clock menu more-arrow no yes

Filed under:

Sabermetrics: Keuchel, Webb, and SIERA

New, 20 comments

Talking Astros' Sabermetrics: Comparing Dallas Keuchel and Brandon Webb and what it says about SIERA

Gary A. Vasquez-USA TODAY Sports

Carson Cistulli recently wrote a retrospective Fangraphs article entitled "The Brandon Webb That Wasn't and Will Be." For those who don't remember: Webb, the Diamondbacks ace pitcher, was among the best starting pitchers in the first decade of this century.  Webb won a Cy Young award in 2006, and was runner up for the Cy Young in 2007 and 2008.  His pitching style featured what Cistulli calls a "generational" groundball rate.

Cistulli's reminiscence brought on, in his words, an epiphany.  Webb was not considered to be a top prospect before his break out 4 WAR rookie season in the majors. He was the 249th overall draft pick in 2000, never made the Top 100 prospect lists, and was never discussed as a pitcher who would average 5 WAR in the majors.  Webb was a low velocity pitcher; in fact, his career pitching velocity would have placed him among the bottom 10 pitchers for velocity in 2014.  Cistulli ties this sub-par velocity to the consistent under ranking of Webb's ability before he reached the majors.

As I read the fangraphs article, the description of Webb's pitching immediately brought to mind Astros' pitcher Dallas Keuchel. Keuchel was the 221st overall draft pick in 2009, and he also did not warrant inclusion in Top 100 prospect lists.

Will Keuchel's 2014 season be a break out similar to Webb's 2003 rookie season?  I can't tell you what will happen in the future.  But the pitching metrics for Keuchel in 2014 can be compared to Webb's pitching.

The table below shows Keuchel's 2014 peripheral stats compared to Webb's break out 2003 season and his Cy Young 2006 season.  Keuchel's line is displayed between the two Webb seasons.

IP K/9 BB/9 HR/9 BABIP FB velo GB%
Webb, 2003 180.2 8.57 3.39 0.6 0.264 88.7 63.90%
Keuchel 2014 200 6.57 2.16 0.5 0.295 89.7 63.50%
Webb, 2006 235 6.82 1.91 0.57 0.287 88.3 66.30%

The next table provides a similar comparison of ERA and advanced pitching stats (FIP, x-FIP, and SIERA).

SIERA ERA FIP x-FIP WAR
Webb, 2003 3.15 2.84 3.34 3.1 4.4
Keuchel 2014 3.11 2.93 3.21 3.2 3.9
Webb, 2006 2.99 3.1 3.18 3.03 6.3

Keuchel's 2014 pitching bears a remarkable similarity to Webb's pitching in 2003 and 2006. And, like Webb, Keuchel exhibited the major's top groundball rate.  Keuchel's velocity is somewhat higher than Webb's, but his groundball rate doesn't reach the extremes that Webb produced.  Although Keuchel's ERA and FIP are comparable to Webb's, he didn't quite reach Webb's 2003 fWAR, and Webb's 2006 fWAR is much higher.  This is partly due to the ballpark and league average adjustments which are applied to FIP for the fWAR calculation. The run environment has declined since 2006 and Webb pitched in a more hitter-friendly home ballpark. On the plus side for Keuchel, he produced a higher BABIP, indicating less liklihood that the results were driven by luck.

Astros' fans would love to see Keuchel follow Webb's path in most respects...except one.  Webb's ML career was abruptly ended in 2009 by a shoulder injury. Though he attempted to come back from the injury, subsequent to 2009 Webb only pitched 12 innings at the AA level in 2011.

SIERA

The comparison of Keuchel and Webb led me to meander down another trail: SIERA. SIERA is a complex formula which goes beyond FIP and considers the interaction of various pitcher skills, including groundball and flyball rates.  SIERA is described in the Fangraphs glossary here. And go here for a more detailed discussion of underrated pitching skills which SIERA attempts to recognize.

In the comparison of Keuchel's 2014 and Webb's 2003 and 2006, above, both the ERA and SIERA were less than fielding independent pitching (FIP).  SIERA attempts to reflect a positive characteristic of extreme groundball pitchers. The Fangraphs library discusses how SIERA is affected by extreme groundball pitchers:

The more ground balls a pitcher allows, the easier they are to field.

Example: Brandon Webb

The infield defense behind Brandon Webb has been very average in his career, but his BABIP on ground balls is not the league-average .233. During his career, it’s been .205. Webb’s sinker is heavy, and it has a tendency to chop off the bottom of the bat. Nearly every pitcher with Webb-like ground-ball skills allows fewer grounders to go for hits...

In other words, a pitcher who increases in ground-ball rate from 45% to 50% will not help his SIERA as much as a pitcher who increases his ground-ball rate from 55% to 60%, because even though both are giving up fewer home runs by increasing their ground-ball rates, the latter pitcher is getting more outs on those extra ground balls.

This is probably the most important non-linear term included in SIERA.

The groundball BABIP tendency described above may partially explain why extreme groundball pitchers in 2014 had an ERA below their FIP.  (11 of 13 pitchers with a GB% above 53% had an ERA below their FIP.)

So, when should we use SIERA instead of FIP or x-FIP as the "go-to" pitching metric?  The usual argument for SIERA is that it is more predictive of next year's ERA than FIP or x-FIP.  But can SIERA help explain why the current year ERA is more or less than FIP?

The answer to the latter question might be "yes," but my review of 2014 pitchers' results leaves some uncertainty in my mind about the answer.

Although SIERA credits extreme groundball pitchers for a lower BABIP on groundballs, almost half of the extreme groundball pitchers in 2014 produced a SIERA higher than their FIP.  SIERA attempts to reflect the interaction of a number of skills in addition to groundball rate.  Because of the formula's complexity, the reason for particular results is not transparent.

Taking a pitching example familiar to Astros' fans, Jarred Cosart is an extreme groundball pitcher who has produced SIERA results higher than his FIP and x-FIP, which in turn are higher than his ERA.  Why?  Hard to say. (At least without trying to work through the complex calculation of Cosart's SIERA.)  Maybe it has something to do with the way SIERA treats high BB% pitchers.

For what it's worth,  the following starting pitchers with an ERA below FIP also produced a SIERA substantially below FIP: Bumgarner, Samardzija, Keuchel, Cueto, Leake, Hammel, Gallardo, Haren, Peralta, Norris, Dickey, Simon, de la Rosa, Stults, Danks, Noesi, and Hernandez.  Many on this list are above average groundball rate pitchers, though several are not. Generally speaking, I think most of these pitchers fit the "weak contact" label--though that is admittedly subjective.  Again, I can't say whether SIERA successfully identified skills in these pitchers which allowed them to produce ERAs below their FIP...or whether it's just randomness.

It's also worth noting that x-FIP produces similar results to SIERA for many of the listed pitchers.  This isn't surprising because SIERA and x-FIP are numerically similar for most pitchers.  Among 2014 starting pitchers, the correlation between SIERA and x-FIP is 0.98 (i.e., highly correlated).  For most pitchers, FIP and x-FIP together form adequate evaluation metrics, and SIERA adds minimal value.  However, in particular instances, SIERA may provide additional information which helps us understand if the pitcher's ERA is sustainable.  Also, as noted previously, SIERA may be the best choice for predicting next year's performance.