November 16, 2007
Quick Answers from Quick ERA
Knowing which pitchers you can rely on from season to season is important in winning any fantasy league, keeper or otherwise. Pitcher performance, at least in certain fantasy statistics such as ERA, are variable and oftentimes do not reflect a pitcher's peripheral work. For this week's piece, we are going to take a look at pitchers whose ERAs are not reflective of their true performance, in both a positive and negative sense.
One of the most effective in-season statistics for analyzing this scenario is Nate Silver's Quick ERA, or QERA for short. QERA estimates what a pitcher's ERA should have been based on his strikeout, walk, and groundball rates. Since these particular items are not very park-dependent, QERA can be used comparatively without major adjustments. QERA is not perfect, of course, as those with severe groundball rates can be penalized heavily if the defense behind them does not do their job; this is easy enough to identify though, and there's more than enough good that can come from using this nifty stat for fantasy purposes.
Jeremy Bonderman has been one of the better pitchers in the American League the past few seasons, but it's been tough to notice this with the poor luck he has run into. In 2006, Bonderman posted a 4.08 ERA despite some excellent peripherals and a groundball rate approaching 50 percent, thanks to a low strand rate of 69 percent and a BABIP of .328. In 2007, the problems were not all luck-related, as Bonderman's homer rate jumped from 0.8 to 1.2 per nine, but his BABIP of .329 was well ahead of where it should have been, given his low 17.6 percent line-drive rate. QERA says that Bonderman should have had an ERA closer to 3.82 rather than the 5.01 he posted, and if not for an even lower strand rate of 66 percent and the increase in homers, he most likely would have approached that.
As long as his sore elbow holds up in 2008, there's no reason not to pick up Bonderman early in a draft, but see if you can swing a low-cost deal on him at auction due to the injury risk and letdown 2007 campaign. The only issue I have with Bonderman comes from his poor strand rates, and oddly enough, he seems to pitch worse when there is a man on third base. This would mean that his bad strand rate is not a product of bad luck, but is instead the result of something that Bonderman needs to fix.
On the other side of the spectrum we have Noah Lowry, whose ERA of 3.92 belies his true performance. QERA pegs him for a substantial jump to 5.94; that is the largest deviation from ERA among starters with at least 125 innings pitched. All of the luck that Matt Cain and his quality season didn't get instead went to his teammate Lowry, who managed a 14-8 record for a last-place Giants' team despite striking out only 87 batters and walking another 87. Lowry is not a groundball pitcher, but he cut down on his flyballs significantly this year, helping him slash his homer rate almost in half. This helped alleviate some of the problems that come with walking five hitters per nine, but to expect that kind of performance again is asking for trouble on your pitching staff, in both real life and fantasy. As for his batted-ball data, Lowry should have had a BABIP close to .315 rather than the .288 he posted; this jump of 27 points would give him an adjusted opponent's line of .293/.391/.422, which is a scary sight to behold. Lowry might not even be worth a late look despite his solid win total and neat ERA from 2007, because his 2008 performance may only be fifth starter-worthy, even for an NL pitcher.
This year's third-place finisher for the American League Rookie of the Year, Brian Bannister may not have the same sort of success in his sophomore season. Bannister posted a 3.87 ERA in 165 innings of work despite striking out just 4.2 hitters per nine. His walk rate was not particularly low at 2.4, and he allowed 0.8 homers per nine. His success stemmed from his .266 BABIP, which helped Bannister keep opponents down to .249/.303/.409. If you adjust his BABIP for his liner rate of 19 percent, you get a huge jump to .312; adjusting the opponent line for that puts him at .295/.349/.455, which is the equivalent of turning every hitter into the 2007 version of Miguel Tejada or Jack Wilson. His QERA reflects these issues at 5.18, the fourth-largest bump among starters with at least 125 innings. Pitchers with low strikeout rates have defied expectations before-in fact, both Fausto Carmona and Chien-Mien Wang are on the same list as Bannister-but Bannister lacks the lofty groundball rates that make this sort of thing acceptable at the major league level.
People will be tempted to overpay for Bannister, especially since he managed double-digits in wins for a last-place team and had a shiny ERA. Steer clear of making the same mistake, as Bannister's peripherals and QERA tell a different story than his more mainstream numbers suggested in 2007.
Brad Penny presents an interesting issue as far as QERA is concerned, since the stat believes his 3.03 ERA was far off the mark from his true performance of 4.63. Looking at his peripherals, it's easy to see where the disconnect in true performance and ERA lies, as Penny only managed to strike out 5.8 hitters per nine while walking 3.2 for a pedestrian K/BB rate just under two to one, though his groundball rate of nearly 50 percent is obviously a plus. The peripheral that improved Penny's numbers the most was his very low home run rate, as he allowed just 0.4 per nine on the season. This number is a little deceiving though, as Penny did not allow a homer for the first two months of the season, and then had a HR/9 of 0.6 from June 3 onward. You would certainly take 0.6 HR/9 though, even if that entails fewer strikeouts.
Penny changed his approach somewhat by relying more on groundballs-his G/F moved from 1.2 in '06 to 1.6 in '07-and his reliance on his sinker helped generate quicker outs and faster plate appearances by the opposition. This compensated for the dip in punchouts and the increase in free passes, and the drop in his home run rates also helped out. Even if Penny were to return to his former homer rates, he could probably manage an ERA in the threes for a team that's going to win games for him. He won't get you strikeouts like in years past, but he'll get outs and innings.
In his first year for Milwaukee and in the National League, Dave Bush posted surprisingly poor numbers, with an ERA of 5.12, despite moving to a significantly easier division than the AL East. He struck out roughly half a batter less per nine innings at 6.5 while bumping his walk rate up the same amount to 2.1, and his homers also increased, from 1.1 per nine to 1.3. The homers were an issue that came from allowing more flyballs in fewer innings; his HR/FB rate was actually lower than in 2006. Despite this, his QERA of 4.15 is a serious improvement on his actual ERA, and given his past performance, may be the more realistic number as far as what you should expect from him going forward. Even more than the homers, his above-average .327 BABIP with a more common liner rate of just under 19 percent caused him trouble; this was Bush's first BABIP over .300 in his career. If everything else stays the same across the board, his hit rates should fall down towards his previous levels. You can expect to see his 2008 performance reflect his 2007 QERA more closely.