Last week, as part of his Thursday Throneberries, Rob Neyer wrote — in re the Why We Watch post that I’d submitted to these electronic pages — he wrote that “the only thing missing [from said post]… is a points system that would let us put a number on each game.” That is, Neyer was curious if it might be possible to assign points to each game on a particular day in order to tell which might be most appealing to the sabermetrically inclined viewer.
Neyer’s challenge put a bee in my proverbial bonnet. And when you put a bee like that in Carson Cistulli’s bonnet — proverbial or otherwise — he’s not gonna stop until that bee is either dead or, if not dead, at least captured and successfully rehabilitated.
Which is why I’ve spent every waking minute of the last five or so days — and some of the sleeping minutes, too — working on the problem.
A few minutes of consideration reveal two facts:
1. It’s a big-ish task, this, to devise a points system for every possible aspect (pitching, hitting, uniform design, stadium, broadcast team, etc.) that might contribute to the viewing experience.
2. Despite the verity of point 1, it seems as though we can say with some certainty that pitching matchups — because the pitchers are constantly playing — go the greatest way towards making a game either compelling or not. Therefore, that’s where I’ve elected to start.
So the question I posed to my own brain is: what makes a starting pitcher interesting to the baseball nerd? And also: of the things that might make a pitcher interesting, which of them are easily measured? And finally: what ought one call a stat designed to address these urgent questions?
The last question is the easiest to answer. Were I to construct a stat designed to appeal to the baseball nerd, I’d call that stat NERD. What would/does it stand for? Hard to say, but it just feels so right.
Now, as for the first two questions there, let’s take a look at some possible answers.
Components of Pitcher NERD
• Pitcher Ability (xFIP)
At the center of the baseball nerd’s quest is the desire to understand, quite simply, who is good at baseball and who is less good. While, as Tommy Bennett rightly notes, metrics that evaluate process (as opposed to outcome) aren’t flawless, Expected FIP (xFIP) is both pretty damn sweet and pretty freaking accessible.
• Strikeouts (SwStrk%)
Swinging strikes correlate very highly to strikeouts — are, in fact, more predictive of future K rate than K rate itself. Also, they’re awesome to watch. Yes, strikeouts are a part of xFIP, but there’s a pleasure to the strikeout that ought to be recognized. It represents a pitcher’s total pwn-ing of his opponent. Consider: despite having absolutely no ties to San Francisco or its environs, I consider it a great privilege to watch Tim Lincecum throw his change-up. Yeah, the crazy wind-up is pretty sweet, but his change-up — which gets whiffs about 25% of the time — is what really gets the party started.
• Strike Throwing (Strike% of Total Pitches Thrown)
It’s nice to watch a pitcher who throws a lot of strikes — even those not of the swing-and-miss variety. Phrased differently, it can be super boring to watch a pitcher who doesn’t throw strikes. Consider Rich Harden of 2009. He had a swinging-strike rate of about 16% in 2009, but he only threw about a league average number of strikes. That’s enough to make him less watchable. (And, of course, he’s much harder to tolerate this season, now that he’s getting about half the swing-and-misses.)
• Luck (ERA-xFIP)
It’s a fact: nerds like watching regression happen. Scientists are efforting day and night to figure out why — to no avail as of yet. Anyway, it’s a fact.
A Note on Weighting the Components
Though each of the components listed above probably all contribute to making a pitcher interesting, that’s not to say that each of them ought to be weighted equally. Value luck too heavily and Charlie Morton is the most interesting player in the world. Do the same to swinging strikes and Brandon Morrow makes his way to the top of the charts.
As the goal of this exercise is not to uncover an objective truth — like, for example, how many runs a player has produced or how much he might be worth on the open market — but rather to address questions of an aesthetic nature, I’ve allowed myself to abide by intuition in assigning weights to the components in consideration.
Here’s how I’ve opted to go about it.
Calculating NERD
To calculate NERD, I found each pitcher’s z-score (standard deviations from the mean) for cats 1-3. I multiplied the xFIP score by 2, divided both the swinging strike and strike percentage scores by 2, and then added Luck to the total.
In re that last part, about luck, a couple notes: I opted only to add bad luck to the overall equation. A pitcher who’s overperforming his xFIP is less interesting to me (and to baseball nerds, in general, I’m guessing) than one who’s underperforming it. If a pitcher’s ERA-xFIP is less than 0 (i.e. lucky), I just counted as 0. In other words, I only real care about pitchers who will seem to be improving.
Also: I capped the Luck “bonus” at 2. Otherwise, seriously, Charlie Morton is the highest-ranked pitcher.
Adding a constant (in this case, 4.69) gives all 150 or so pitchers (with 20+ IP) a score between 0 and 10, with average exactly at 5. I had to round the top two guys down to 10 and the bottom three guys guy up to 0, but that’s it.
The Final Equation
Looks like this:
(xFIPz * 2) + (SwStrk%z / 2) + (Strike%z / 2) + Luck + 4.69.
Results
Here are the current top 20 starters (with 20+ IP) by NERD:
Here are the bottom 20:
Discussion
Bill James once suggested that, if a stat never surprises you, it’s probably worthless. This is an idea I embraced while attempting to fine tune NERD. Which is to say, I wanted it to be occasionally surprising.
For example, one might wonder how Randy Wells ranks higher than the very talented Tim Lincecum. Well, in addition to actually being quite good so far this season, Wells’ ERA is almost a full point above his xFIP. With time, the former is likely to crawl back down towards the latter. When it does, Wells will very likely not occupy his current spot. Still, in the meantime, it’s worth wondering when we’ll see Wells’ luck turn for the better.
Future Considerations
Pitcher NERD is definitely not complete. Undoubtedly, it makes sense to consider at least a couple-few more components. Components such as (though not limited to):
• Player Age/Experience
Rookies are exciting. Young players, generally, are exciting. It makes sense to factor something like player age or service time into the equation.
• Repertoire
My good friend Leo writes, “Should a guy like Justin Verlander be higher because he throws 100 MPH? That’s fun to watch.” The same friend also would like see Ubaldo Jimenez further up the charts. And who can blame him: watching Jimenez pitch is fun.
• Fat Heads
Vincente Padilla has one. That should count against him somehow.
If you’re interested in fooling around with the weights, by all means utilize this spreadsheet that I’ve uploaded to Google Docs. It also includes data for 2009, which saw Javier Vazquez, Ricky Nolasco, Roy Halladay, Tom Gorzelanny (!), and Tim Lincecum finish in the top five.
