Kenny Peoples-Walls And How Shortstop Fielding Develops

A little over a month ago, I wrote this piece, examining how minor league catchers cut down on their passed ball rates over time. Indeed, anyone who peruses Baseball-Reference minor league catcher pages probably notices the tendency for teenage backstops to let pitches by them at alarming rates, only to ultimately settle into a more acceptable range as they reach their mid-twenties. On at least an anecdotal level, one can observe a similar phenomenon with left-side infielders–third base and shortstop are the most error-prone positions, and these tendencies can be blown up by inexperience. The worst qualified fielding percentage by an MLB shortstop in 2013 was Jed Lowrie‘s .962, and the worst from a qualified third baseman was Pablo Sandoval’s .940, but in, say, the Low-A South Atlantic League, ten of the fourteen third basemen who got over fifty games at the spot fell below that .940 fielding mark, and twelve of the fifteen shortstops didn’t break .962. And the short-season levels are a level of magnitude worse than that.

As with catchers, then, we can assume that left-side infielders cut down on their error totals significantly as time goes on. In this piece, I’m going to examine the development of shortstops in this area.

Since it’s Cardinals Day here at FanGraphs, let’s take St. Louis shortstop prospect Kenny Peoples-Walls as an example. Peoples-Walls was picked in the fourth round of the 2011 draft before he turned 18, and being very young for a high school draftee, it took him a while to adjust to pro ball. He spent his first two seasons in the Rookie-level Gulf Coast League, putting up a .239/.313/.250 line in 25 games in his age-17 season and improving to .260/.317/.367 as an 18-year-old. Finally, he was moved up to the Advanced-Rookie affiliate in Johnson City for this past sesaon and continued to improve his batting line, hitting .300/.352/.468. It’s a bit disconcerting to see a player spent three straight years in Rookie ball, but Peoples-Walls is also a shortstop who hit .300 with some power as a 19-year-old–you don’t just write those guys off. He has some refinement to do at the plate–a 73/15 K/BB in 2013 isn’t going to get it done–but he’s a teenager with something to build on.

Unfortunately, Peoples-Walls, like a lot of Appalachian League players, is prone to, well, this:

The quality of defensive play in the league is such that, when I see a play like this, my reaction isn’t “Man, what an awful throw,” it’s “That was a good play; I just wished he finished it.”

Here’s the thing about this particular error, though: the biggest mistake on the play is mental. As you can (somewhat) see in this .gif, the runner, Pulaski first baseman Kristian Brito, has true 2-grade speed. Peoples-Walls’ throw beat Brito to first base (or, I suppose, the very-loosely-defined general vicinity of first base) by about three steps. There was no need for Peoples-Walls to throw the ball so quickly and off-balance–he still could have set and thrown and the ball would’ve beaten Brito anyway.

It is this lack of situational awareness that tends to be the biggest plague on low-minors fielders. Often, one sees, the reverse situation play out, actually–a player tries to make a heroic throw from shortstop or third base when there’s obviously no play (sometimes, rather comically, as a result of him mishandling the ball in the first place, thus losing the opportunity to beat the runner), resulting in a rushed, bad throw like the one above, except in a situation where the fielder had nothing to gain by throwing the ball.

Of course, there’s also a higher instance of more garden-variety miscues, like this one, as well:

One might be tempted to look at Kenny Peoples-Walls‘ Baseball-Reference page and say that he’ll never eradicate his error problems. Unlike his consistently improving triple-slash lines, his fielding percentages have remained quite static in his career. He fielded .917 as a 17-year-old, .918 as an 18-year-old, and .916 as a 19-year-old–it doesn’t get much more static than that.

But how does he compare to the typical teenage pro shortstop? And how do we expect teenage pro shortstops to develop?

Before I answer those questions, I should probably address another one, one that looms large in any discussion of stuff like this: Why should we care about fielding percentage? Sabermetric theory has thoroughly debunked the traditional model of fielding percentage as the most important defensive statistic–range is paramount, right?

I certainly agree with this notion, but there are a couple of reasons this is important:

1.) We don’t have advanced defensive statistics for the minors–fielding percentage and range factor are about as advanced as they get. Of course, I wish we had better data, but let’s do something with what we have.
2.) This is the correlation of fielding percentage to UZR/150 for all qualified MLB shortstop seasons since 2002:

A .22 r^2–a player’s error rate explains about 22% of his fielding ability as measured by UZR/150 (which of course has its own flaws, but that’s neither here nor there). That makes sense–fielding percentage might not be anywhere a complete measure of a fielder’s ability, but at a position like shortstop, where errors are fairly common even in the big leagues, the difference between the highest and lowest-error players can add up quite a bit over the course of the season. Twenty-five of the top 27 fielding percentage seasons resulted in positive UZR/150s, and 21 of the 23 worst resulted in negative UZR/150s. It may not be the whole story or anything near it, but it’s not an insignificant portion of it.

Notice, of course, that the lowest fielding percentage of this entire twelve-year span is .947. The average is .975. According to the regression equation in the graph, a shortstop needs to field around .972 to project for a UZR/150 of zero.

Obviously, Kenny Peoples-Walls and most of his low-minors cohorts aren’t at .947, let alone the .970s. Should we expect them to get there?

I wanted to map out the fielding percentage development of shortstops from age 18 to age 28, so I went through Baseball-Reference and found all shortstops who had at least 50 games at the position as 18-year-olds in US pro ball from 1989 to 2003–a nice fifteen-year period–and then tracked their fielding percentages at shortstop for every season they played at least 50 games there through age 28.

Unfortunately, this didn’t yield as robust of a sample as I wanted it to. From 1989 to 2003, there were only 112 players who got in 50 stateside games at short as 18-year-olds. That’s enough for a good sample, but as you’d expect, not all of these players stuck around through age 28 (50 did), and even of those that did, not all of them held the position. In fact, only sixteen (14.3%) played 50 or more games at shortstop as 28-year-olds–a combination of injury attrition, moving to other positions, or manning a backup/utility role that didn’t involve enough games to qualify for the sample.

As such, the above chart is a little misleading, in that there are 112 players comprising the age-18 data point, but only 77 remain at age 19, 63 at 20, 53 at 21, 43 at 22, 42 at 23, 36 at 24, 23 at 25, 21 at 26, 20 at 27, and 16 at 28. Still, it gives us a starting point for understanding how shortstop fielding percentages evolve–they start out in the .920-.930 range at age 18 and steadily rise up past .950 by age 22 before climbing slowly up to the .970 range by age 25. That matches the traditional conception of position player development in general–we expect rapid growth until the early twenties and then slow growth into a player’s prime around age 26-28. It’s possible that error rate, constituting the elimination of bad plays rather than the increased ability to make good ones, could continue to improve past a player’s “prime” as he gets veteran know-how and all that sort of thing, but that’s a topic for another day.

In any case, this graph, flawed as the methodology may be, strongly points toward the age 18-22 window as the key time for improvement. If a player isn’t in the mid-.900s in fielding percentage at 22, he’s clearly well behind the curve. Let’s focus in a bit more on this by examining the fielding percentages of just the 43 players who were still manning the shortstop position at 22:

Weeding out the players who were done at short by age 21, we can notice a few things. First off, age-18 fielding percentage didn’t really change much–the overall average was .924, but here it’s .926. It’s not as if those who stuck at the position for awhile were all fielding .940 or .950 from the start of their careers. However, we do see a pronounced jump in the age-20 fielding percentages here (.944 overall, .951 for these 43). Further, the fielding percentages of this group improved by .25 from age 18 to age 20, but just .04 from 20 to 22, suggesting that pre-age-20 cutdowns in miscues are vital.

We can also look at this phenomenon in terms of what the lowest fielding percentage for each age was:

18: .885 (Pokey Reese)
19: .882 (Pepe Frias)
20: .890 (Ramon Hernandez…no, not that Ramon Hernandez)
21: .893 (Steve Lackey)
22: .922 (Eddy Martinez)
23: .910 (Ricky Magdaleno)
24: .922 (Enrique Cruz)
25: .940 (Damian Jackson)
26: .905 (Tomas de la Rosa)
27: .949 (Danny Klassen)
28: .954 (Edgar Renteria)

There are a couple of outlier seasons in there–de la Rosa’s age-26 season was way out of line with his career numbers and .038 worse than the next-worst age-26 campaign, and even at 20 and 21, Hernandez and Lackey were at least .02 worse than everyone else in that age group. If we throw those out, we can see that sub-.900 percentages may be tolerated for teenagers, but by the time the twenties kick in, .910 or so is the basic “floor,” jumping up to the .920s at 22 and the .940s at 25, rising up to the familar .950 or so for players in or near their primes. Players teetering on the edge of these marks are obvious position-switch candidates. On the other hand, here are the highest marks:

18: .966 (Manny Amador)
19: .973 (Edgar Renteria)
20: .981 (Cesar Izturis)
21: .983 (Rob Valido)
22: .979 (Jimmy Rollins)
23: .980 (Rollins)
24: .986 (Alex Rodriguez)
25: .986 (Rollins)
26: .987 (Rodriguez)
27: .989 (Rodriguez)
28: .985 (Alex Gonzalez)

Oddly, Amador was never a regular shortstop again. Right behind him at age 18 were Jose Reyes (.964) and Rollins (.960), though, and the rest of these players were shortstops, and good ones, for quite some time. Oddly, Renteria had the highest fielding percentage of 19-year-olds and the worst of 28-year-olds, though.

So through these different means, we can start to get a handle on how to treat teenage fielding percentages. Above .940 is well ahead of the curve and below .900 is well below it. Ostensibly, we can see that most of the low performers didn’t work out and most of the high ones did–if they could hit, anyway. But what if we test this idea on a larger scale?

This is a graph of age-18 fielding percentage (x-axis) vs. age-24 fielding percentage for the 36 players in the sample who played 50+ games at 24. As you can see, there is a correlation, but it’s not a particularly strong one, and there are some odd data points–we have Pokey Reese going from .885 to .965 on one hand, and Aaron Holbert going from .951 to .934 on the other.

Other age groups don’t really fare any better. Here’s age-18 fielding percentage vs. age-20 fielding percentage:

And 22:

And, for comic effect only, 28:

So, while we do clearly see a marked increase in fielding percentages of shortstops as they mature, exactly what point they start from doesn’t really dictate where they end up. It’s not like a .900 fielder at 18 is destined to top out at .960, whereas a .925 fielder at 18 is destined for .980.

Here’s something that’s a little more helpful:

This is age-20 fielding percentage vs. age-24 fielding percentage, and it’s over twice as helpful as the age-18 marks were. This is due to a couple of reasons: first, age-20 players are more developed, and second, they’re more often in full-season ball, meaning that most of their marks were in samples far larger than 50 games, whereas many of the 18-year-olds in the sample were just over the threshold. There are still some data oddities, mind you–Magdaleno shot up from .917 to .967, while Gookie Dawkins fell from .959 to .951 and Holbert fell from .947 to .934–but we still get a good sense of how shortstops can project from age 20 to age 24–they’re all expected to reach .940 or so, “average” moves from .950 or so to the low .960s, and players who are fielding in the .980 range can be expected to more or less maintain that but not improve.

So what does all of this tell us? Nothing’s set in stone, of course, and player development is fluid, as anyone who studies prospects knows. There are players who beat the odds the numbers lay out for them, and there are near-certain stars who fail to work out at all. But with those obvious caveats, there are a few things to be gleaned here:

1) If a shortstop prospect is fielding above .920 as a teenager, above .940 at 20-21, above, .950 at 22-23, and above .960 from there on out, he’s basically on track to become a solid or better fielder in terms of error avoidance.

2) As the first point indicates, shortstops can be expected to reduce their error totals fairly dramatically from age 18 to age 20 and progressively less dramatically from there on, but still improving into their prime years.

3) Where exactly a shortstop’s fielding percentage lies at age 18 isn’t particularly important. Having a really solid one at a young age is a good sign, but other than that, it doesn’t mean much. Miguel Cabrera fielded .928 at short as an 18-year-old, while Alex Gonzalez (the good one) fielded .925. Cabrera went on to become a stiff first baseman in his mid-twenties, whereas Gonzalez played a quality short well into his mid-thirties. Damian Jackson fielded a solid .933 as an 18-year-old, but couldn’t ever top .955 as a shortstop in eight more years at the position. Pokey Reese, who was the worst fielding 18-year-old at .885, ended up becoming a solid defender at both up-the-middle positions. Young kids usually make a lot of errors–standing out by making only some is a big positive, but beyond that, exactly how many the player makes doesn’t really determine whether he’ll make it at the position long term. And thus…

4) The most important thing, it seems, is for players to follow the improvement curve. To bring this back to Peoples-Walls, his .918 fielding percentage as an 18-year-old was slightly below the average of 18-year-olds in the sample (note, it was just 42 games, so it’s quite unreliable), but it’s not far off, and as I just said, the 18-year-old fielding percentage doesn’t seem to be a huge deal. By staying static at age 19 with a .916 mark, though (just 47 games, so still unreliable, I know), he’s now gone from .06 behind the average to .18, or, to put it another way, from 64th of 113 to 64th of 77. If he doesn’t improve at age 20, he’d be somewhere around 60th of 64. Players can take unusual routes to solid performance, and they can start from various degrees of rawness, but the key is improvement, and it appears that that improvement has to start fairly early.

With these in mind, there is obviously a lot more research that can potentially be done on this area (and on the other infield positions in similar fashion). Don’t make the mistake of dismissing teenage shortstops with .900 fielding percentages out of hand–their error totals may not be as singularly awful as you think. On the other hand, there is certainly a large number of instances where a player fails to improve to an MLB-caliber level at the position. These numbers seem to point to the key not being the innate starting point of a player, but how he’s able to develop over time, an important thing to keep in the back of our minds as we consider all aspects of prospects, not just the error rates of shortstops.