This is a post introducing and explaining NERD scores for teams. I’m including the results first and then the background, methodology — all that junk — second.
Curious as to what NERD is? The short answer is: it’s a number, on a 0-10 scale, designed to express the “watchability” of teams for those of the sabermetric persuasion.
For more information, consult the index right after the results.
Here are the results for team NERD:
If you’ve read this site with some regularity over the last couple-few months, you know that one particular concern of mine has been to develop a method by which the sabermetrically oriented fan might be able to choose, with some accuracy, the most compelling game of any particular day/night.
It was with that goal in mind that, at the beginning of June, I submitted for the readership’s consideration a metric called NERD. The object of NERD was/is to grade on a 0-10 scale each starting pitcher’s likely interest to those of the sabermetric persuasion. Adding together z-scores (or, standard deviations from the mean) for a number of categories — xFIP, swinging-strike percentage, age, etc. — I was able to approximate the entertainment value (again, to the sabermetrically oriented) of each pitcher with 20+ innings as a starter.
Because much of the readership is composed of curious and occasionally not-that-cruel nerdicators, a number of thoughtful suggestions were given towards the improvement of NERD, some of which (i.e. the suggestions) were incorporated into a newer iteration of the metric. Thus, NERD was refined into something more or less useful.
Over the last few months, I’ve used those pitcher NERD scores to inform the One Night Only game previews I submit here. It’s worked well, I think.
Or, I should say, it’s worked well except for its obvious weakness — that is, the fact that it entirely omits everything besides starting pitching. Of course, one can eyeball the relative pros and cons of these other things. One knows, for example, that, because they’re young, fast, talented, and constructed by a progressive front office, that the Tampa Bay Rays are bumpin’ like the City of Compton. Likewise, there’s this year’s iteration of the San Diego Padres, who began the season with a new GM and the league’s second-lowest payroll, and currently sit atop the NL West standings.
On the other side of things, there are teams like the Astros and Royals: teams whose roster-building we’ve decried here with such regularity that it would be indecent to enumerate their flaws once again.
But recently, I’ve begun to think that eyeballing such factors isn’t enough — that, if we’re gonna get this NERD party started, we better get it started right.
So it was, this past Friday, that I broke out my disgustingly sleek netbook, double-clicked the crap out of the Excel icon, and set to work on a team-level NERD.
It was also about that time that I issued the following SOS via the Twitters:
Hey, help. Trying to make a NERD score for teams. What do people like to see? I have Avg. Age, for one. HR/FB? O-Swing%? Team UZR?
It needs to be said: the responses were enthusiastic and bountiful. To everyone who played along in this sordid game, I salute you.
Now, does that mean I took them all (i.e. the suggestions offered by the Twitterverse)? Hell, no! But — truly, no BS-ing — even the suggestions that I didn’t take were helpful, and helped inform the components that ultimately ended up in the present version of NERD,
What are the components, then? Here they are:
Components of Team NERD
• Age (AGE)
Young players are more fun to watch — for a couple reasons, probably. For one, their peak years are still in the future, so it’s possible to dream about what they might do. For two, younger players tend to be cost-controlled. That’s bad for them, of course, but it’s a fact: talented, cost-controlled players tend to populate well-constructed (and, thus, nerd-friendly) teams.
• Park-Adjusted wRAA (BAT)
In an early version of the team-level NERD, I didn’t take actual, you know, production into account. The results were fine, but the A’s were like third. This looked odd to a bunch of people, and I couldn’t blame them.
• Park-Adjusted Home Run per Fly Ball (HR/FB)
Home runs are great. (If you don’t believe me, watch what Russell Branyan did to a baseball the other day in the Bronx. The ball literally disappears.) Why park-adjusted HR/FB? To neutralize the effects of the team’s ballpark*, is why. I used Dan Turkenkopf’s four-year weighted numbers.
*I discuss it down below, but will also mention it here: it’s my intention to eventually add a NERD “park factor,” too, based on average attendance, home run park factor, perhaps broadcast team, etc.
• Team Speed
Team speed certainly needs to be part of a watchability index, but the question is how to incorporate it into a team-level NERD. Like this, I think is the answer:
Stolen Base Attempts per Opportunity (SBA): This is a measurement of how often a player attempts to steal when the base in front of him is unoccupied. Obviously, teams that attempt more stolen bases are creating more action. Action, generally speaking, is pleasant to watch.
Stolen Base Runs (SBR): Of course, stolen base attempts alone shouldn’t be what we regard as great and good. No nerd worth his TI-82* is gonna ignore stolen base efficiency. For example, the White Sox are second in the majors with a 10.3% SBA, but are second-worst in the majors with -4.6 runs (per linear weights, that is) on said attempts. Smart Ball, indeed.
Extra Base Taken Percentage (XBT): This is a measurement of the times a player takes an extra base on a hit — like from first to third on a single, for example. Or first to home on a double, also for example. League average is about 40%.
*Is that what the kids are using these days?
• Bullpen Strength (BULL)
How exactly to measure the value of a bullpen is still a matter of some debate among the sabermetric community; however, the aesthetic virtues of a shutdown bullpen are not. The measure here is the xFIP of the bullpen as a whole. Padre relievers have a collective 3.05 xFIP, over three standard deviations above the mean!
(Note: I was unable to find, per Drew Fairservice’s request, BMI numbers for all the major league bullpens. Sorry, dude! Would’ve worked great.)
• Team Defense (UZR)
Defense is this year’s black. Actually, it was last year’s black, too. It’s probably gonna be next year’s black, as well.
• Luck (LUCK)
I’m quoting myself when I say, “Nerds like watching regression happen.” It’s as true now as it was all the way back in June of this year when I first wrote it. The question is: How do we define “luck” here?
In this case, I’ve subtracted actual runs from Base Runs, or:
Base Runs – Actual Runs Scored
If you haven’t seen it yet, internet denizen John Wright maintains a spreadsheet with BaseRuns standings at his site.
• Payroll (PAY)
This just in: Moneyball was neither written by Billy Beane, nor is it about filling a team’s roster with plodding walkmeisters exclusively. It is about building a ball club as efficiently as possible, and — especially in these WAR-filled days — such a thing has become of paramount interest to the nerd. Hence, its inclusion.
To calculate NERD, I found each team’s z-score (standard deviations from the mean) for all the categories above (except Payroll, which I’ll mention post-haste). I multiplied AGEz by 2, multiplied each of the speed factors by .33 (so’s to have one fully weighted speed category), and then added all the z-scores together.
As with pitcher NERD, a couple of categories act as bonuses only (i.e. no negative scores included): age and luck. In the case of age, it occurs to me that it’s not necessarily bad watching older teams; it’s just more excellent watching younger teams. With regard to luck — well, this might be a case of editorializing on my part. I’m certainly not above deriving pleasure from another’s misfortune, but I like to reserve that sort of joy for people I really dislike — not, you know, Livan Hernandez.
Also, note: I capped the Luck “bonus” at 2. Otherwise, the Orioles creep up like 3.8 points.
Adding a constant (in this case, 3.21) gives all 30 teams a score between 0 and 10, with average exactly at 5. I had to round the top two teams down to 10 and the bottom two to 0, but that’s it.
Why I Counted Youth Twice as Much
Probably for a bunch of reasons, but mostly because of the Pittsburgh.
The Final Equation
Looks like this:
(AGEz*2) + BATz + HR/FBz + (SBAz*.33 + SBRz*.33 + XBTz*.33) + BULLz + UZRz + PAYz + LUCK
It’s important to remember that this is strictly a measure of what has happened. Like, I personally enjoy watching the Phillies, as Jayson Werth and, in particular, Chase Utley are very clearly kings among men. But Utley (along with Placido Polanco and Shane Victorino) has missed time this season. The addition of Wilson Valdez and whomever else in Utley’s place — well, that’s not nerd-friendly viewing. The Phillies’ score reflects that fact.
As for the top few teams, I think they’re pretty inuitive. The Rays are not only young and well-run, but — in case you haven’t heard — the subject of a forthcoming book written by a nerd! The Diamondbacks have made some very questionable moves of late, but remain young, athletic, and power-y. The Rangers have a low-ish payroll, a serious MVP candidate, and some pretty serious glovemen.
Nor ought the bottom teams surprise anyone: the Royals and Astros finished 29th and 30th, respectively, in our organizational rankings. The Dodgers are 11 games out despite an old-ish, well-paid roster. (Note: Giving over 500 combined PAs to Garret Anderson, Brad Ausmus, and Jamey Carroll is not good for watchability). The Orioles have Matt Wieters, yes, but they’ve also been terrible in almost every way imaginable. Additionally, Garrett Atkins.) The Mariners? Well, they’re kinda fast. That’s about it, though.
Is NERD complete? No. Adding a team-level variety is definitely an improvement, but there are at least three stones that remain unturned, as follows:
• Home Park Adjustments
What about a park makes a game watchable (on TV, in particular)? That’s a good question. My guess is, the answer includes the following factors: (a) attendance as a percentage of stadium capacity and (b) the park factor of the field in question.
Vin Scully makes a game better. Don Orsillo and Jerry Remy, I’d argue, make a game better. Joe Buck makes a game borderline unwatchable. How do we integrate that into NERD? That’s the question that we’ll need to answer.
• Individual Player NERD / Adjustments
As I mention above, the absence of Chase Utley, et al., really hurt the Phillies’ NERD score. Ideally, it’d be possible to represent Utley’s return immediate, rather than waiting for the team totals to catch up.
Using NERD to Preview a Game
I’ll discuss this in a post later today. Probably. I think.