Projection Systems Don’t Like Chris Withrow, Should We Be Worried?

Chris Withrow had a surprisingly good season in 2013. After being stuck in AA for over three years, he managed to make an impact at the major league level, which was far beyond what we expected. Withrow’s season came out of nowhere, but it was not without its warning signs. As a result, all four major projection systems that have been released see Withrow taking a major step back next season:

	IP	ERA	FIP	K/9	BB/9	HR/9	WAR
2013	34.2	2.60	3.57	11.16	3.38	1.30	0.1 (F), 0.7 (RA-9)
2014 Steamer	30.0	3.34	3.55	10.44	4.69	0.81	0.0 (F)
2014 ZiPS	54.1	3.48	3.77	9.20	4.35	0.85	0.3 (Z)
2014 Oliver	74.0	3.89	4.12	8.76	5.11	0.85	-0.3 (F)
2014 PECOTA	45.2	4.43	4.27*	8.28	4.53	0.99	-0.2 (BP)

*Using 2013 FIP constant

PECOTA and Oliver both have Withrow below replacement level next season. Steamer projects a replacement level season. ZiPS is the only projection system that has him above replacement next season, and not by much. All of the systems are forecasting a big increase in ERA, and most are seeing a jump in FIP (Steamer is the only system that doesn’t). By digging deeper into the reasoning of the projections, it will be possible to have an informed opinion on their validity.

BABIP

Withrow had a BABIP of .205 at the major league level last season. A BABIP that low is not sustainable over a large sample size. The league average BABIP is a bit below .300. Even pitchers who are good at controlling hits on balls in play (like Clayton Kershaw) can’t sustain multiple seasons below .260 or so. Withrow’s BABIP will regress and will cause him to allow more hits, leading to more runs allowed.

Strand Rate

Last season, Withrow had a strand rate of 88.5%. That was 13th highest among relievers with at least 30 innings pitched. While strand rate can be a skill controlled by the pitcher (better pitchers have better strand rates), this rate is still pretty high. Strand rate has a moderate negative correlation with BABIP (as BABIP goes up, strand rate tends to go down). When Withrow’s BABIP regresses, his strand rate will go down, again resulting in more runs allowed.

ERA/FIP Gap

Withrow had a large gap between his ERA and FIP last season. There isn’t much data on Withrow at the major league level, and his BABIP and strand rate are going to regress. As a result, the projection systems are expecting his ERA to go towards the FIP and not the other way around.

Track Record

Withrow only has one full season above AA and half a season in the majors, so the projection systems are expecting him to regress towards his AA numbers. This regression is applied to both Withrow’s run prevention and peripherals.

Should we be worried?

Yes, but not very. While I think all of the issues listed above are valid, I’m not too concerned about Withrow going forward.

Some of the 2013 stats that I left out of the table are more optimistic. Withrow’s FIP was 3.57, but that included a HR/FB ratio of 14.7%, well above league average. Withrow’s xFIP, which adjusts his HR/FB ratio to league average, was an excellent 3.03. The HR/FB ratio decrease is also supported by Withrow’s time in Albuquerque, where he allowed zero home runs in a highly inflated run environment.

There is also data that partially supports Withrow’s control of balls in play. While he won’t allow a BABIP of .205 again next season, his SIERA was only 2.59. Withrow’s SIERA was great for a few reasons. Withrow’s strikeout rate was very high last season. He also allowed more fly balls and fewer line drives than average, which supports a low BABIP. As I found when I reviewed Withrow’s 2013 performance for a guest post on MSTI, Withrow only allowed one line drive on his slider all season, and that was during his first major league appearance. Withrow threw more sliders as the season progressed, which could point to continued success in limiting line drives next year.

Regressing Withrow’s peripherals towards his minor league numbers makes sense, but a lot of those numbers are from when he was still starting. His stuff plays up in the bullpen, and it might have taken him time to adjust. The fact that Withrow’s strikeout rate was similar in AAA and the majors gives me hope that he can maintain it next season. Withrow’s walk rate was lower in the majors than in AAA, but since Withrow’s pitch distribution changed significantly after he came up that could be a real change and not just a small sample size. We need more data to be confident in that claim.

Withrow will need to reduce his walk rate in the future if he wants to have sustained major league success. However, the potential warning signs aren’t as pronounced as the projection systems make them seem. There’s still plenty of hope for him moving forward. When the seven man bullpen is constructed for the regular season, there’s a decent chance that Withrow won’t be in it. But I’d still rather have him than Chris Perez or Brandon League.

—

This post uses the following statistics:

• FIP: Fielding Independent Pitching. Attempts to create an ERA-like number using only plays that do not require defense to complete (K, BB, HR) by assuming the pitcher is playing in front of a league average defense. Explanation here.
• xFIP: Expected FIP, which is the same as FIP but normalized with a league average HR/FB ratio. Explanation here.
• SIERA: Skill interactive ERA, an ERA estimator that uses batted ball data and other inputs. Explanation here.