I feel like it’s become a tradition on this site to post articles at the end of April which throw cold water on the hot start to the season from certain Pirates pitchers. This tradition dates back to when Jeff Locke was breaking out as a starter in 2013, and has continued any time we’ve seen the chance for regression from an unexpected start.
Typically, the focus is on one pitcher. As I wrote yesterday, the Pirates have seen good results from their rotation from an ERA and FIP standpoint, but have some concerns with the xFIP numbers.
The only difference between FIP and xFIP is that xFIP normalizes the pitcher’s home run per fly ball ratio. The Pirates are currently at 8.8%, while the league average is at 14.5%.
Some people prefer FIP and feel that xFIP is ridiculous because of the importance placed on the HR/FB ratio. I feel it’s just as ridiculous to say that FIP is legitimate (which helps to normalize strand rates and batting average per balls in play), but that normalizing a home run total is going too far.
The formula for FIP, as detailed on FanGraphs, is the following:
FIP = ((13*HR)+(3*(BB+HBP))-(2*K))/IP + constant
The formula only factors in the stats that a pitcher can control, based solely on the batter vs pitcher battle, with all outside factors removed. The only difference with xFIP is that it doesn’t assume that a pitcher has total control over his home run totals.
That’s not always true. Some pitchers have a HR/FB rate that is constantly higher than normal, while some have a HR/FB rate that is constantly lower. Just like any stat, you’d take those individual circumstances and adjust your analysis. In those cases, I would prefer FIP to xFIP. But that’s only after a pitcher has a track record of going against the average. I wouldn’t assume that a pitcher doing well with a low HR/FB rate is the exception until he shows that he’s the exception over several years.
The HR/FB rate for pitchers typically stabilizes at around 400 fly balls. That’s a lot. Jameson Taillon has 391 so far in his MLB career, spanning 459.2 innings. Taillon’s HR/FB rate is at 11.8% on his career, but 8% this year.
I bring all of this up because if you are only focusing on FIP, the starting rotation looks great. The group currently ranks fourth in the majors, and second in the NL just behind the Reds. If you switch over to xFIP, the Pirates rank 15th in baseball, and 10th in the NL.
This is a problem, considering the rotation is the strength of this team, and their biggest hope of making the playoffs. With their position player struggles, they need a top five rotation to have a shot at competing.
Here is the breakdown of the rotation, sorted by ERA:
Jordan Lyles: 2.05 ERA / 3.67 FIP / 4.36 xFIP
Joe Musgrove: 2.18 / 2.52 / 3.94
Trevor Williams: 3.38 / 3.53 / 4.48
Jameson Taillon: 4.06 / 3.12 / 3.79
Chris Archer: 4.33 / 4.83 / 4.71
A few things instantly stand out here. First is that there is a big expected regression going forward for Lyles, Musgrove, and Williams. Taillon has actually been slightly better than his ERA indicates, and is the best pitcher in the rotation by xFIP right now. Archer has struggled, and his xFIP shows that he should be lower than his current ERA.
We’ll focus on the three guys who project to see negative regression.
Joe Musgrove
Musgrove is looking great on the surface, but the xFIP gives a red flag. He currently has a 3.1% HR/FB ratio, while his career total is 12.3%. Even if we say that last year’s 10.2% is legitimate, he still has a long way to drop.
The good news here is that Musgrove is doing the things you need to do in order to counter the HR/FB regression. He’s striking out a little less than a batter an inning, and limiting his walks to a 1.8 BB/9.
The league average ERA is 4.35 and the league average xFIP is 4.31. Even if Musgrove regresses to the xFIP number, he’ll be above average. If he can maintain a “lucky” HR/FB ratio, he’ll be in great shape. But by that, I mean a few points off his career totals, and not the extreme difference we’re seeing now.
Overall, Musgrove isn’t someone I’d be concerned about right now, as he’ll still be productive, even with regression.
Trevor Williams
Williams was the featured pitcher in this analysis last year around this time. That led to the usual “What if he’s the exception?”, followed in May and June with “I guess he wasn’t the exception.”
The interesting thing is that Williams went back to a lucky HR/FB ratio in July (4.3%, leading to a 2.33 ERA and a 5.90 xFIP). He maintained that in August (7.7%, 1.16, 4.66), and allowed zero homers in September, which helped to balance some unlucky BABIP and strand rates.
In his career, Williams has performed better than most with his HR/FB ratio. He’s at 9.4% in his career, spanning 384 fly balls, which means it’s a pretty stable sample size. He was at 8% last year, and is at 7.5% this year.
So the regression doesn’t look that bad if you assume he’ll stop at his career totals, and could be minimal if he repeats last year’s total. This would be a case where a pitcher has displayed a trend where he posts better than average HR/FB rates, making his FIP a more reliable measure.
While the xFIP suggests a big regression, I could see Williams remaining below a 4.00 ERA, especially since he’s shown good career numbers with his BABIP (.277) and strand rate (73%). He should see a bit of regression from his current ERA, but I don’t think it will be as extreme as the xFIP indicates.
Jordan Lyles
Lyles is off to a great start this year, with a 2.42 ERA in his first five starts. That’s actually inflated by his outing last week, and for the most part he has been shutting down opposing lineups. However, he’s also one of the biggest regression candidates in the pitching staff right now.
Lyles has a 3.65 FIP and a 4.45 xFIP. The FIP is much lower due to his unsustainable strand rate, currently at 85%, and sitting at 100% prior to his last two starts. That will continue to come down to earth and settle around the 70% range (his career is actually 65%, so hopefully he improves on that), which will result in more runs and a higher ERA.
The HR/FB ratio is at 9.4%, while he’s been at 12.8% in his career, and was at 12.7% last year. So you can expect more home runs than we’ve already seen.
A key for Lyles going forward would be to cut down on the walks — currently not horrible at 8.8%, but higher than the last two years at 7.6 and 6.8% — and to reduce his fly balls to reduce the home run impact. He’s got a 48.5% fly ball rate this year, which is way up from 37.3% last year, and up from 30.7% in his career.
This could be explained by pitch type. He’s throwing his four-seam and curve more often, and his sinker and slider less often. The sinker was one of his main pitches when he had low fly ball totals. He dropped the usage last year and the fly ball numbers went up, and continued that approach this year only to see those numbers go up again.
This is a challenge, because you could argue that the increased usage of his curveball, which is his best pitch and plays off the four-seam, has led to his success. That successful approach also brings on risk with more home runs. But Lyles can still find a way to reduce the fly balls, even with this approach. His current fly ball percentage on fastballs is 62%, up from 52% last year, and 37% in his career.
If he can reduce the fly balls, and maybe cut out a few walks, he could really negate the impact of his pending regression.
