Explaining the Recent Success of Randall Delgado and Trevor Cahill

When the season began, Trevor Cahill and Randall Delgado made up two-fifths of the Diamondbacks’ starting rotation.  By mid-April both had been relegated to the bullpen. Cahill was likely on a short leash after an inconsistent 2013 campaign. Similarly, the club indicated that they did not envision Randall Delgado as a starter; he was not projected to crack the rotation until Patrick Corbin went down for the year. They’ve been in the bullpen for almost two months now, and they have been pitching significantly better as relievers.

In looking at the statistics of both pitchers, I looked at the career results of each as a starter and reliever. Delgado has logged the majority of his innings as a starter, posting a 4.72 FIP, 16.2 K%, and 8.0 BB%. Although he has only pitched 24.1 career innings out of the bullpen, he posted a 3.40 FIP, 31.8 K%, and 10.9 BB%. The only issue has been the walks, but the dramatic increase in strikeouts makes the walks tolerable. Cahill’s statistics as a reliever also look much rosier. As a starter, he had a 4.36 FIP, 15.7 K%, and 9.2 BB%. Through 27.2 career innings as a reliever, he has a 2.85 FIP, 24.8 K%, and 11.1 BB%. Same story here—the strikeouts are up dramatically, but the walk rate has gotten even worse.

Based on the statistics, it seems like both men have improved their pitching. But we have to consider that it’s easier to be a reliever than a starter. The average ERA for major league starters this season has been 3.94, and the average for relievers has been 3.61. That’s despite the fact that teams generally use their better pitchers as starters. So far, in 2014 the average starter strikes out 7.42 batters per nine, while the average reliever strikes out 8.44 per nine. This discrepancy exists for a variety of reasons. Two of the big reasons are that relievers throw fewer pitches so they can throw harder, and relievers do not have the face the same batter multiple times.

This problem has been examined before. Noted baseball statistician Tom Tango examined all pitchers who pitched as both starters and relievers. On average, when a starter converts to a reliever, their BABIP should be 17 points lower, HR% should be 14.53% lower, and K% should be 17% higher. Relief pitchers are also supposed to give up about one less run per nine innings. Although starters walked slightly fewer less batters than relievers, Tom Tango insisted that “Walk rate is FLAT.” Our circumstantial evidence suggests otherwise.

So far, it seems like our converted relievers have outdone themselves. Delgado’s FIP is more than a run lower and he’s striking out batters at twice the rate. Cahill’s FIP decreased by 1.51 and his K% is about 66% higher. (I am ignoring HR% because that has the highest tendency to get skewed by a small sample size.) Interestingly enough both men have also seen increases of more than 30 points in their BABIP, the opposite of what was supposed to happen. Both of their walk rates have increased as well.

Cahill and Delgado are doing better in their new roles than the average pitcher who has made the switch. Why is this so? They may simply be on hot streaks that will regress as time passes — but there are other reasons to think these two men are benefitting from the switch to the bullpen more than the average pitcher.

One theory is that these two guys were suffering disproportionately when facing batters a second or third time, a phenomenon we’ve seen with Josh Collmenter. This is usually the case when pitchers have two good pitches but not three. Randall Delgado falls squarely into this category, as he is primarily a fastball/changeup guy. The statistics support this theory, as batters’ OPS jumped over 100 points between the first and second time through the order. Strangely enough, Delgado does not get worse between his second and third times through the order. Cahill’s progression is more typical, and his improvement likely does not fall under this theory.

Another theory is that the statistics are flawed. As with any study, there can be certain biases and margins for error. In this instance, it’s possible that many of the data points are from relievers that successfully become starters. The ones who failed likely would quit the attempted conversion after a trial period.

There are many moving parts here, and there are even more factors that could explain the dramatic changes. A pitcher’s mindset and splits against lefties and righties could also be relevant. At this point, there are too many potential possibilities. We’ll just have to see how they pan out.