Wednesday, November 17, 2010


Here in the interstice between Cy Young Award announcements, it might be worth a few moments to trot out a little method for ranking the candidates.

It's always struck me that relying on any single statistical measure--whether "advanced" or dinosaur-like--is the quintessence of what I've taken to calling "Baseball Stalinism" (and you know what those initials spell). Hence a method that incorporates a series of new and traditional "metrics" (a word that really deserves to die a horrible death...) is going to be preferable.

With that, here's how what I'm calling "Weighted Cy Young Award Points" works. We have seven categories (as you'll see on the tables delineating the results for AL and NL 2010 below):

--Wins Above Replacement (WAR): as taken from
--Wins (W): be still, faint of heart!
--Winning Percentage (W%): see above...
--Adjusted Earned-Run Average (ERA+): as taken from
--Adjusted On-Base Plus Slugging (OPS+): the batters vs. pitcher version
--Win Efficiency (WEff): WAR measured against batters facing pitcher (BFP) totals
--Innings Pitched (IP): for durability...

Before anyone raises the now-popular objection about W and W%, understand that these categories are weighted. Despite the pronouncements of many, there is no reason to completely dismiss these stats: they continue to tell us something. They simply don't tell us as much as other stats, however. That's why in the "WCYAP" method, they have the lightest weighting.

We weight WAR and ERA+ the most (four), OPS+ and IP next (three), WEff next (twice). W and W% are both single-weighted. (The weightings are displayed on the diagrams below, and color-coded just for "fun.")

Why bother with a stat such as Win Efficiency? Well, pitchers whose WAR is achieved in fewer BFP have  added another type of value to their team that isn't really accounted for anywhere else, and after looking over the historical stats, it just made sense to include it here.

Let's look at the National League first.


Remember, a "5" actually means that the pitcher is first in the category, while a "1" means that they finished fifth.

Oh yes, I forgot to mention that there are "bonus points" awarded for the number of categories that a pitcher ranks in the "top five." As you can see, only two NL pitchers made it into the top five in all seven categories (Halladay, Wainwright).

The results here indicate that Jimenez and Johnson were closer to Halladay than the voting results would tend to indicate. I'd have little or no problem casting a ballot based on the results shown above.

Now let's go to the American League.


I was surprised how well the system graded Buchholz, but he shows up well in the heavily weighted stats, which allows him to gain ground on the competish. Nevertheless, the method taps Felix Hernandez as the top pitcher in the AL during 2010. What the method shows is that the accomplishments in the AL were spread around a good bit more this year than they were in the NL.

One minor drawback here is that we can't use the method all the way back in time: OPS+ data is not available (at least not yet) prior to 1950. But I can live with that if you can.

Our results here don't produce any real "award controversy." (Aw, shucks.) But possibly we can walk away with a more nuanced look at how the various stats can be combined to produce a ranking that takes into account as many perspectives as possible. As we used to say at BBBA, 'tis a consummation devoutly to be wished...