Was Colby Lewis Really That Much Better In The Second Half of 2014?
The predominant thought and narrative surrounding Lewis’ 2014 season was he was horrible in the first half due to just returning from a long absence, and as he became more comfortable as his body readjusted to toeing the rubber every fifth day, Lewis became more productive in the second half.
Tom Tango popularized a metric, kwERA, which is similar to FIP, but strips home runs out of the equation and instead utilizes only strikeout and walk rate. In a Hardball Times piece from 2012, writer Glenn DuPaul found that when trying to predict a pitcher’s second half performance, strikeout and walk differential, and not ERA, FIP, or XFIP was the better indicator of future performance. As DuPaul points out, since kwERA utilizes strikeout and walk differential, kwERA indeed accomplishes the goal of better predicting second half performance. Despite that horrendous 6.54 ERA in the first half of ‘14, Lewis’ first half K%-BB% was actually 11.8%, which when used to calculate his kwERA, leads one to calculate a first half kwERA of 3.98. As mentioned above, Lewis’ second half ERA was…….3.86. Lewis might have been getting hit hard, but the strikeout and walk numbers were at least palatable. The following chart will now examine Lewis’ batted ball profile from 2010 through 2012, and 2014.
So, I guess that should conclude my piece. Lewis was not as awful in the first half as his ERA suggests, and his second half resurgence should be accredited to regression. Well, not exactly. As the author mentioned above, the pitcher still has to hold himself accountable for some of his poor performance. Let’s take a look at some of what Lewis did to improve his chances for regression in the second half. The charts were obtained for the magnificent Brooks Baseball.
A few interesting items of note is while the narrative is Colby pitched much better in the second half due to his lower ERA, one could argue Colby actually pitched better in the first half despite the bloated ERA over 6.00. Lewis allowed five more long balls in the second half, and his FIP and xFIP were both lower in the first half while he was being shelled. There have been numerous articles in the past explaining why ERA is not very reliable in determining overall pitcher performance, and one believes Lewis’ first half is pretty indicative of this. While the author will cede Colby was mechanically and strategically more effective in the second half, the fact his ERA was three full runs worse before the break does not necessarily support the fact his first half was far worse.
Ranger fans have been enamored with Colby Lewis since he returned to the team in 2010 as he was superb for them in the playoffs in both ’10 and ’11, and his presence on the mound reminds us of the most successful period in franchise history. His resilience and willingness to stay in Texas are both admirable, and make rooting for the guy fairly easy. However, relying on Lewis this year due to his ’14 second half might be fairly difficult. Steamer projects Lewis to yield a 4.67 ERA and 4.86 FIP, while PECOTA portends a 4.27 ERA for the right-hander. Reminder, the AL average ERA and FIP in ‘14 was 3.92 and 3.85, respectively.
So, while one understands the “veteran presence” and ”this guy is a good fourth starter” clichés spewed since Lewis re-signed last December, one must also be weary of giving too many innings to a replacement level starter. Yes, eating innings is valuable, but if the innings being pitched produce below average results, that is the opposite of valuable. There is also Lewis being the third most likely pitcher to suffer an injury according to Jeff Zimmerman’s injury predictor model. While one year deals are rarely considered short-sighted or poor investments, and the odds of Lewis generating surplus value are not exactly astronomical, relying on Lewis to be one of the better pitchers in the rotation in 2015 might not be wise.
The thought that Lewis is ready to return to prior levels of performance based on his second half are rather silly because other than some mechanical improvements, he really did not improve much statistically in the second half of ’14, and much of that improvement could be attributed to regression. And if one would be satisfied with a 4.00 ERA in ’15 from Lewis because he is your fourth starter, and you think the fourth starter has to be a Lewis type or something similar, remember a 4.00 ERA is not really that good anymore in this pitching dominated era.
The author’s intention was not to poopoo on Colby Lewis (although it may seem that way) as one is quite fond of him, and appreciative of the wonderful baseball memories he has provided, but this is 2015, and relying on a 35 year old Lewis might not be a shrewd strategy for a team that fancies itself as a contender. While Lewis is currently listed as the Rangers fourth starter, he might not be the fourth best starter currently in the organization. Yes, Lewis did make some changes after the break last season, but regression also helped lower his ERA, and he was not pitching as poorly as the ERA for a guy who was not even throwing between starts suggested it did.