When Eric Seidman and I introduced SIERA in February, we were very careful to show that it predicts future ERA better than current ERA does. While Defense Independent Pitching Statistics are not a foolproof way to measure pitchers, using them as a guide to dig further into the numbers can be very helpful. Last October, I spent a couple articles analyzing Cole Hamels’ performance, and I highlighted how little was different between his 2008 and 2009 season, and how I expected his performance to improve as his luck neutralized. Sure enough, Hamels has seen his ERA fall back toward 2008 levels in 2010. In June, I disappointed Rockies fans by explaining the luck that had led to Ubaldo Jimenez’s 1.16 ERA at that time. Sure enough, he has a 4.36 ERA since that article was posted. Eric and I wrote on the Diamondbacks’ starters, stressing the bad luck that Dan Haren had seen to that point in the season. He had a 5.35 ERA, but it has been 3.59 since that article was posed and Haren has also been traded to the Angels. My point is not to cherry pick successes, but to prove that this type of analysis works. I certainly cannot be right every time I say a pitcher’s ERA is likely to fall or rise, because luck plays a role in pitching to a very large degree and luck by its very nature can reoccur. However, this type of analysis will prove prophetic more often than not.
In today’s article, I will take seven pitchers with significant divides between their ERAs and their SIERAs, and figure out if they are likely to regress and why they might. Luck has many components and simply playing what I call “saber pepper” by throwing a BABIP against the BABIP fungo bat is not going to cut it. Instead, I will find the source of the luck and why it is occurring.
Josh Beckett: 5.91 ERA, 3.69 SIERA
The largest gap between ERA and SIERA belongs to Beckett, who has struggled with back problems this year. Beckett is a classic example of why one must always be careful in declaring a pitcher’s bad luck to be the cause of such a differential. The main source of his differential is his .358 BABIP, including a .385 mark with men on base. However, this is a result of batters hitting the ball much harder off of him, which is likely related to his injury. While bad BABIP is rarely repeated with any sort of consistency, that does not mean it is not the residue of temporarily poor mechanics. A back injury can lead to poor mechanics, and that can lead to high BABIPs. Will Beckett’s 2011 ERA be closer to his 5.91 ERA than his 3.69 SIERA? Almost undoubtedly the ERA! But that is a different thing than saying that Beckett is simply the victim of bad luck. His .815 line-drive BABIP, .250 BABIP on fly balls, and 15.9 percent home runs per outfield fly ball all indicate that his pitches are being transformed into rockets by opponents’ bats. Look for Beckett to improve in 2011, but don’t call 2010 bad luck. It was temporary and unsustainable bad performance.
Clay Buchholz: 2.25 ERA, 4.28 SIERA
On the other side of drastic SIERA/ERA differences among Red Sox pitchers is Clay Buchholz. The two-run difference between his ERA and his SIERA is the largest of all major-league pitchers with 100 innings pitched, and is largely a result of his .262 BABIP, which has led to 17 fewer hits than average. This is not the only reason, however, as his .229 BABIP with men on has further minimized damage. The primary explanation for his low BABIP is the rate at which ground balls reach the outfield. His ground-ball BABIP is .175, which is far below the league average of .234. Since 50 percent of his batted balls are ground balls, this is the majority of the difference between Buchholz’s BABIP and the league average. Across the league, 15.2 percent of ground balls find the hole, but only 9.6 percent of Buchholz’s ground balls have done so. While some of this may be a skill for inducing worm-beaters, the evidence of this is far from conclusive at this stage, and most pitchers with this kind of result are simply lucky. While Buchholz may yet prove to be a pitcher who can pitch ahead of his peripherals, the odds of him being able to stay two runs in front of them are incredibly unlikely. At this stage, one would have to expect Buchholz to be relatively average unless he shows an improvement in strikeouts or walks. Odds are that Beckett and Buchholz will have far closer ERAs in 2011, with Beckett even being more likely to outperform Buchholz.
Tim Hudson: 2.41 ERA, 3.60 SIERA
Hudson has been the subject of many forecasters’ predictions of regression in 2010, but his ERA has not risen very much. He is a Cy Young contender, despite a strikeout rate that has fallen from a career-high 16.2 percent last season to 14.9 percent in 2010. However, Hudson has the best ground-ball rate of his career at 65.8 percent. Although his ERA is still below his SIERA by a large margin, his SIERA is lower than his xFIP of 3.80. The reason for that is SIERA has a negative coefficient on the square of ground-ball rate, which in English means that the more ground balls you induce, the more those ground balls prevent runs. The reason for this is something that I would like to research further, but Hudson provides a nice case study. One possible reason is that ground balls are frequently singles when they are hits, but the more ground balls induces then the more than can turn into single-erasing double plays. Hudson leads the majors in ground-ball double plays induced with 29. However, Hudson’s ERA is really so low because his BABIP is just .249. Breaking things down further, we see that his ground-ball BABIP is a preposterously low .187. The league average is .234, meaning that Hudson has allowed 18 fewer ground-ball hits than league average. This would seem like good luck, just as it appeared to be with Buchholz. However, this is not all that different than his .208 career ground-ball BABIP and Hudson's track record calls into question typical assumptions about these things. This year has brought down his overall career BABIP to .284, despite playing on teams that have collectively allowed .295 BABIPs. With 6,836 career balls in play, that is a statistically significantly below-average BABIP. Hudson is likely the kind of pitcher who is often going to pitch ahead of his peripherals, which means that even though he is not going to perennially put up 2.41 ERAs, he should safely be in the mid-3.00s from here out and is a legitimate mid-level ace that could carry the Braves in the postseason. He has been lucky but his luck is rather muted when you consider his ability to erase baserunners and induce weak ground balls.
Joe Blanton: 5.15 ERA, 4.16 SIERA
Many people missed Blanton’s improvement in 2009 that Eric documented well coming into this season, and fewer people were bound to notice it when Blanton masked it by struggling mightily to start 2010. However, despite failing to miss bats early on, Blanton has rebounded and begun to strike out more hitters. However, he has been the victim of a .329 BABIP and a 14.9 percent home run per outfield fly ball rate. The latter is not that abnormal considering the league average is 12.2 percent and he plays in a reasonably small park that could explain half the difference, but the BABIP is really the problem for Blanton. There are three main sources of his high BABIP: 1.9 percent more line drives than the league average (20.9 percent vs. 19.0 percent), 43 points higher line-drive BABIP than league average (.760 vs. .717), and 16 points higher outfield fly-ball BABIP than league average (.195 vs. .179). Combined with the fact that his home run per fly-ball rate is a little high, it’s pretty clear Blanton is getting hit harder than he should. The rate of batted balls reaching the outfield in the air is higher this year (49.0 percent) than last year (47.4 percent) as well, providing further evidence of this claim. However, things like line-drive rate and home runs per outfield fly ball do not tend to last and Blanton’s track record provides no reason to expect them to persist. While the decision of whether Blanton is “at fault” or not is more of a philosophical one than a statistical one (this is not bad defense, but hard-hit balls), the conclusion is that Blanton is more likely to pitch like his 4.16 SIERA suggests more than he is to pitch like his 5.15 ERA. Pitchers who strike out hitters at least as reliably as Blanton does generally do not allow .329 BABIPs.
Johan Santana: 2.98 ERA, 4.18 SIERA
If I told you that Santana would have a 2.98 ERA in 2010 before the season started, you probably would not have been surprised. However, the way that Santana has reached that mark is quite different than you might have expected. Since joining the Mets, Santana’s strikeout rate has fallen considerably. He struck out 26.8 percent of hitters in his last year with the Twins, but immediately plummeted to 21.4 percent and 20.8 percent in 2008 and 2009. Now, he is just below the league average of 18.3 percent this year, checking in with a mere 17.6 percent. This is why his SIERA is so high. However, his BABIP is just .276, which is similar to his .278 career rate. Is he inducing the same weak contact that he historically has induced? Not really. He has seen far fewer hits reach the outfield (52.3 percent in 2010 compared to 48.3 percent in 2009). His infield pop-up rate explains this difference, falling from 14.0 percent last year to 10.0 percent this year. However, Santana has kept his BABIP especially low with runners on base at .237. This is well below his .304 BABIP with bases empty. Thus, he has allowed fewer hits when it has mattered and kept his ERA low. This trend is not new to Santana, who has typically kept his home-run rate lower with runners on. His BABIP has been similar in his career with men on versus with the bases empty, but the fact that Santana is still showing some trends of buckling down is a good indicator that there is some skill involved in preventing runs in important situations. Santana also has kept BABIP on fly balls down to just .128, well below the league average of .179. Why Santana’s BABIP is so low is less clear. Traditionally, BABIPs of high-strikeout pitchers are low, and so Santana’s career .278 BABIP is not all that surprising. However, he has kept it low despite losing the high-K rate. It’s way too early to predict Santana is due for an ERA north of 4.00 next year, but south of 3.00 seems particularly unlikely without a revival of his strikeout rate.
James Shields: 4.92 ERA, 3.50 SIERA
Shields’ improvement in strikeout rate from 18 to 21 percent since last year should have marked an improvement in his ERA, but instead it has spiked from 4.14 to 4.92. The reason is that opponents are hitting the ball much harder. Shields has allowed 30 home runs on 172 fly balls, which is about nine more than you would expect for a pitcher who pitches in Tropicana Field for half of his games. Shields’ .332 BABIP explains a big part of the difference. His line-drive rate is about 2 percent above league average, which explains about 10 points of increased BABIP while the rest is evenly distributed. He has allowed about six hits on line drives above league average. He has also allowed two extra hits on grounders, two extra hits on bunts, and two extra hits on fly balls. With an improvement in strikeout rate accompanying this, it seems very unlikely that the BABIP problem is likely to persist and seems to be one of the more obvious cases of bad luck among pitchers with huge gulfs between their SIERAs and ERAs. Look for Shields to emerge quickly and possibly surprise an opponent or two in the postseason.
Trevor Cahill: 2.72 ERA, 4.28 SIERA
Cahill has induced ground balls on 57.3 percent of batted balls in 2010, well above the league average of 45.6 percent. More importantly, he has let very few balls find holes. His ground-ball BABIP is .137, which is very low even for an Athletics team that has a fantastic .196 ground ball BABIP. Keep in mind that Cahill has allowed a very normal 16 hits in the infield, Given Cahill’s low 3.6 percent pop-up rate, odds are he is not inducing incredibly weak contact. His 15 home runs on 121 outfield fly balls are pretty much spot on the league average, further highlighting his normalcy in allowing hard-hit balls. While Cahill appears to be a good match for the ground-ball guzzling A’s, he still have been lucky and should expect a more average-looking ERA going forward.