The original table I posted was constructed by grabbing BABIP for both Arroyo and mlb from the fangraphs website. Then I originally calculated BABIP for "Staff" by summing the fangraphs data for the staff after removing Arroyo's data then using the pitcher's formula for BABIP {(H-HR)/(TBF-HR-K-BB)}. Apparently this gives a slightly different value than the Fangraphs website BABIP (ex. fangraphs reports Arroyo's 2006 BABIP as .270 and the formula above calculates it at .268). So to avoid this slight discrepancy, below is a table reporting the BABIPs for each category as uniformly calculated by the pitcher's formula for BABIP after grabbing relevant data from fangraphs (so the pitchers BABIP formula was applied to Arroyo's counting stats, counting stats for all mlb pitchers, and to the counting stats for all Reds pitchers after subtracting Arroyo's stats from the group). Though there are slight changes (with an emphasis on
slight), as can be seen, the conclusions in my original post remain unchanged:
Code:
BA Staff mlb
2006 0.268 0.302 0.294
2007 0.303 0.306 0.296
2008 0.311 0.306 0.293
2009 0.261 0.282 0.292
2010 0.237 0.294 0.290
2011 0.276 0.278 0.287
2012 0.284 0.281 0.290
2013 0.264 0.270 0.290
BA Staff mlb
06-'08 0.294 0.305 0.294
09-'13 0.262 0.281 0.290
Again, the conclusions remain unchanged-Arroyo's BABIP was nearly identical to his teammates 5 out of 8 years that he was a Red (actually his was greater than his teammates twice based upon strict values but I don't consider those differences meaningful). The change in defense behind them lowered both Arroyo's and his teammates' BABIP similarly and the actual difference between the magnitudes can readily be attributed to the fact that Arroyo's penchant for giving up alot of contact likely amplified the quality of the defense behind him. Basically Arroyo had a two year stretch (2009-2010) where his BABIP deviated significantly from his teammates and that was sandwiched between five seasons of nearly identical BABIPs between Arroyo and his teammates. Again, that's a classic regression to the mean pattern. As can be seen by breaking up his tenure as a Red using the 2008 off season, Arroyo's BABIP was different from league average really only when he was in front of the reds stellar defense.
Those wanting to conclude that Arroyo has the ability to suppress BABIP, have to explain why he failed to have a BABIP that was meaningfully different than his teammates for the majority of his career as a Red. In other words, why would it be reasonable to conclude that being wrong about 65% of the time equates to being right? To argue that Arroyo can meaningfully suppress BABIP one has to essentially argue that Arroyo was able to greatly manifest his ability to extreme levels during three seasons but decided not to manifest it for the other five while the argument is implying his BABIP was not subject to randomness. Is that a more compelling position than viewing his few extremes as a pretty common thing that occurs with pitcher BABIP especially given the context that, 1) BABIP can vary greatly for pitchers from season to season, 2) he did not display a consistent ability to suppress BABIP before joining the Reds, 2) his BABIP was not different from mlb ave BABIP given long stretches of performance until his defense improved dramatically, 4) he is a high contact pitcher, and there are obvious effects of the quality of defense behind him on his BABIP, 5) as a Red he actually demonstrated poorer results on defense-independent contact as suggested by his higher than expected HR/FB%.
So back to the original, original point. Giving the pitcher full credit for what happened while he was on the mound, like using ERA to assess a pitcher's production does, is inappropriate. Some have argued that assessing pitchers via their peripherals overvalues pitchers who strikeout a lot, walk few, and don't give up gopher balls. Really, what those types of pitchers do is insulate themselves from their defense to a greater degree than pitch to contact guys (anyone want to argue with the notion that such a skillset isn't more valuable?). The problem with ERA is that it can greatly skew the evaluation of high contact pitchers (both up or down; and really ERA can do this to all pitchers) by ignoring the influence of the quality of defense behind a pitcher (among a list of other things ignored as well). Unfortunately the argument that "high contact/pitch to contact guys like Arroyo or Leake can suppress BABIP therefore their ERA is more valid than a multifaceted approach using peripherals, batted ball tendencies, luck.randomness metrics/environment/quality of defense", simply doesn't hold water as the analysis above definitively demonstrates. Arroyo really doesn't suppress BABIP so much. And to the minimal degree that he might, it doesn't explain the difference between his ERA and FIP as a Red in light of what we know about the quality of defense behind him.
Arroyo's lower than expected BABIP is largely a mirage. Suggesting he repeatedly suppresses BABIP to a degree that can significantly influence his ERA attributes a skill to Arroyo that largely really reflects an attribute of his defense or the interaction of the contact he allows with the quality of the defense. It also ignores that for the majority of seasons where diffrences existed between his ER and FIP, there were no differences between his BABIP and the BABIP of his teammates. If the Reds defense would've remained similar to the quality of the 2006-2008 seasons, would we even be having this discussion?
Frankly, the relevant question I think is why has Arroyo's ERA been lower than his FIP as a Red in the absence of a consistently suppressed BABIP?
One thing that jumps out of the data is the unusually high strand rate (LOB%) he's posted as a Red which is interesting especially given the unexpectedly poor strand rate he had for his career prior to 2006 (LOB%=66.1% versus mlb average LOB%= 71.2% for the period between 2000-2005) and the fact that pitchers that yield a high rte of contact generally can't control their LOB% (high K pitchers have been shown to generally post slightly higher than expected strand rates). Here's a summary of Arroyo vs mlb ave again:
Arroyo's Pitcher trifecta (peripherals intrinsic to a repeatable skills set) and luck/randomness metrics (things that tend to regress) for 2006-2013:
Code:
Make 'em miss Command BatBall Luck
ERA FIP K/9 K% Contact% SwStr% BB/9 BB% GB% BABIP LOB% HR/FB
Arroyo 4.05 4.6 5.92 15.7 83.8 7.1 2.31 6.1 40.9 0.278 75.1 12.0
MLBave 4.32 3.95 6.68 17.3 81.2 8.2 3.01 7.8 44.1 0.300 71.4 10.6