keyboard_arrow_uptop

I discovered something interesting when I wrote the Transaction Analysis for the Braves’ signing of Sean Rodriguez (yep, they save me for the really important deals). Here’s the sentence that sent me down the trail toward this article:

As positional versatility grows in value in baseball—there were two players who played six or more positions in 2013, compared to 17 in 2016—Atlanta acquired one of the most versatile.

Now, I know what you’re thinking: “Dude, you wrote 600-something words about Sean Rodriguez. It’s inevitable that you’d come up with some random stuff that’s mildly interesting.”

OK, I’ll admit, some of what I wrote—like drawing the distinction between the poor man’s Ben Zobrist and the unable-to-escape-the-cycle-of-poverty man’s Ben Zobrist—doesn’t really rise to the level of sophisticated empirical analysis. But positional versatility is something that’s suddenly come into vogue. That’s interesting.

Last year Rodriguez played seven positions. In descending order, he appeared in 57 games at first, 29 at second, 27 at shortstop, 17 in right, 11 at third, 10 in left, and five in center. Let’s use that as our floor for multi-positional players: five games per position. That way we don’t give a player credit for a weird extra-innings emergency stint or two at a position[i].

I like graphs. And I like the period from 1969 to present, because in 1969:

  • Baseball expanded from 20 teams to 24
  • Divisional play and the multi-tiered postseason began
  • The pitcher’s mound was lowered
  • The strike zone expanded
  • They took those rocks off the moon, and nothing’s been the same since

So here’s a graph of players since 1969 playing five or more games at seven positions, as Rodriguez did last season:

OK, that’s useless. How about players who played five or more games at six positions:

If you’re wondering when I’m going to knock it off with the stupid graphs, the answer’s now. But first, who are the guys in the graph above? Here’s an alphabetical list of players since 1969 to have played five or more games at six or more positions:

Now, here’s my point. Playing six or seven positions is extreme. Nobody since 1969 has played catcher and five other positions, so everybody on the list above had to play five or more games at all but one of the seven non-catcher infield and outfield positions. That’s a lot. So let’s move the bar down just one notch, to five positions, five or more games each. Now we’ve got something:

I can even add a trend line to that, though it understates what’s happened the last couple years:

This divides it up between players playing exactly five, six, and seven different positions:

For many years, a player putting in time at five different positions was as rarity. Then, around the turn of the century, it became more common[ii]. That fad passed a little, but it’s become more popular than ever lately.

It isn’t hard to understand why positional versatility has become more valuable of late. With 12- and 13-man pitching staffs, there are a limited number of hitters, and therefore fielders, available on the bench.

For example, on August 13, 1975, in a game against the Red Sox, Tigers pinch-hitter deluxe Gates Brown batted for shortstop Tom Veryzer in the bottom of the eighth. He singled. Manager Ralph Houk brought in Billy Baldwin to run for him. Then, in the top of the ninth, Gene Michael came into the game to play short. On an American League team with 13 pitchers, those three players—one a pinch-hitter, one a pinch-runner, one a defensive substitute—are your entire bench! A pinch-hitter or pinch-runner who can stay in the game and play a position isn’t a luxury, it’s a necessity.

That’s why Alexi Amarista, Mike Aviles, Chris Coghlan, Nick Franklin, Marwin Gonzalez, Enrique Hernandez, Brock Holt, Michael Martinez, Hernan Perez, Jurickson Profar, and Andrew Romine all played five or more games at five positions, Shawn O’Malley played six positions, and Sean Rodriguez played seven.

How are these multi-positional players being used? Are they regulars who wind up, over the course of the season, playing all over the field? Are they late-inning fill-ins who mix and match their way to five or more games at several positions? Or are they something in between?

To answer that, I took each player since 1969 who played five or more games at five or more positions and looked up the number of plate appearances he received. That gave me 186 player seasons of data. Then, for each year, I calculated the average plate appearances. (And if you’re saying, “mean or median?” the answer is mean, though I calculated both, and they were pretty close.) This should indicate not only how multi-positional players are deployed but also whether that deployment has changed over time:

That’s a little noisy. I’m going to remove the data points for seasons in which only one or two players qualified, since in those years, one player can have a huge influence:

The takeaway here, I think, is that generally speaking, players who play five or more positions typically get 250-420 plate appearances per season. Of the 30 data points on the graph, only four are less than 250 and only four are greater than 420. So let’s say multi-positional players are, by and large, part-time players—not regulars, but not rarely-used substitutes. Of the 186 player seasons denoted by the red bars, only 27 qualified for the batting title (the most recent being Gonzalez in 2016, Rodriguez in 2015, and Josh Harrison and Zobrist in 2014), but only 11 had fewer than 150 plate appearances (Martinez in 2016 and Elian Herrera in 2014 are the only ones in the past decade) [iii].

Does this mean that, Moneyball-style, there’s a market inefficiency for multiple-position players? Well, they’re certainly not paid a lot! Of the 13 players who played five or more games at five or more positions last year, four started the year in the minors. The other nine earned a mean of $1.70 million and a median of $1.35 million. If we assume the other four earned the major-league minimum, the mean drops to $1.31 million and the median to $606,000. Same in 2015: Mean $1.49 million, median $1.29 million, or if assuming minimum salary for players with no listed salary, mean $1.38 million, median $1.08 million.

But a market inefficiency assumes that the market is underpaying for a valuable asset. How valuable are the multi-positional players? The 13 who played five or more games at five or more positions contributed, in total, 2.4 WARP in 2016. Over 3,519 plate appearances. Six were sub-replacement level. Technically speaking, that’s not good.

And things weren’t much better in 2015: 6.8 WARP in 3,139 plate appearances, four of nine players sub-replacement. Frame of reference: 6.8 WARP in 3,139 plate appearances is like 1.1 WARP in 547 plate appearances, which is what Adeiny Hechavarria generated this past year.

However, maybe that’s not the right measure of the value of a multi-positional player. Circling back to Rodriguez, his versatility was rewarded by the Braves. He received a $1.5 million signing bonus and will be paid $5 million in each of 2017 and 2018, more than any multi-positional player since Zobrist in 2014. Maybe the Braves aren’t paying him just to platoon with Jace Peterson at second base, but also to provide breathers to Freddie Freeman, Dansby Swanson, Adonis Gracia, Matt Kemp, Ender Inciarte, and Nick Markakis, enabling all of them to be more productive. In that sense, perhaps a player who can play multiple positions is not just a handy guy who contributes a little bit of WARP, but also somebody who can help everyone around him do better.

There might be something to that. The players who played five or more games at five or more positions played on teams that averaged 84 wins in 2016, including World Series opponents Cleveland (Martinez) and Chicago (Coghlan) as well as divisional champions Boston (Holt), Los Angeles (Hernandez), and Texas (Profar). In 2015, teams with a multi-positional player won an average of 83 games, though that average drops to a .500, 81-win season excluding Rodriguez and the 98-win Pirates.

Maybe players who can play all over the infield and outfield provide outsized benefits to teams by allowing all players to be better rested. Or maybe they’re becoming a necessity in an area of constrained benches. Either way, the growth we’ve seen in recent years seems likely to continue as long as pitchers comprise around half of major-league rosters.

Thanks to Meg Rowley and Patrick Dubuque for research assistance/Charles Gipson de-mythologization.



[i] I once played in a fantasy league where players qualified for a position if they played there one game. In 2002, Albert Pujols hit .314/.394/.561 with 34 homers and 127 RBI. And on August 2 that year, he played the last two innings of a loss to the Braves at shortstop. Man, something like that could make your season in that kind of league.

[ii] What was going on at the turn of the millennium? I don’t really know. It could be that the preponderance of old sluggers—there were an average of over five players per year from 2000-2004 with 30 or more home runs who were 35 or older; the average over the past three seasons is two—and the need to carry legs for them. Or it could be that teams were just starting to adjust to growing reliever corps. Or maybe it’s just coincidence that Jolbert Cabrera, Miguel Cairo, Joe McEwing, Melvin Mora, Desi Relaford, Donnie Sadler, who combined for 17 seasons of five or more positions played from 2000 to 2004, were all born within 40 months of one another.

[iii] The outlier is Charles Gipson of the 2000-2001 Mariners. Over those two seasons, he appeared in 55 games in left field, 40 games in right, 22 in center, 14 at third, 11 at shortstop, and one at second—that’s 143 games—but had only 105 plate appearances, total. The 2000 Mariners won 91 games with only three regulars under the age of 31 and the 2001 team won 116 with only four regulars younger than 32, so there were ample opportunities for Gipson to come into the last innings of games that the Mariners were leading to take over a position from an old guy. In 2000 he entered 42 of his 59 games in the eighth inning or later, and in 2001 the count was 61 of his 94 games. He’s the only player since 1969 to play five or more games at five or more positions in a season without getting 100 plate appearances, and he did it twice.

You need to be logged in to comment. Login or Subscribe
CrashD
12/20
I like graphs, too, particularly graphs that show you something you miss on a spread sheet. Keep on graphing!
mainsr
12/20
Oh, I will! I think this means we're visual learners or something.
ErikBFlom
12/20
What about playing many positions early in the history of baseball? Honus Wagner supposedly could play every position, but in the dead ball era, shortstop was by far the place to put him.
mainsr
12/20
Yes, that's a good point. I was thinking of going back before 1969 but it's a pretty laborious project, in that you have to do 35 play index searches per season to identify every player who played at least five positions (since there are 35 possible combinations). And, of course, there's the issue of incomplete data, particularly in the pre-Retrosheet era. Just randomly, I checked 1916. There were two players that year who played five or more games at five positions: Harry Heilmann (34789) and Howie Shanks (35679). I'm guessing that was pretty much the norm, given that lineups back then were amazingly durable (Heilmann's team, the Tigers, had players who started 141 or more games at each of five positions) and benches were large enough that teams could carry, easily, a backup infielder who wouldn't play the outfield and a backup outfielder who wouldn't play the infield. Wagner, based on the data we have, played 1,887 games at short, 374 in the outfield (no breakdown between positions available), 248 at first, 210 at third, and 57 at second. He didn't play shortstop at all until 1901, his Age 27 season and fifth year in the majors, and didn't become a regular there until 1903.
myshkin
12/20
Hmm, what are the 35 combinations? If you are setting aside P and C, then I count 1 7-position, 7 6-position, and 21 5-position combinations. Also, you can probably just combine the seasons and then collate the results elsewhere, e.g. there are only seven 34579 seasons between 1901 and 1969. The incomplete data starts to become a real issue anywhere earlier than the 30s, though. Non-linear methods of player and roster valuation are the way to go here. A simpler, but similar, question is how much extra value does a platoon bat on the bench (or having a L-R platoon for one spot) add over the alternatives.
mainsr
12/20
Oops, sorry, 35's a typo; I meant 56. (If you saw my spreadsheet, you'd know how I'd get them mixed up.) But that includes C. (And there were, in fact, two occurrences since 1969: Phil Clark in 1993 and Phil Nevin in 1999 both played 5+ games at catcher and the four corners. If you look just at positions 3-9, yes, there are 21 possible five-position combinations. Re platooning, agreed, that confers a benefit, though not the (not easily proved) benefit of giving regular position players games or innings off by plugging in a Brock Holt or a Sean Rodriguez all over the place. Incidentally, your finding that there were seven 34579 positions between 1901 and 1969 is wild--there have been seven in the last five years!
phealy48
12/20
I appreciate the subtle Moneyball-amazon link.