On top of that, $11.5 million is paying Nolasco as if he’s approximately a 2.5-win pitcher. Nolasco easily exceeded that in both 2008 and 2009 and even matched that in an injury-shortened 2010 season. All the Marlins need is for him to do is repeat 2010 three more times and they’ll have broken even. Is that feasible for Nolasco?
Of course, this sparked yet another argument about how FanGraphs chooses to calculate its pitcher Wins Above Replacement (WAR). As we all know, FanGraphs uses FIP to calculate WAR, while Baseball-Reference uses Rally’s WAR formula which takes straight runs allowed and adjusts for quality of oppnent, then subtracts defensive run impact. I mentioned before that at times these methods yield drastically different results, and that in particular was the case for Nolasco in 2009.
So the question that is once again asked is how can we determine how Nolasco will pitch in the future? Are his recent 2009 and 2010 struggles (of varying degrees, of course) a product of simple bad luck or is there something more to it that needs to be considered when accounting for WAR? Let’s review what each version of WAR measures and see how these match up with Nolasco’s game.
Another lame (pitcher) WAR joke
FanGraphs pitcher WAR uses Fielding Independent Pitching (FIP), a statistic developed by Tom Tango that looks at ERA if it were solely dependent on K, BB, HR. As Tango incessantly points out on his blog, FIP does not take a stance on balls in play; it treats balls in play in a similar fashion that OBP does in equating walks with home runs.
Anyway, as I keep saying: it’s not a problem that OBP treats a walk identically to a HR. So, it’s not a problem that FIP ignores all in-park batted ball events. That’s a feature, not a bug.
Of course, what Tango is saying is that FIP is just a part of a pitcher’s game. The key is that, unlike OBP, it actually turns out that FIP’s components are a major part of a pitcher’s game, encompassing a good deal of a pitcher’s value. FanGraphs takes that stance a step further and claims that this FIP’s components contain enough of a pitcher’s value that the remainder will be insignificant in terms of WAR. This WAR value is also park-adjusted.
On the other hand, B-Ref’s version of WAR, initially developed by Rally, adjusts for a few more things. The quality of opposing hitters is adjusted for, as is park. In addition, B-Ref starts with pure runs allowed and and backs out the defensive contribution based on its defensive stat TotalZone. This seems like a good idea, but it too has some flaws:
1) It assumes a defense performs equally behind all pitchers, which is hardly the case.
2) It relies on a play-by-play defensive metric, which can have its own problems (search for Colin Wyers’ articles on Baseball Prospectus for more on that)
So both versions of pitcher WAR are imperfect, which cannot possibly be surprising, since after all, we are making our best-effort guesses at separating pitcher performance from a variety of different factors (defense, ballpark, and opponents among them) using imperfect tools. Still, what do we know about both of them?
1) FanGraphs WAR (fWAR) accounts for defense (via FIP) and park
2) Rally’s WAR (rWAR) accounts for defense (via TotalZone runs subtracted from pitching performance), park, opponents, and sequencing (impact of pitching with runners on versus bases empty)
How does this affect Nolasco? Well, his 2009 season was very unlucky for a variety of reasons, but he was also quite lucky in his 2008 year as well. Let’s compare some of the factors for each of those years:
As has been harped on many times before, these numbers look pretty similar over the last three years. The major differences came in BABIP, and if you just had to guess based on a quick survey of these numbers, you would estimate that Nolasco should have a BABIP closer to 2010′s .328 rather than 2008′s .284 due to the relative proximity of the 2010 data. If you buy into defense-independent pitching statistics (DIPS) theory (and believe me, you should), then you have to figure that a large portion of that BABIP is the responsibility of the relatively porous Marlins defense, and that Nolasco should still be somewhere in between the leaguewide .300 average and Nolasco’s career .314 mark.
But there is a problem that continues to persist that I have examined in the past and of which I was reminded once again by The Hardball Times’ Vince Caramella. The problem of how Nolasco pitches between when there are runners on as opposed to the bases empty. Check out the xFIP splits since 2008:
|Year||xFIP, Bases Empty||xFIP, Runners On|
Nolasco’s actually been stingier on home runs with runners on, but xFIP yields league average HR/FB% for the home run input for FIP, and Nolasco’s career HR/FB% is pretty close to the league average. In each of the three seasons, encompassing two good and one bad year, Nolasco has pitched significantly worse in terms of strikeouts and walks with runners on as compared to bases empty situations. At the end of last season, I examined this phenomenon and saw that Nolasco was wider with his zone with runners on, which could lead to him getting behind in the count more often. Whether this effect has anything to do with his mechanics out of the stretch remains to be seen, but I suspect the Marlins know something about this.
The key is that, in WAR calculations, Rally’s methodology on B-Ref is more likely to incorporate any problems that Nolasco has in terms of skill with runners on, whether they include skill differences out of the stretch or pitch selection with runners on. If the split that we have seen is repeatable because of some skill, then rWAR is more likely to capture Nolasco’s true value then fWAR will. In other words, because of this potential aspect in his game, we would expect Nolasco to regress to a mean generally higher than his FIP or xFIP.
It seems the Fans and projection systems feel the same way. FanGraphs’ projection page for Nolasco has the projections from both the Bill James Handbook and the Fans (13 in total). Bill James projected a 3.92 ERA, while the Fans have projected a 4.10 ERA. Dan Szymborski’s ZiPS system projected a 3.78 ERA. My own crude projection has him at arounda 3.93 ERA, so each of the four systems mentioned here have Nolasco at around the same value. Taking just his ERA puts him at a 3-win value, whihc more than justifies his contract for the future, regardless of the supposed fear of his runners on numbers or his large ERA/FIP split. As Marlins fans, I think we should all be happy; expect some regression for Nolasco, but know that he isn’t the guy from 2008 just like he isn’t the guy from 2009.