By the Numbers: Evaluating pitching as easy as ERC
Many still consider ERA to be the best single measure of a pitcher's ability and the strongest correlate of his Fantasy production. Though you may have the occasional Jair Jurrjens or Randy Wells creep onto the ERA leaderboard without great strikeout or WHIP ratios, by and large, the pitchers who are the best at preventing runs are also among the leaders in these other categories. They are also some of the most reliable sources of wins.
Still, ERA can sometimes leave something to be desired. If a pitcher gets the benefit of a good defense, his bullpen regularly cleans up his messes or he just has a streak of good luck, his ERA will overrate him. Should those fortunate circumstances change, that pitcher's Fantasy value will take a turn for the worse, and ERA would let you down as an indicator of future performance. ERA can underrate a pitcher as well. Think about the perception of Jered Weaver just over a year ago, when he was taken after the 10th round in most leagues. He was coming off a 2008 season in which he had become a much better strikeout pitcher while still not allowing too many walks or homers, yet his ERA did not reflect the improvement. It actually rose from 3.91 to 4.33, so many owners couldn't see the 3.75 ERA and 16 wins that were to come.
While Weaver was missing more bats and putting fewer men on base, he was allowing more runners who reached base to score. His left-on-base (LOB) rate of 68.8 percent was unusually low for him and below the major league starting pitcher average of 72.4 percent for the '08 season. A measure of Weaver's performance that weeds out the effects of defense, bullpens and luck would have shown the potential for his improvement in Fantasy. Component ERA, or ERC, is just that sort of measure, and Weaver's ERC of 3.80 in 2008 foreshadowed the improvement that was to come in 2009, once his LOB rate got back its norm.
While BABIP, the measure used in our weekly Lucky/Unlucky analyses, can also provide clues as to which pitchers are due for improvement or decline, its main impact is on the batting average that a pitcher allows. ERC takes balls in play into account, as well as homers, walks, hit batsmen and total bases. It can't always predict changes in future performance, since a pitcher's skill trends (e.g., strikeout, walk, and batted ball type rates) can affect their Fantasy stats at least as much as other factors. However, if a pitcher's skill and BABIP numbers are roughly the same from year to year, ERC can help us to anticipate otherwise unexpected changes in performance.
By looking at the ERC differential -- that is, the difference between ERC and ERA -- from 2009, we can target pitchers who are good bets to improve in 2010. Below are the 15 starting pitchers whose ERCs were the lowest in comparison to their ERAs. The solid bars represent each pitcher's 2009 ERA, while the hash marks represent where ERC sees that their ERA should have been. Also note that all but two of them had LOB rates at or below the major league average for starting pitchers. It is no coincidence that Ricky Nolasco had the lowest LOB percentage in the majors and also posted the lowest ERC differential. Even if he had a porous defense and rickety bullpen "supporting" him (which he arguably does), Nolasco would be hard pressed to allow so many baserunners to score again. Nolasco's ERA has dropped so far this year, and we have already seen improvements from Jonathan Sanchez , Ubaldo Jimenez and Carl Pavano (though some of that is skill-related), while skill and BABIP decline -- at least the short-term variety -- has kept pitchers like Javier Vazquez and Joel Pineiro from enjoying the success that ERC predicted for them. Of these starters, Dan Haren may be the least likely to improve in 2010, as a low .272 BABIP and above-average 74 percent LOB rate cancelled out whatever rough breaks he endured in 2009.
With the exception of Joe Saunders , each of the 15 starting pitchers with the highest ERC differentials in '09 stranded an above-average proportion of their baserunners. For the most part, we can expect each of these pitchers to do worse this year than last, though a couple of exceptions stand out. A casual glance at the list would suggest that Ricky Romero is due for a substantial fall this year, but last year's high BABIP and near-average LOB rate show that he wasn't as lucky as ERC suggests. ERC also says Roy Halladay wasn't as good as he seemed in his final Toronto season, but an elevated BABIP reassures us that he really was that good. J.A. Happ , on the other hand, had a double-whammy of favorable BABIP and LOB rates. Putting his forearm injury aside, these statistical trends offer sufficient reasons to be lukewarm on Happ this season.
Even more relevant for owners, now that we are entering the second month of the season, is to determine whose 2010 stats we should distrust. Jair Jurrjens , Freddy Garcia and Chris Volstad have some the lowest ERC differentials so far, but all have benefitted from ultra-low BABIP rates. As these BABIPs rise over the next few weeks, not only could their ERAs grow further, but their ERCs will catch up to their ERAs. Bronson Arroyo , Kyle Lohse , Jeremy Bonderman , Justin Verlander and Kenshin Kawakami , however, have all genuinely pitched well enough to have ERAs below 4.20, even though all currently have ERAs of at least 4.50.
ERC has a much harsher assessment of Rich Harden , Jon Garland , James Shields , Randy Wolf and Dontrelle Willis . Each of these starters, aside from Harden, has an ERA below 4.00, though ERC suggests that all of them should have ERAs that are at least a run higher than their current level. Several of the pitchers with large positive ERC differentials have gotten raw deals on balls in play, though discounting the impact of a high BABIP just puts makeup on the proverbial pig for Justin Masterson , Nick Blackburn and John Lannan . Jonathon Niese and Brett Myers , however, may really be as good as their ERAs look, as they have managed low ERAs despite BABIP rates in excess of .350.
The ERA/ERC bullet graphs you see here will be a recurring feature this season, as part of a rotating weekly schedule of sabermetric leaderboards. Every few weeks throughout the season, we will check in and see which starting pitchers have ERCs that portend future changes in their ERAs, while also keeping an eye on BABIP and skill trends that will add depth to our analysis.
Runs Created per 27 Outs (RC/27) -- An estimate of how many
runs a lineup would produce per 27 outs if a particular player
occupied each spot in the order; ex. the RC/27 for Miguel Cabrera
would predict the productivity of a lineup where Cabrera (or his
statistical equal) batted in all nine spots; created by Bill James
Component ERA (ERC) -- An estimate of a what a pitcher's ERA would be if it were based solely on actual pitching performance; created by Bill James
GO/AO -- Ground out-fly out ratio
GB/FB -- Ground ball-fly ball ratio
Batting Average per Balls in Play (BABIP) -- The percentage of balls in play (at bats minus strikeouts and home runs) that are base hits; research by Voros McCracken and others has established that this rate is largely random and has a norm of approximately 30%
Isolated Power -- The difference between slugging percentage and batting average; created by Branch Rickey and Allan Roth
Walk Rate -- Walks / (at bats + walks)
Whiff Rate -- Strikeouts / at bats
Al Melchior will be providing data-centric advice columns Fantasy owners all season. Click here to send him a question. Please put "Melchior" in the subject field.