keyboard_arrow_uptop

I’m guessing that many of the people reading this will have come to appreciate the game of baseball the way I did, through the magic of baseball cards. When I was growing up, a baseball card told you three things on the front of the card: the player’s name, the player’s team, and the position that he played. It’s something that’s deeply ingrained in how we understand baseball. Players have their positions. And we know all nine of them. Well, of course, there were the utility infielders. I remember when I was 7 that I called them “switch-position” guys. They got little dashes in the little circle where it showed their position. SS-2B. 3B-1B. In baseball, it’s perfectly legal to shift around the park, but the good players, they had a regular spot to hang out in.

Now that I’m older, I can appreciate the importance that the position that a player holds down has in terms of determining his value. The idea of the defensive spectrum is something that’s been around since Bill James introduced it in the 1980s. James initially conceptualized the defensive spectrum by noting that there were several shortstops in baseball who held down regular jobs despite being awful hitters. There were no first basemen who could make the same claim. James lined up the positions in terms of their average production and found that it neatly aligned with the accepted wisdom of which positions were toughest and easiest to play. Center fielders and shortstops were allowed to get away with .230 batting averages because those spots are the hardest to play. If a first baseman tried to hit like that, he wouldn’t be a big leaguer much longer.

When we try to figure the “true” value of a player, we have to take into account his position, but how we do that is a little tricky. The most common way—at least when putting together uber-metrics like WAR—has been to use a positional adjustment to give a player credit for playing a more difficult position (or penalize him for playing an easier one). Smith might not hit very much, but the fact that he can play a competent shortstop means that third base wasn’t clogged up and his team is more likely to find someone who can hit if it dips into the third base bin rather than scrounge in the shortstop bin.

But that works the other way around too. The assumption also runs that if Smith’s team were to find another shortstop whom they liked, Smith could easily slide over to second base and the fact that he can play an average shortstop is evidence that he would be an above-average second baseman and an amazing left fielder, if needed. We assume that while his offensive performance might be even worse compared to his new peers at second base, his defensive value would go up. He gets some credit for that. When comparing Smith to a left fielder with a great slash line but so-so defense, we should consider that in the Smith’s favor.

I think there’s a very flawed assumption behind that logic. It implicitly conceptualizes defensive ability as a single dimension. A spectrum, if you will. You’re either good at “defense” or you are not. It’s like saying that “intelligence” is a single thing, but then mumbling when asked to define what that thing is. Instead, I propose that we take a fresh look at the defensive spectrum through the lens of the actual data.

Warning! Gory Mathematical Details Ahead!

Let’s talk about what’s happened between 2011 and 2015. And first things first, let’s start with what everyone knows to be true intuitively: catchers are a completely different animal. We can see that from the data. The obvious way to study how well a shortstop would do at second base is to look at players who have spent time in both spots. The problem with catchers is that you don’t get many case studies.

Among catchers who spent more than 360 innings at the position in a season, there were very few cases where they spent even a small amount of time (I used 27 innings as a cutoff). Less than 10 percent (9.6 percent) moonlighted as first basemen, and there were a couple of isolated cases of playing left and right field. That’s it. On the flip side, among those who spent 360 innings at first base, only 1.5 percent got any meaningful time as a catcher (probably some of the same people as above). There was one regular third baseman who moonlit as a catcher (Jordan Pacheco, 2012). That’s really it. Catchers should really be put in their own box.

Let’s move on to the other seven positions then, shall we? Specifically, let’s move on to shortstop, because shortstop presents us with a vexing question. What do we do with the utility infielder? Suppose you have a player who plays 25 games at second, 15 at third, and 10 at short. When trying to figure out his positional value, should we consider him a second baseman? Should we prorate his work over the three positions? Should we look at the fact that he proved that he could handle shortstop at the major-league level? What would happen if a utility infielder played 150 games at shortstop?

Let’s look at this from an empirical angle. I calculated how many innings that a fielder had spent at each position other than catcher. He needed 360 total innings in the season to be included in the sample. I calculated the percentage of those innings that he spent at each position, and then subjected all of those player-seasons to a hierarchical cluster analysis (for the initiated, I used Ward’s method.) For those familiar with cluster analysis, you know well that interpreting it is one part art and one part science. I finally settled on a 14 cluster solution. The idea is that each player season is treated as a unique case (and there were 1,549 of them!) and the program begins by looking for the two cases which are most similar. For instance, two cases where a player spent his entire season at first base. In that example, the program would see that they are identical cases and “cluster” them together. It then repeats the process until it finds no more identical cases, but perhaps turns up a guy who was mostly a first baseman but did a little emergency duty in left field one day. Eventually, it gets to the point where you’re connecting things that don’t make sense to put in the same basket. That’s the point where you stop.

The 14 clusters were all fairly recognizable player types. Seven of them, not surprisingly, were guys who spent more than 90 percent of their time at each of the seven non-battery positions. Then, there were 7 utility types:

· The jack of all trades guy (49% 2B, 19% 3B, 16% SS, 8% LF, 4% 1B)

· The fourth outfielder who is primarily a center fielder (53% CF, 29% LF, 11% RF)

· The fourth outfielder who is primarily a right fielder (53% RF, 25% LF, 20% CF)

· The fourth outfielder who is primarily a left fielder (68% LF, 18% RF, 9% CF)

· The guy that the team is just trying to hide somewhere (45% 1B, 40% LF)

· The more traditional utility infielder (41% SS, 38% 3B, 13% 2B)

· The utility-ish guy infielder who doesn’t play short (62% 3B, 14% 1B, 13% 2B)

But back to looking at shortstops… or at least people playing shortstop. There are of course guys who have “shortstop” on their business cards, but we also see a couple of guys (jack of all trades, traditional utility infielder) who also play there at least on a non-cameo basis. And occasionally, some of the other guys randomly end up there as well. What happens when they all try to play shortstop? I used revised zone rating (weighted by the number of “in zone” plays) to look at how many balls each group turned into outs for each cluster group when they were playing shortstop. Then I looked at the average number of “out of zone” plays made per inning in the field for each group as well.

Cluster

RZR

OOZ plays per inning

Regular Second Basemen

81.0%

.0621

Regular Shortstops

80.2%

.0550

Regular Third Basemen

75.8%

.0495

Utility Infielders

80.0%

.0509

Jack of all Trades

79.3%

.0566

Everyone Else

81.5%

.0517

Regular shortstops converted 80.2 percent of the balls hit in their “zone” into outs. Jack of all trades guys nailed down 79.3 percent, and utility infielders came through at a 80.0 percent rate. Oddly enough, second basemen playing at short and guys making cameos(!!!) wound up with better numbers than the regular shortstops! But, #SmallSampleSize.

Looking at utility infielders for a second, we see that they get to about 0.2 percent fewer balls in the zone and make .004 fewer out of zone plays per inning than a regular shortstop. Over 1,400 innings and 400 balls hit to a regular shortstop over the course of a year, that’s 0.8 extra plays not made on balls in the zone and 5.6 balls not gotten to out of the zone. Assuming that those are all singles (and assuming that turning a single into an out is worth about 0.75 runs), we can assume that your average regular shortstop is about 5-ish runs better over the course of a season than a utility infielder, assuming both played a regular shortstop’s workload.

Jack-of-all-trades guys (at least the “average” one), if they were forced to play short on a regular basis, would miss 3.6 extra balls in the zone, although they might get to a couple extra balls out of the zone that a “real” shortstop would have gotten to. Of course, mileage will vary by the individual player, but the average utility infielder probably could handle shortstop at pretty decent level. The fact that he gets most of his appearances at second and third is much more a matter of circumstance. It seems a shame to penalize him for that.

(Sidebar: It’s possible that we have a sample bias problem here. Managers might have a guy who would actually make a dandy shortstop, but is a jack-of-all-trades type more through circumstance than anything. These also may be the guys who actually end up playing short more often than the jack-of-all-trades guys who should really only play there in emergencies. This is the trouble of trying to do comparisons across positions. Only the ones that the manager thinks can handle it end up doing it. Since we pay managers to introduce bias into the sample, it’s possible that this is what we’re seeing.)

Let’s take a look though at whether shortstops really do make better second basemen. Here are some aggregate stats for players when they are playing second base.

Cluster

RZR

OOZ plays per inning

Regular Second Basemen

80.4%

.0376

Regular Shortstops

81.3%

.0313

Regular Third Basemen

77.1%

.0434

Utility Infielders

81.7%

.0434

Jack of all Trades

81.8%

.0338

Everyone Else

80.1%

.0398

All told, regular shortstops are better at getting to in-zone balls at second base (when they moonlight there) than are regular second basemen, but they don’t get to as many out-of-zone balls. (Again, a bit of a small sample size warning on that one too.) All in all, it’s a bit of a wash between them. However, utility infielders have them both beat!

We see some of the same patterns among different cluster members when they are playing third base.

Cluster

RZR

OOZ plays per inning

Regular Second Basemen

73.0%

.0314

Regular Shortstops

70.8%

.0318

Regular Third Basemen

72.3%

.0339

Utility Infielders

73.4%

.0344

Jack of all Trades

70.2%

.0319

Everyone Else

70.4%

.0322

Regular shortstops actually make comparatively bad third basemen. Again, it might be the fact that the naughty shortstops are the ones who get exiled to third once in a while, so maybe we’re just dealing with the dregs of the position here, but what to make of the utility infielder? Usually, utility infielders are guys who “can handle” shortstop (and we saw that above), and who have just enough of a bat to be allowed to learn the ropes at second and third and stick around for a few more seasons. But why can’t the guy who’s an actual shortstop do the same thing?

I’d suggest that the answer is that regular shortstops moonlighting at third or second might be terrific athletes, but what they don’t have are a lot of in-game reps at those positions. He’s probably grown up always being “the shortstop.” The utility guy, on the other hand, has likely seen game action at the position and a little practice never hurt anyone.

But of course, it’s not that easy as just working hard and taking a few million groundballs. We also see that jack-of-all-trades guys, who are generally drawn from the less athletic end of the spectrum and despite being very familiar with playing a few spots, turn out to be lesser fielders, so it seems like you need to have a little of both athleticism and practice.

Why Studying Utility Infielders May Have Screwed Up the Boston Red Sox
There are a few interesting implications to this. Assuming that the utility infielder effect is true, what we have is evidence that it’s not enough to just be “good at defense.” There’s some actual specialization and practice involved. A shortstop can’t just go to second base or third base and magically outplay the normal occupant. He probably wouldn’t be horrible, but there would be some lag while he got acclimated to the position so that his true talent (if he has it) could shine through. The utility infielder has already had that acclimation time.

From a player development standpoint, I think this has a somewhat less obvious corollary. Teams often will have their players specialize at one position in anticipation of playing that spot at the big-league level. The reality is that not everyone who makes it to the big team will be a starter with a consistent position. It might actually be worth it for teams to mix and match a little more on the farm, even with the guy who looks like he’s the “shortstop of the future.” Plus, we know that a multi-positional guy actually has a decent amount of value (perhaps a few runs) which comes exclusively by virtue of his willingness to wear several gloves. And sometimes a guy has to move because a team is blessed to have two good shortstops. Might as well prepare for the possibility in the minors. Maybe even something as simple as having the fifth and sixth innings be a time when the second baseman, third baseman, and shortstop rotate through to the other two positions. That way, when he gets to the majors and is called up to fill in for the utility infielder who is filling in for the second baseman, he has some experience in actually being a utility infielder.

But back to determining a player’s value. We want to model reality as closely as possible. A shortstop who has only played shortstop and does so at an average level would actually probably put up merely average defense if he suddenly found himself at second base. At least at first. He might learn the position and take to it, but our models assume that he would put up above-average defense instantly. That’s a dangerous assumption.

The evidence suggests that a fielder needs some seasoning at the position he’s playing. At this point, we don’t know how much seasoning, but to properly value a shortstop and to give him credit for “well, he could go to third and just nail it there,” we need to ask whether he’s ever played third before. If not, we need to temper that value a bit. I don’t know that there’s an easy way to shoehorn that into WAR, but it would be a better reflection of reality. Maybe have a “previous positional diversification” factor in the calculations. You may also want to think about how quickly the player learns. It’s going to matter.

I think it’s time I brought up Hanley Ramirez’s name. When the Red Sox signed him, they made the assumption that because of his previous shortstopping experience, he would do fine if they just plunked him in left field. Well, something did go plunk. Much of the assumption about how players move around the diamond and what we might expect from them comes from studying the people who do that most regularly, utility infielders (and outfielders). Because they are the same person when playing second as they are when playing short, they make great case controls. We could see how they did at short and how well it correlated with what they did at second and create a chain of value from there. We didn’t stop to think that utility infielders weren’t a good model for guys who were going into a position cold, specifically because utility infielders had experience in roaming around the field. Turns out that one of the factors driving that chain of value was the actual experience of being a utility guy. It might not apply as well to a guy who was going into the position cold. Trying to apply it to Hanley Ramirez, who had played third before, but never the outfield, turned out to be a bad idea.

Lesson: We need to pay closer attention to our assumptions.

You need to be logged in to comment. Login or Subscribe
Richie
2/09
Outstanding article, IMO. Well-conceived, and then well-executed. Thank you.
Machaut
2/09
Yes, a thousand times yes! I always thought the assumption that "stud infielder can convert to outfield easily" is a lot like assuming a classical musician, being the most technically refined of her kind, will automatically excel at jazz. There are plenty of multitalented performers out there, but they're usually those who actually work hard at multiple disciplines. That said, I believe that Harold Reynolds will make a transition into politics seamlessly.
therealn0d
2/09
Classically trained trumpet player here.That is a great example. I was in both concert band and jazz band, first chair in both. I could handle jazz just fine, could play the leads just fine, but when it came time to get out the Fake Book and turn to page whatever...there are the changes, have at it...I was completely lost. I had spent plenty of time playing it but very little time actually learning it. I could play a cadenza just fine, because that's the language I knew.
jfranco77
2/09
Who were the 2Bs playing SS? Were they Addison Russell or Javier Baez? If so you might have the classic "good enough to play SS, but we have one of those, so moved to 2B, but then moving back to SS when our SS needs a day off" problem.
pizzacutter
2/09
I think if there's a hole in this, it might be that. I tipped the cap to it in the article, but I wish I would have had time to go deeper on this very issue.
GeorgeKimmet
2/09
Agreed, excellent article. I'd be curious to know whether the effect is the same in the outfield?
therealn0d
2/09
Just (this moment) finished reading your article in the annual (after having just read this). You just keep getting better and better Russell. Thank you for being here.
bhacking
2/09
Another example was Alex Rodriguez worked out at 1B this past year and couldn't get good enough the Yankees would even consider using him there. Remember this was one of the top SS in the game who then moved to 3B but couldn't make the switch to 1B.
therealn0d
2/09
That brings up how this wasn't controlled for age, because ARod is kinda super old to learn a new position.
Richie
2/09
Well, wasn't controlled for a lot of things, which is fine for a first look at the matter. Fine for a Web site article, as opposed to, say, something for the Annual. Fine given what has been uncovered, something illuminating and robust enough that it would take a whole lotta found counter-caveats to reverse.
lichtman
2/10
I have to say that I don't see much value in any comparison that does not use some kind of "delta method" whereby you are are comparing the same pool of players. Clearly these pools are not the same. You warn about that "possibility" but it is a certainty and I believe fatal to your analysis. Tango (and I) have done plenty of work on this (you should have cited at least SOME of that work!) subject using the delta method. That is how we come up with the positional adjustments in the first place. Well, at least one of the methods. I would bet for example, that when you compared players to themselves, you would find that regular SS did indeed have better numbers at 2B than regular 2B. Likely true at third also, although third and SS have greater differences in skill-sets than 2B and SS.
harrypav
2/10
I'll just echo and agree with MGL's note on the lack of citations. We all need to do much much better in this arena.
ericmvan
2/10
Two comments. First, positional adjustments are yet one more example of why we ought to sepaate WAR into a retrospective version, that attempts to measure the actual value of a season just completed, and a prospective version, that measures the expected value were the season to be repeated, and hence involves isolating the talent underlying the performance. For retrospective WAR, all you really need is an adjustment that is largely based on the average offense at the position in question. How long a period of years you use for smoothing depends on whether you are comparing players within or across seasons. I would argue that in a period where, for instance, SS offense is unusually high or low, a WAR designed to identify an MVP should reflect that. However, if you're trying to identify HOF worthiness, you'd want to use the long-time historical trend. (OK, so that's three different WARs. So far.) There will be some complications -- right now, for instance, LF offense is less than RF offense because there aren't enough good OFers to go around. But the proper adjustment for that can be approached empirically in several ways -- for instance, it's also true (last time I looked) that fewer LF innings are played by identifiable regulars than any other position. Prospective WAR is actually good for little more than pre-season player rankings, but of course it's based on work crucial for accurate projections (and hence comes in two versions as well, park-neutral and park-specific). Positional adjustments here are much trickier. Now you do want to identify how well players actually play other positions, the value of being a utility guy, and so forth. The second comment concerns the Red Sox Hanley disaster. The year before, the Sox had thrown Brock Holt into the OF despite zero innings of experience there at any level of baseball, and in 382 innings (mostly in RF) he put up a +28 R/150 DRS and +17 UZR. That may have made them overly optimistic about the Ramirez conversion.