' /> <param name='site_root' value='' /><param name='name' value='wCH&#47;Sheet1' /><param name='tabs' value='no' /><param name='toolbar' value='yes' /><param name='static_image' value='https:&#47;&#47;publicrevizit.tableausoftware.com&#47;static&#47;images&#47;wC&#47;wCH&#47;Sheet1&#47;1.png' /> <param name='animate_transition' value='yes' /><param name='display_static_image' value='yes' /><param name='display_spinner' value='yes' /><param name='display_overlay' value='yes' /><param name='display_count' value='yes' /><param name='showVizHome' value='no' /></object></div><div style='width:1370px;height:22px;padding:0px 10px 0px 0px;color:black;font:normal 8pt verdana,helvetica,arial,sans-serif;'><div style='float:right; padding-right:8px;'><a href='http://www.tableausoftware.com/public/about-tableau-products?ref=https://public.tableausoftware.com/views/wCH/Sheet1' target='_blank'>Learn About Tableau</a></div></div>

Three notable outliers sit atop the graph: Jorge de la Rosa, the aforementioned Cole Hamels, and Kris Medlen (2.49, 2.31, and 2.24 wCH/C respectively). These three pitchers decisively have the best changeup over the past five seasons. It is also important to note that, of these three, Cole Hamels is the only one with an above-average fastball over this five year sample (despite it being below average for the past two seasons as previously stated). To illustrate how narrowly Hamels’ fastball is above-average, the median fastball is Roy Oswalt’s at -0.01 wFA/C while Hamels sits four hundredths above at 0.03 wFA/C. Kris Medlen on the other hand has the worst fastball (-1.00 wFA/C) of the entire group, while maintaining the third best changeup. This is not necessarily proof-positive, however. In these three cases there is evidence that an average-to-terrible fastball, at the very least, doesn’t negatively affect the same pitcher’s changeup.

Of course, Felix Hernandez deflates this theory a bit, but there’s a reason he’s called the King. He and Johnny Cueto are the only pitchers who appear in the top ten of both the best fastball and best changeup lists. Jeff Sullivan wrote an article recently that suggests Hernandez’s changeup has steadily improved into the best pitch in all of baseball. Throw Stephen Strasburg and Alex Cobb in there and you have four pitchers with outstanding fastballs and changeups. In fact, if you remove the three outliers at the top (de la Rosa, Hamels and Medlen), the naked eye can notice a trend that seems to start at the lower left with Tommy Milone and goes to the upper right near Chris Sale. That line of best fit would look like this:

<script type='text/javascript' src='https://public.tableausoftware.com/javascripts/api/viz_v1.js'></script><div class='tableauPlaceholder' style='width: 1370px; height: 543px;'><noscript><a href='#'><img alt='Sheet 1 ' src='https:&#47;&#47;publicrevizit.tableausoftware.com&#47;static&#47;images&#47;wC&#47;wCHLineofBestFitOutliersOmitted&#47;Sheet1&#47;1_rss.png' style='border: none' /></a></noscript><object class='tableauViz' width='1370' height='543' style='display:none;'><param name='host_url' value='https%3A%2F%2Fpublic.tableausoftware.com%2F' /> <param name='site_root' value='' /><param name='name' value='wCHLineofBestFitOutliersOmitted&#47;Sheet1' /><param name='tabs' value='no' /><param name='toolbar' value='yes' /><param name='static_image' value='https:&#47;&#47;publicrevizit.tableausoftware.com&#47;static&#47;images&#47;wC&#47;wCHLineofBestFitOutliersOmitted&#47;Sheet1&#47;1.png' /> <param name='animate_transition' value='yes' /><param name='display_static_image' value='yes' /><param name='display_spinner' value='yes' /><param name='display_overlay' value='yes' /><param name='display_count' value='yes' /><param name='showVizHome' value='no' /></object></div><div style='width:1370px;height:22px;padding:0px 10px 0px 0px;color:black;font:normal 8pt verdana,helvetica,arial,sans-serif;'><div style='float:right; padding-right:8px;'><a href='http://www.tableausoftware.com/public/about-tableau-products?ref=https://public.tableausoftware.com/views/wCHLineofBestFitOutliersOmitted/Sheet1' target='_blank'>Learn About Tableau</a></div></div>

This line of best fit seems to indicate that my hypothesis is wrong. But, as you know, we can't simply omit outliers without good reason so the actual trend line dips a bit downward and looks like this:

<script type='text/javascript' src='https://public.tableausoftware.com/javascripts/api/viz_v1.js'></script><div class='tableauPlaceholder' style='width: 1370px; height: 543px;'><noscript><a href='#'><img alt='Sheet 1 ' src='https:&#47;&#47;publicrevizit.tableausoftware.com&#47;static&#47;images&#47;wC&#47;wCHLineofBestFit&#47;Sheet1&#47;1_rss.png' style='border: none' /></a></noscript><object class='tableauViz' width='1370' height='543' style='display:none;'><param name='host_url' value='https%3A%2F%2Fpublic.tableausoftware.com%2F' /> <param name='site_root' value='' /><param name='name' value='wCHLineofBestFit&#47;Sheet1' /><param name='tabs' value='no' /><param name='toolbar' value='yes' /><param name='static_image' value='https:&#47;&#47;publicrevizit.tableausoftware.com&#47;static&#47;images&#47;wC&#47;wCHLineofBestFit&#47;Sheet1&#47;1.png' /> <param name='animate_transition' value='yes' /><param name='display_static_image' value='yes' /><param name='display_spinner' value='yes' /><param name='display_overlay' value='yes' /><param name='display_count' value='yes' /><param name='showVizHome' value='no' /></object></div><div style='width:1370px;height:22px;padding:0px 10px 0px 0px;color:black;font:normal 8pt verdana,helvetica,arial,sans-serif;'><div style='float:right; padding-right:8px;'><a href='http://www.tableausoftware.com/public/about-tableau-products?ref=https://public.tableausoftware.com/views/wCHLineofBestFit/Sheet1' target='_blank'>Learn About Tableau</a></div></div>

This lends itself nicely to the idea that a good changeup might actually be helped by a below average fastball. Could this relationship work the other way as well though? Could a bad changeup actually help a pitcher's fastball succeed? This seems far less intuitive, but here's the graph:

<script type='text/javascript' src='https://public.tableausoftware.com/javascripts/api/viz_v1.js'></script><div class='tableauPlaceholder' style='width: 1370px; height: 543px;'><noscript><a href='#'><img alt='Sheet 1 ' src='https:&#47;&#47;publicrevizit.tableausoftware.com&#47;static&#47;images&#47;wF&#47;wFA&#47;Sheet1&#47;1_rss.png' style='border: none' /></a></noscript><object class='tableauViz' width='1370' height='543' style='display:none;'><param name='host_url' value='https%3A%2F%2Fpublic.tableausoftware.com%2F' /> <param name='site_root' value='' /><param name='name' value='wFA&#47;Sheet1' /><param name='tabs' value='no' /><param name='toolbar' value='yes' /><param name='static_image' value='https:&#47;&#47;publicrevizit.tableausoftware.com&#47;static&#47;images&#47;wF&#47;wFA&#47;Sheet1&#47;1.png' /> <param name='animate_transition' value='yes' /><param name='display_static_image' value='yes' /><param name='display_spinner' value='yes' /><param name='display_overlay' value='yes' /><param name='display_count' value='yes' /><param name='showVizHome' value='no' /></object></div><div style='width:1370px;height:22px;padding:0px 10px 0px 0px;color:black;font:normal 8pt verdana,helvetica,arial,sans-serif;'><div style='float:right; padding-right:8px;'><a href='http://www.tableausoftware.com/public/about-tableau-products?ref=https://public.tableausoftware.com/views/wFA/Sheet1' target='_blank'>Learn About Tableau</a></div></div>

Again, we start with a glaring outlier, this time it's Adam Wainwright. His fastball is a bewildering 1.97 wFA/C while his changeup is below pedestrian as far as major leaguers go at -0.79 wCH/C. To give you an idea of what a pedestrian changeup would look like, the median in the same sample is Randy Wolf's, which is worth -0.16 wCH/C.

<script type='text/javascript' src='https://public.tableausoftware.com/javascripts/api/viz_v1.js'></script><div class='tableauPlaceholder' style='width: 1370px; height: 543px;'><noscript><a href='#'><img alt='Sheet 1 ' src='https:&#47;&#47;publicrevizit.tableausoftware.com&#47;static&#47;images&#47;wF&#47;wFALineofBestFit&#47;Sheet1&#47;1_rss.png' style='border: none' /></a></noscript><object class='tableauViz' width='1370' height='543' style='display:none;'><param name='host_url' value='https%3A%2F%2Fpublic.tableausoftware.com%2F' /> <param name='site_root' value='' /><param name='name' value='wFALineofBestFit&#47;Sheet1' /><param name='tabs' value='no' /><param name='toolbar' value='yes' /><param name='static_image' value='https:&#47;&#47;publicrevizit.tableausoftware.com&#47;static&#47;images&#47;wF&#47;wFALineofBestFit&#47;Sheet1&#47;1.png' /> <param name='animate_transition' value='yes' /><param name='display_static_image' value='yes' /><param name='display_spinner' value='yes' /><param name='display_overlay' value='yes' /><param name='display_count' value='yes' /><param name='showVizHome' value='no' /></object></div><div style='width:1370px;height:22px;padding:0px 10px 0px 0px;color:black;font:normal 8pt verdana,helvetica,arial,sans-serif;'><div style='float:right; padding-right:8px;'><a href='http://www.tableausoftware.com/public/about-tableau-products?ref=https://public.tableausoftware.com/views/wFALineofBestFit/Sheet1' target='_blank'>Learn About Tableau</a></div></div>

This one is a lot more difficult to discuss, but there does seem to be a slight relationship between the best fastball pitchers having a great fastball if their changeup is also better. One could suppose that if, Matt Garza for instance, worked on his changeup, his fastball could get even better. This is a lot less intuitive to me for a couple reasons: 1) how often are pitchers known as changeup pitchers? 2) How does a batter wait on a changeup? If they aren't offering at the fastball, the pitcher will continue to throw the fastball.

What I liked about this exercise was seeing all the pitchers that got omitted because of being slightly below a certain hurdle. For instance, both Garrett Richards and Shelby Miller were both omitted due to their inexperience (398.2 and 370.0 IP respectively). Both have above average fastballs (0.62 and 0.55 wFA/C) and abysmally bad changeups (-6.29 and -2.02 wCH/C). I think, what this research does, is help their case as young, developing pitchers. The biggest takeaway for me is to not put too much stock into a player's scouting profile. If their changeup is 70 and their fastball is 50, maybe it doesn't matter as much as scouts would have us believe. Obviously this isn't proof of anything, but I believe it to be a step in the right direction to understanding that a pitcher's changeup doesn't necessarily rely on their fastball dominance.

. . .

All statistics courtesy of FanGraphs.

Michael Bradburn is a Featured Writer for Beyond the Box Score. You can follow him on Twitter at @mwbii. You can also reach him at michaelwbii@gmail.com