Predicting the future performance of baseball players is a wildly frustrating endeavor. When projecting a hitter’s stat line, it’s impossible to account for the player having a BABIP that’s nowhere close to their career average. When projecting a pitcher, it’s hard to know when their HR/FB rate is going to swing drastically away from their career average. The shifts in “luck” make projecting seem like a fool’s errand at times.
But it may be even more frustrating when something within the player’s control changes completely. When predicting the future, you largely have to rely on the assumption that what the player puts into the equation will remain relatively the same. But of course that’s not the case. Hitters will lose plate discipline out of nowhere. Pitchers will lose the ability to induce swings and misses. It just happens. But it’s often very hard to know when those changes are going to happen.
Another example of the input changing is a shift in the mix of pitches a pitcher will use.
For example, when I projected Cole Hamels as a top ten pitcher, I was operating under the assumption that he would continue to use his changeup about 30% of the time like he did last year seeing as how his change is by far his best pitch according to our pitch values. And my assumption was even stronger that he would continue to use his cutter around 10% of the time like he has for the last three years, even though the pitch wasn’t as effective last year as it was in 2011. But those assumptions were wrong. Hamels is throwing the change about 25% of the time and the cutter about 17% of the time.
Hamels is presumably throwing the change less often because he is struggling with his control of the pitch. For his career, his change has been called a ball 26.61% of the time he throws it. But this year it’s being called a ball exactly 11% of the time more (37.61%). But it doesn’t really matter that he’s throwing it less because his control of it is off. We couldn’t have expected his control of the pitch to decline so suddenly. When the inputs change without warning, it makes predictions look silly.
I decided to take a look and see whether the pitchers who have the biggest differences between their 2012 and 2013 xFIPs have made significant changes to their pitch mix. I used the PITCHf/x percentages here on the site and calculated the difference in usage between 2012 and 2013 for all pitchers who are currently qualified and who also had 100+ IP last season.
What I found was that almost all of those who have improved the most have made some significant changes. But those struggling seem to be using a similar mix and are suffering from other problems. Aside from Hamels, Jon Niese is the only other potentially useable starter who has seen a significant change in pitch usage and has seen a decline in skill.
Below are the pitchers with the biggest improvements in xFIP along with the differences in pitch usage. The chart may look like it is cut off, but changeups are the last column.
Pitch Usage Gap
Of that group, Ubaldo Jimenez, Homer Bailey and Ryan Dempster are probably the buy-lows. Everyone else on the list other than Peavy has an ERA under 3.00, and Peavy’s ERA isn’t too far above his xFIP. But Ubaldo’s ERA is almost two runs higher than his xFIP, Dempster’s is about three-quarters of a run higher, and Bailey’s is about half a run higher.
There are some differences between the pitch classifications here at Fangraphs and those over at BrooksBaseball.com. Brooks has Ubaldo throwing more splitters and sinkers this year as opposed to changeups like we do. But however you want to classify it, he’s throwing more balls with sink and fewer traditional four seamers. Given the precipitous drop in velocity he’s experienced in the last few years, this probably isn’t a bad idea.
Admittedly, his control is still awful. But his groundball and strikeout rates look more like they did back when he didn’t suck. His current strikeout rate (26.2%) is likely to regress, but his swinging strike rate is up a little from last year, so he should be able to continue to strike out hitters at an above average rate. I couldn’t blame you for refusing to consider him, but I believe he’s going to have some good starts in the future. However, he is home to Detroit and at Cincinnati in his next two starts. You might want to give him another look in June.
Dempster has added a cutter to the mix at the expense of his two-seamer. He’s also upped his four-seamer and splitter usage at the expense of his slider. And even though his slider is a plus pitch for him according to our pitch values, it’s probably good that he’s not throwing it an absurd 40% of the time like he did last year. This is another situation where classifications may not be exact because we have him ditching his change after 2012 and Brooks never really had Demp throwing a change. But his splitter has roughly the same velocity as what was classified as his change here and obviously has similar movement.
I didn’t realize it until I delved a little deeper into Bailey’s numbers in order to write this paragraph, but like Ubaldo and Dempster, Bailey has also increased the usage of his sinking pitches. Our classifications have him throwing fewer four-seamers and more two-seamers, while Brooks has him throwing more sinkers and splitters. Either way, all three of these buy low candidates have shifted towards more pitches with sink. And they’ve all seen increases in their swinging strike rates with Ubaldo and Bailey also seeing big gains in their groundball rate.
One thing these three have not seen is their ERAs following their improved xFIPs. And given a noticeable shift in their pitch usage, it’s reasonable to assume that their improvements in skill are legitimate to some degree and that their ERAs should eventually move toward their xFIPs.