Bullpen the cause for sinking Marlins?


This according to Baseball Prospectus’ Russell Carleton (formerly known as Pizza Cutter). I’ll break down his take (BP Premium subscribers only, I believe, though I think I have fantasy access and still got to read everything) bit by bit, starting with this:

"The Marlins outplayed their Pythagorean projection by about 5 games, winning 87 rather than the 82 that would have been expected of them given their total runs scored and runs allowed. The Braves, on the other hand, underplayed their projection by 5 games. In general, a team significantly over-performs by winning a lot of close games, and loses a lot of blowouts."

I went ahead on checked on the Marlins and Atlanta Braves in terms of close games and blowouts. My assumption coming in was that Carleton was correct on this point. Sure enough, the Marlins had indeed performed a little too well in one-run games, going 30-20 last season in games that they probably should have split. In fact, splitting those games gives up those five games that Pythagorean says we should not have won! The Marlins split extra-inning games as expected, but also won two rain-shortened games, contests that may or may not have been appropriate to win.

How did the Braves fair in these same contests? The Braves mostly split one run games, though they were two games over on those games. They also won blowouts at a large rate, which helps to drop their actual win total compared to their Pythagorean rate. The Braves also won one extra extra-inning game and one rain-shortened blowout. Really, shouldn’t we be dropping their win totals too?

Obviously, that’s not a good way to analyze things, but it doesn’t really matter. We know the Marlins outperformed their Pythagorean as well as their context-neutral WAR performance. The question Carleton was getting at was whether the Marlins would perform to 2009 levels in 2010. And here, the bullpen was at the heart of the issue.

Because I do not know what constitutes fair use of BP’s Premium content, I won’t quote everything here. Here’s the crux of Carleton’s argument:

– the bullpen outperformed their expected performance last season based on their component statistics

– the bullpen is mostly unchanged and perhaps worse than last year’s model

– because the pen outperformed their peripherals, they should regress to the mean and become significantly worse this year

On the surface, I agree, though perhaps not so much with the severity of the argument. Let’s dig into some numbers. Here are two skill-intensive ERA estimators, FanGraphs’/THT’s xFIP and BP’s SIERA, looking at the pen’s 2009 performances. And here this year’s CHONE projected FIP, serving as our defense-independent performance metric.

Player2009 xFIP2009 SIERAProj. 2010 FIP
Leo Nunez4.414.024.26
Dan Meyer4.113.624.07
Brian Sanches
Renyel Pinto5.084.744.45
Burke Badenhop3.46*3.713.92

*Includes only relief appearances

Just by scanning the numbers there, it seems like a lot of the regression is going down, not up. Of course, all of these pitchers (with the possible exception of Badenhop) outperformed their FIP last year. But I would say that there is good reason for some of them.

Carleton brings up that Meyer, Nunez, Sanches, and Pinto all had below average BABIP. I would agree, and qualify that the regression projected by CHONE is appropriate; he’ll give up more hits and less home runs this year. PECOTA seems similarly inclined. Pinto, I would venture, has something in his repertoire that is inducing weaker contact, as his career BABIP in 936 batters faced is .270. CHONE projects a .290 BABIP, which I find fair. Clearly his projected BABIP would still need a good deal of regressing, but I think there is something to be said about Pinto’s contact numbers, especially his strangely low LD% for his career. Sanches had a .287 BABIP and allowed 157 non-bunt BIP in 2009. Regressing those numbers to a .300 BABIP would yield two extra hits. Using standard linear weights, that could produce something like 1.5 extra runs.

The other point mentioned was the Marlins’ 2009 strand rates. As a team, they stranded 73.5% of runners, a number I do not find all that anomalous. The team 17th in the league in stranding runners according to FanGraphs. The six individual pitchers mentioned above all had above average rates, but Badenhop and Meyer would only have a small amount of regression to go. My concern is with Nunez (2009 LOB% 79.4%), Pinto (79.5%), and Sanches (85.3%). CHONE projects them all to drop and have more runners score. Even though their BABIP were not extremely anomalous, the difference may have been in the distribution of those hits.

In all, the Marlins pen was worth 22.5 runs above replacement according to FanGraphs. That of course discounts things like timing and sequencing, meaning it generally assumed neutral defensive assistance by the Marlins. Looking at the 2010 version, the pen does not appear to project all that different than the defense-independent performance of 2009. Whether that means it drops the Marlins to an 83-win team remains to be seen, but the Marlins did not remain entirely static between 2009 and 2010. Something tells me replacing players like Emilio Bonifacio and Jeremy Hermida with (hopefully) more competent personnel like Cameron Maybin and Gaby Sanchez should give us a two to three wins back, for example.