June 24, 2013
The Zobrist Effect
Last week, I began discussing a question that has puzzled the sabermetric community for a while. How do we put a value on a player's ability to play multiple positions? Most teams have guys who are capable of pulling duty at several places on the field, but they are bench/utility players who serve as backups. What to make of the player who hits well enough to be a starter and fields well enough at multiple positions to be worth starting there?
And so we introduce the Zobrist Effect, named for Ben Zobrist, the Tampa Bay Rays second baseman/right fielder/occasional shortstop. Much of the value that someone like Zobrist brings to a team isn't directly because of what he does on the field. It's in how Zobrist allows the Rays to construct their roster and use players to take advantage of the fact that they can move Zobrist around, if they need to. Perhaps we might begin to put some numbers to that fungibility.
In my estimation, there are three things that a versatile player allows a team to do. First, he allows a team to build a mixed position platoon (whether by handedness or some other factor) that they can use. Second, he allows the team to shift guys around in the late innings and replace an all-bat fielding butcher in one position with an all-glove no-hit player at another (with our super-utility guy bridging the gap). Thirdly, positionally flexible players enable a team to be much more efficient when it comes to giving players days off. But let's take each in its turn and see how the Zobrist Effect might play out.
Then again, the platoon effect is one of the strongest statistical effects there is, usually checking in around 15-20 points worth of OBP in the league-wide aggregate. Champions are made by such shifts in OBP. The thing about constructing a good platoon is that often, the two players must play the same position. After all, if you're going to pull your righty-mashing third baseman against a lefty, someone has to play third base. What are the chances that a team can find two players who are of an opposite handedness, but roughly equal hitting talent (otherwise, one would just start), who play the same position, are worth the roster spot for their production from one side, and are happy playing in a platoon role? Because of the Zobrist Effect, a team might not have to find two players who both play the same position, just two guys whom the super-utility guy can be the bridge between. It expands the number of players that they might use to cobble together a platoon.
Exactly how valuable is that platoon effect for fringe regulars and other part timers, though? One thing to note is that the platoon advantage for this group is likely to be bigger than it would be for the general population, because the general population is loaded with players who are regular starters against both sides and have that role because they don't have much of a platoon split.
To look at this, I used the odds-ratio method to isolate all cases where a batter who had between 100 PA and 400 PA in the season in question (2003-2012) was at bat and opposing a pitcher who faced at least 250 batters. I looked at the expected OBP of the matchup, coded whether the batter and pitcher were of the same hand or different handedness, and ran a logistic regression using both inputs. The effect for fringe regulars and part-timers of the platoon effect was about 22 points’ worth of OBP. For regulars (400 or more PA), it was 19.
That means that when constructing a platoon, the type of players who would be playing in a platoon are more likely to get a slightly bigger bang for their platoon advantage than is generally expected. And roughly 22 points worth of OBP gained by the platoon advantage is one on-base event every 45 plate appearances. Each position on the diamond gets roughly 700 PA per year. Players who had more than 600 PA held the platoon advantage over their pitching opponent about half of the time. I suppose that even if a team could somehow finagle it (by breaking the rules of substitution) that the "correct" half of the platoon was always at bat, those extra 350 PA in which a full-time player might have given up the handedness advantage would be worth an extra seven or eight on-base events (around a win?) Of course, that would never actually happen, and the effect would be smaller, but every little bit helps. Even if we're talking about two- or three-tenths of a win, a two-win player is nothing to sneeze at and he may have just added 10 percent to his output by simply having multiple gloves in his locker.
Individual situations may vary, of course. A team might find itself holding one player who hits .350/.420/7 gagillion against lefties, but can't hit righties, and another guy who does the reverse. A platoon does improve the results that a team can expect out of the combination of two otherwise flawed hitters, and our multi-instrumentalist makes it more likely (but does not guarantee) that a decent platoon can be fashioned.
A few years ago, I created my own toy defensive metric (OPA! or Out Probability Added Above Average). I estimated that a good defender is worth between .01 and .02 runs per inning above the average player at most positions and that a poor fielder kills about as much value. We learned last week that there is no reason to believe that moving positions within games saps fielders of their defensive prowess. Let's assume that our utility infielder is .015 runs per inning better than the average fielder and that our Zobrist is an average left fielder, but represents an upgrade of roughly the same amount over the butcher. The problem is that while they contribute .03 runs per inning of value, they will get to do it only in the late innings of games where it makes sense to make that move. Over 80 innings (say, the ninth inning of roughly half of games?) that will net a team 2.4 runs. Not bad, but not earth-shattering. If a team has this sort of roster construction by chance, it's not a bad strategy to follow, and the multi-position ability of our Zobrist does facilitate that extra value.
Efficiency in Handling Days Off and Injuries
A multi-positional player allows a manager to make moves like this. Hitters who ride the pine do so for a reason. Either no one has yet recognized their brilliance, or there wasn't any brilliance there to recognize. But there are still differences between the best guy on the bench and the worst. For example, from 2003-2012, the 10th-most-used hitter on a team (by plate appearances) had an average (weighted) OBP of .312 (he averaged 284 PA in a season). The 11th-most-used checked in at .310 (238 PA). When you got down to no. 13, it was an OBP of .298 in 163 PA. Being able to use the first guy on the bench to replace a guy who is hurt or who needs a day off is worth 14 points of OBP over the last guy on the bench.
Yet how often does the light-hitting utility infielder play because he's the only one who can fill in at shortstop? What if we could take some of those 163 plate appearances (or the 129 that the 14th-most-used guy gets, in which he hits for an OBP of .294), and give them to the first guy on the bench?
There's an interesting thing to note about those numbers. The realization, whether conscious or unconscious, that the platoon advantage was worth 15 to 20 points of OBP (not to mention a good amount of other stats) led to the development of an entire genre of pitcher, the LOOGY. This despite the fact that a LOOGY might have only a few dozen high-leverage plate appearances in which his LOOGYness is specifically needed. What then of a slightly smaller swing in OBP (12 points) but one that could be applied to a larger pool of plate appearances? Would that not be worth developing a genre of player? Why has there not been the same evolutionary pressure to develop multi-position players? It probably has to do with the fact that to make a LOOGY, you just have to tell a guy to throw with his left hand, which he was going to do anyway. Creating a multi-positional player takes more work.
I recognize that you're not going to take just any second baseman and turn him into an elite shortstop, or a right fielder into a good third baseman, but perhaps a franchise can find a guy who projects as a regular hitter and can be passable in a few different spots? This way when the injury bug bites (and it will) or a player needs a day off (and with apologies to Cal Ripken, they all do once in a while), a better quality of substitute can be used.
Again, if you stole half of the plate appearances from guy no. 13 on the bench (80 PA or so) and from no. 14 (65 PA), and maybe a few extras down the line, and gave those 150-200 PA to guy no. 10, you’d improve them by 12 points of OBP and squeeze out an extra run or two over the course of a year. Again, it's not a huge effect, but it's a competitive world and everything counts in large amounts. (A billion points for the reference.)
A little extra value
But as always, let's keep things in perspective. There is added value to being versatile that doesn't show up in a seasonal stat line. It comes in the form of allowing a team to cover weaknesses on other parts of the roster. But it's not a huge amount, and maybe instead of developing more versatile players, a team might do better to figure out why that third baseman can't hit lefties to begin with.