Pitching and Fielding, Part Three
January 28, 2004
Introducing yet another way to allocate defensive Win Shares
Splitting Win Shares between pitching and fielding correctly is an elusive quest, like catching a shadow. Hard to see, hard to imagine, and you’ll never know if you’ve actually done it. At least not without play-by-play data. Before this is all over, I intend to incorporate some of Pinto’s and MGL’s play-by-play fielding data into our Win Shares spreadsheets, just to see what we get.
In the meantime, I’ll offer you another shadow.
I’d like to come at this fielding/pitching thing a little differently, along these lines:
One, Fielding Independent Pitching, or FIP, is a perfect stat for allocating Win Shares to pitchers, because fielders have no impact on home runs, strikeouts or walks, the components of FIP. I sometimes forget to mention this, but kudos to Tangotiger for inventing FIP.
Second, all the other runs allowed (non-FIP runs, if you will; I like to call them DRA) are basically determined by the team’s Defense Efficiency Ratio, or DER. If you need convincing, this article graphs the historic relationship between the two.
Third, DER is the responsibility of the pitchers as much as it is the responsibility of the fielders. Maybe more so. In other words, the state of a ball when it leaves the bat has as much, if not more, impact on DER than what fielders subsequently do.
You may gulp a bit at this one, because we’ve learned that pitchers don’t control the outcome of a batted ball as much as we thought. But the pitcher is the man on the spot. If a ball falls somewhere on a playing field where a great fielder cannot reach it, then the pitcher should bear that responsibility.
Fourth, and finally, pitchers may not impact DER very much, but they can make DER irrelevant. As we’ve shown in this article, pitchers with low and negative FIP’s reduce the range of DRA. If a pitcher has a FIP of -1.00, you just about know what his DRA will be. Fielders won’t impact it as much.
With that in mind, here’s my system (warning, this is very technical and math-oriented. Proceed at your own risk):
- First, compute the team’s runs allowed per game (RA/G) on a park-adjusted basis.
- Second, compute team FIP, using park-adjusted home runs.
- Third, compute team DRA (Defensive Runs Allowed?), which is runs allowed per game minus FIP.
- Fourth, compute all three of these figures for the league in total.
Now, create a baseline allocation of Marginal Runs Allowed to pitching. To do this,
- First, calculate the team’s Marginal Runs Allowed by subtracting its RA/G from 152% of the league’s RA/G.
- Second, use this formula: (league RA/G threshold minus (team FIP plus league DRA)) to establish an “A” factor.
- Third, use this formula: (league RA/G threshold minus (league FIP plus team DRA)) to establish a “B” factor.
- Fourth, divide “A” by (“A” + “B”) to calculate your initial percent allocation to pitching.
- Finally, multiply this percent times Marginal Runs Allowed to calculate Pitching Marginal Runs Allowed. We’ll call the leftover runs DRA Marginal Runs Allowed.
This step represents the portion of Marginal Runs Allowed that should be allocated to pitching based on only its FIP. For 2003, the baseline allocation ranged from 35% (Devil Rays) to 59% (Yankees).
Next, determine the portion of DRA Marginal Runs Allowed that should be allocated to pitching. The first step is kind of complicated:
- Start by allocating a percent of DRA to pitching based on FIP. Here’s my proposed formula: (60% - FIP/10). If FIP is less than 0, the percent allocation is fixed at 60%.
So if a pitching staff has a FIP of 1.00, it gets 50% of DRA Marginal Runs Allowed credited to it. If its FIP is 2.00, 40% gets credited to the pitchers. Whatever is left in DRA Marginal Runs Allowed is credited to the fielders.
This may throw people for a loop, but it’s tied to the idea that fielders behind pitching staffs with low FIP don’t contribute as much to DRA.
The 2003 pitching allocations, at this stage of the calculations, range from 60% (Rays) to 81% (Yanks).
The other thing I like about this approach is that every step, up to now, can be applied to individual pitchers. It’s always sort of bugged me that pitchers receive Win Shares via a different system than the team’s pitchers do. So, part of my proposal is that the system I’ve described, up to now, also be applied to each individual pitcher on the staff when individual Win Shares are calculated.
Tell you what. Think about it, and let me know.
After this step, I pretty much follow Charlie’s system. I use linear weights for wild pitches, errors, double plays and outfield assists, versus the league average, to move Marginal Runs Allowed between pitching and fielding.
As an aside, I don’t add claim points to the numerator and denominator. I’ve always been confused by that methodology. I think it’s easier to just move weights between the actual Marginal Runs Allowed.
So what’s the bottom line?
Well, the net effect of all this is that my system actually allocates more defensive Win Shares to pitching than James or Charlie. In the NL, 72% are allocated to pitching, vs. 68% in the original/James version. In the AL, the numbers are 71% and 68%. Like Charlie’s system, however, my results vary widely, from 59% to 83%. Guess which teams.
I will post a table with the results of each system by team. I also hope to explore these systems a little further in the next article, and then I plan to move on. Enough is enough!
Again, with the tweak I sent you (the linear strikeouts), I’m granting about 72% to pitching. The real key to these things is taking the clear non-pitcher portion and assigning responsibility, no matter what system you’re talking about.
I’m looking forward to reading the second article right about now ...
Page 1 of 1 pages
Commenting is not available in this channel entry.