The Hatteberg Effect

Alex Rodriguez isn't the only player to struggle in his initial appearance at first base. (via Keith Allison)

Alex Rodriguez isn’t the only player to struggle in his initial appearance at first base. (via Keith Allison)

Billy Beane: It’s not that hard, Scott. Tell him, Wash.
Ron Washington: It’s incredibly hard.

On April 11, 2015, Alex Rodriguez played first base for the first time in his professional career. It didn’t go well. Rodriguez dropped the first throw he saw from the left side of the infield. Later, his foot came off the bag reaching for a wide throw from third. That’s two errors on throws to first, both on plays major league first basemen usually make.

This is a fielder who was once a Gold Glove-caliber shortstop. Granted, age had taken a pretty serious toll on that once-golden glove, and by this point A-Rod was almost exclusively a DH, but it’s not like taking throws at first requires any of the speed or range that had abandoned him.

We like to think anyone can play first base. It’s where teams stick their DHs when they have to play the field, or where they move aging and injury-riddled veterans who can no longer cover the ground needed at their old positions. It’s where a nearly-49-year-old Julio Franco trotted out on a major league baseball field, and a well-past-50 Franco kept on trotting out in the minors and independent leagues for years.

There’s a reason first base is all the way at the end of the defensive spectrum. After all, there is not much a first baseman has to do that other infielders don’t also have to do better, since being so close to first means they rarely have to make difficult throws and generally have more leeway to bobble or knock down batted balls and still have time to beat the runner to the bag. All things considered, first base is probably the position where a poor fielder can do the least damage and the position where a good fielder will have the most trouble making an impact.

The only thing a first baseman has to regularly do that other fielders don’t is catch throws to first, and that is one of the simpler tasks executed on a baseball diamond. Watching the replays from that A-Rod game, though, you start to hear Washington’s voice (or rather, Brent Jennings’ voice portraying Washington) echoing in the back of your mind, and you begin to wonder: Is it really such a simple task?

We know where Washington stands, and probably where A-Rod stands. Most likely, they’re right that it actually is pretty hard, or at least a lot harder than it looks. How much harder, though? Or, more precisely, how much difference does it actually make when you stick someone without any experience at first?

To test this, I looked at players since 2000 who have appeared at first base with limited experience at the position and whether there were more errors on throws from second, third, or short with these players at the bag than with other first basemen.

Part-time 1B (≤10 Games in a Season)

The first group of players I looked at consists of players who appeared in at least 100 defensive games in a given season, with no more than ten of those coming at first. This ensures each player is a regular at another position (or multiple positions) who slid into the first base role only as a secondary option, at best.

This gives us a good starting point by singling out players who are committed to playing somewhere other than first. That means they not only have limited game experience at the position but also likely don’t get much training at first-base-specific skills like receiving throws since they will be spending their time working on their primary position instead. (Some players from other positions, like Matt Holliday this past offseason, do work out at first base to expand their versatility, but it is still generally much less experience than regular first basemen get.)

One interesting facet of this sample is that the players we’re looking at are virtually guaranteed, at least as a group, to be better fielders than the average first baseman because they all regularly play positions further left on the defensive spectrum. If this group of players does allow more throwing errors, we can be confident it’s not because they are simply worse fielders. (It’s not a given that the ability to save throwing errors is necessarily correlated with overall fielding ability at the major league level, but it’s nice to be able to rule that out as a possible explanation.)

There were 257 player-seasons since 2000 that qualified for this sample, covering approximately 2691 throws to first from either second base, third base, or shortstop. The number of throws isn’t exact because it’s not always easy to infer from the play-by-play data when there was a throw to first. For example, Retrosheet doesn’t explicitly list where someone was throwing on an error, and it is hard to tell whether there was a throw on an infield hit when there is no putout or error to indicate a throw.

I used the following guidelines to estimate which plays involved throws to first from the data at hand, but there are undoubtedly plays I missed or misclassified (for those familiar with SQL, I’ve attached the code I used at the end of this article so you can see the full details):

  • The play is a ground ball fielded by either the second baseman, third baseman, or shortstop.
  • The first baseman is credited with either a putout or a dropped ball error, or another infielder is credited with a throwing error (excluding throwing errors in double play situations unless the error comes after the first putout).
  • The play is not scored as a hit or a fielder’s choice.

As mentioned above, this gives us 2691 plays where we can be pretty confident a throw to first occurred while the players in our sample were playing first base. A total of 52 of those plays resulted in throwing or dropped ball errors, or 1.93 percent.

A Hardball Times Update
Goodbye for now.

We can then look at all the infielders who were involved in those plays and see what that same mix of infielders did throwing to first basemen who aren’t in our sample. When we do that, we find their throws resulted in errors 1.49 percent of the time. That’s roughly a 30 percent increase in throwing errors with our sample of players at first.

There are approximately 6.4 throws to first in an average game, so that translates to about 0.029 extra errors per game. Assuming a swing of ~0.8 runs per error, that’s about 0.023 runs per game , or the equivalent of about 3.7 runs per year.

Limited Experience at 1B (≤10 Games in a Season, ≤50 Games Past Experience)

One shortcoming of this sample is that, while each of the players has limited experience at first base in the season we are looking at, some of them may also have past experience at first. A few are even former full-time first basemen, like Miguel Cabrera the year he moved back to third after the Prince Fielder signing, or Kevin Youkilis, who spent much of his career shifting between first and third depending on who else Boston had in their lineup.

Then there are players like Casey Blake, who was never a full-time first baseman but who made at least one appearance at first every year from 2000 through 2011. (In fact, Blake so frequently played the role of part-time first baseman that he had five separate seasons qualify for our sample.)

So this isn’t a perfect sample to explore what happens when you put just anyone at first. The majority of players in the sample do have considerably less experience receiving throws at the bag than your average first baseman, but players like Cabrera and Youkilis probably don’t tell us much about what we are really looking for. To better isolate players with limited experience, we can pare the group down a bit further.

I repeated the above test but this time eliminated anyone from the sample who came into the season with more than 50 career games at first base. This cut players like Cabrera, Youkilis, and Blake in the latter part of his career, while leaving room for players like Jim Edmonds, who occasionally filled in at first to get into the lineup with a nagging injury but was never anything but a full-time outfielder (the full list can be found here).

Removing these players from the sample bumps the error rate up to 2.06 percent over 2133 throws, or the equivalent of about 4.7 runs per season worse than the average first baseman.

LIMITED EXPERIENCE AT 1B
Time Spent at 1B Throws to 1B Error Rate Baseline Extra Errors / 1000 Throws Run Value Prorated to 162 G
≤10 G in a season 2691 1.93% 1.49% 4.4 -3.7
≤10 G in a season, ≤50 G previous experience 2133 2.06% 1.49% 5.7 -4.7

No Experience at 1B

The first group we looked at focused on players who were committed to playing a position besides first in a given season. We can also come at this from a career angle—in other words, players who have firmly established careers at another position who are playing first base for the first time. I defined this group as players with at least 500 career games in the field before the first season in which they appeared at first base.

This sample includes the Rodriguez performance described above. It also includes a couple players who immediately became full-time first basemen after years of never playing there (including, once again, Cabrera—the other is Nomar Garciaparra in 2006), as well as a handful who transitioned into a bench role where they mostly played first base. However, most of them, like A-Rod, spent only a few games filling in at first, and none had any past experience in the majors at first base. The full list of 79 players in this sample can be found here.

In order to focus on games where inexperience was most likely a factor and to prevent a small number of players from dominating the sample (Cabrera alone has as many games in his first season at first base as roughly 3/4 of the rest of the sample combined), I limited my data to the first 10 games each player appeared at first base in that first season.

This group allowed errors on 2.32 percent of throws to first. The same mix of infielders throwing to other first basemen resulted in errors on 1.66 percent of throws. This translates to a 39 percent increase in throwing errors, worth the equivalent of about 5.4 runs per year.

NO EXPERIENCE AT 1B
Time Spent at 1B Throws to 1B Error Rate Baseline Extra Errors / 1000 Throws Run Value Prorated to 162 G
≤10 G in a season 2691 1.93% 1.49% 4.4 -3.7
≤10 G in a season, ≤50 G previous experience 2133 2.06% 1.49% 5.7 -4.7
No Past Experience 1681 2.32% 1.66% 6.6 -5.4

To give these results some context, I’ve graphed the number of throwing errors per 1000 throws (approximately the number of throws a first baseman will see in a full season) for all first basemen with at least 100 games at the position since 2000. The blue line represents the average for all first basemen, while the other three lines represent the averages for each of the samples listed above. The sample averages have been shifted left or right so that their baseline falls on the league average of 1.57 percent to make direct comparisons possible.

throwing_errors_to_1B

As you can see, the vast majority of regular first basemen allow fewer throwing errors than those with little to no experience at the position. (As a quick note, I didn’t adjust the error rates for individual first basemen to account for the infielders throwing to them, so some of the more extreme error rates could be due to having infielders who are more or less wild than normal.)

One last thing I decided to check was whether catchers fare any better at the transition to receiving throws at first than fielders from other positions. The mechanics of receiving pitches are vastly different from those of a first baseman, but catchers do have a great deal of experience reading and reacting to wild throws, especially those in the dirt. It’s possible this experience transfers at least somewhat to first base.

To check this, I went back to the sample of players with at least 100 defensive games in a season, with no more than 10 of them at first base and no more than 50 games of past experience at the cold corner. I then restricted this to players with at least 100 games at catcher (either in that season or in past seasons). At this point, I’m splitting up the sample too much to really learn anything (there were only 345 throws to first in this entire sample), but the limited results don’t give any indication that catchers do better than their peers. They allowed errors on 2.32 percent of throws against a baseline of 1.37 percent for the same infielders to other first basemen.

A-Rod can feel at least a little vindicated that he’s not the only one to struggle with the transition to first base, though since he is taking grounders at first base again, he might not have much time to celebrate. Players with little and especially no experience at the position do indeed perform noticeably worse receiving throws at first than regular first basemen.

This makes perfect sense. Things like picking throws out of the dirt, positioning yourself to handle difficult hops, and recognizing when you need to leave the bag early or go for a swipe tag involve a lot of split-second decision making and reacting. None of that is easy. Like everything professional ballplayers do, they make it look easy, or at least much easier than it actually is, but major league first basemen work very hard to make those plays routine. Of course we wouldn’t expect someone who hasn’t taken those same countless reps to be equally good at it.

The only question was how much better the regular first basemen are and how much it costs teams to put someone unfamiliar there instead. If a team is considering playing someone out of position at first, does the increase in throwing errors offset other potential gains, or is it negligible compared to other considerations?

Throwing errors are rare enough that the effect isn’t huge, but it is definitely there. The increase in throwing errors we saw for players with no past experience at first base is equal to about a 0.010 drop in wOBA. If we typically think of a really good fielder being about 10 runs better than average at his position and a really poor one being about 10 runs worse, then this is about half of that.

It’s far from the only things teams should consider in deciding whether to put someone at first (after all, Billy Beane’s A’s did win 103 games with Hatteberg at first base), but it also probably shouldn’t be ignored.

Disclaimer on title: Hatteberg doesn’t actually qualify for any of the samples in this article because he had only 370 games on defense when he moved to first base, and the A’s didn’t actually allow more throwing errors with Hatteberg at first than you’d expect, so maybe he’s not quite the best guy to name this after. Oh well, it has a nice ring to it.

References & Resources


Adam Dorhauer grew up a third-generation Cardinals fan in Missouri, and now lives in Ohio. His writing on baseball focuses on the history of the game, as well as statistical concepts as they apply to baseball. Visit his website, 3-D Baseball.
17 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments
Rally
7 years ago

Good work. I wonder if, among the inexperienced 1B, there is a difference between lefty and righty throwers. I’ve primarily played shortstop or outfield in my baseball/softball career, but as I’m old I’ve moved to first base. It is so much easier to get your footwork right if you are a lefty thrower, so I’ve actually started using a lefty glove when I play there. I still throw righty when I play anywhere else.

Paul G.
7 years ago

I do remember the Yankees trying to move Gary Sheffield to first after he came off the disabled list and he had nowhere else to play. From my observations, he was genuinely terrible.

Is the conclusion here is players are moved to first less because it is “easy” and more because they will do less damage?

Adam Dorhauer
7 years ago
Reply to  Paul G.

I think in general, teams care more about how much practical impact a decision like where to play someone in the field will have than how easy or hard it is. To some extent the ideas are probably related, but thinking about the actual impact is a much more clearly defined question that has more practical value.

For example, asking someone to play out of position at catcher and asking someone to play out of position at first base are both probably hard, but the former is extremely rare because there is a huge impact to having a catcher who doesn’t know what he’s doing behind the plate since he’ll be involved in every pitch. Or, moving a poor defensive second baseman to shortstop might be an easier transition than moving him to first base, but could still have a more negative impact, so you don’t see it as much.

Spencer Jones
7 years ago

I traditionally play shortstop and center field, hurt my arm this spring and had to play first, figured it’d be a piece of cake given where it is on the defensive spectrum, it’s way harder than it looks. Having to hold runners and then get off the bag and get into a good fielding position, picking throws out of the dirt, finding the bag on throws from other infielders, figuring out when to get the ball yourself or go to the bag and let the pitcher or second baseman take it etc. There’s a lot to it.

LHPSU
7 years ago

I would say that it’s the least demanding position physically, but it’s a more mentally demanding position with more decisions to make than any other position other than catcher.

Played7positions@D1
7 years ago
Reply to  LHPSU

This is a bananas statement. More decisions to make? There is literally one play where you have to think at all — the slow chopper to the right side where you have to decide to take it yourself or flip to the pitcher. Everything else is go to the bag and catch the ball. There’s a reason why so many players “downshift” defensively to 1st base, it takes the least time to learn and is the easiest to adjust to.

John C
7 years ago

Watching Hanley Ramirez at first base this season for the Red Sox many times, it seems like picking throws is something he’s definitely learning on the fly. I’ve seen him make very nice plays, but occasionally misjudge a bounce throw across the diamond and cause Xander or Shaw to get an error. He’s done well on ground balls and line drives, probably because of his experience as a shortstop, but has some problems on pop fouls behind him. All in all, a work in progress, but certainly not the worst first baseman in baseball.

His being there in the first place is obviously a response to his defensive skills and the needs of the team. Hanley played shortstop for years; the Dodgers could get away with that because they have a pitching staff full of strikeout and fly-ball pitchers. (Same reason Davey Johnson used to play Howard Johnson at short during the 1980’s Mets heyday.) Hanley was spectacularly bad in left field and hurt himself in the process, and the Sox needed a first baseman, so he’s it. Most guys who get moved to first end up there at least in part because they can’t do but so much damage there.

MGL
7 years ago

Very nice work. Would love to see the same analysis with UZR or DRS range. FWIW, I found that the only position where the peak age for defense is in the 30’s is 1B, suggesting that there is indeed a long learning curve associated with it. The other positions have a peak defensive age of early to mid 20’s, suggesting it’s all about speed and agility (and willingness to abuse your body) with not much of a learning curve.

Adam Dorhauer
7 years ago
Reply to  MGL

That’s fascinating that 1B defense could have a peak age that late. I would have never guessed that.

Carl
7 years ago

Am I correct that your analysis (great methodology by the way) shows that on average a bad defensive first baseman is worth approx. -.5 fWAR per year, and if so a good hitting but bad fielding first baseman (i.e. Jason Giambi) is therefore worth it over a better fielding but worse hitting (if more than .5 fWAR per year) first baseman?

Adam Dorhauer
7 years ago
Reply to  Carl

This is only looking at preventing throwing errors, so overall (considering range, fielding grounders, etc) a bad defensive first baseman could easily cost more than .5 fWAR compared to an average defensive first baseman. But yes, a good-hitting but poor-fielding first baseman could certainly be a better option than a weak-hitting but good-fielding alternative. You would definitely want to look at a player’s overall contributions, offense and defense, together to see what their total impact is.

Frank Firke
7 years ago

Does the effect appear to be fairly constant, or is there evidence that some new 1B are worse than others?

Also, did you compute / could you present confidence intervals on your error rates? It’s not clear from eyeballing how much of the difference between experience levels could/should be attributed to randomness, given how small the rates are.

Adam Dorhauer
7 years ago
Reply to  Frank Firke

This methodology doesn’t lend itself that well to looking for differences between individual first basemen because the samples are so small. Some of the players included in the samples have only a handful of innings at first base, and they’re all limited to 10 games or fewer (with the exception of players who had multiple seasons in the first two samples).

The data is basically a bunch of players with zero throwing errors in a small number of throws, and then occasionally someone with one or possibly a few throwing errors, so there will be a lot of variance between individual players, but most of that will just be random noise. It is likely some new first basemen are better at this skill than others, but it would be near-impossible to tell which is which using only the recorded number of throwing errors that occur in such a limited number of games for each player. You might be able to get a better idea using scouting data in conjunction with the play by play data.

The effect for each of the samples by itself is roughly 2 standard deviations higher than average. Here are the random SDs for each of the samples based on the number of throws and the baseline error rate, along with a 95% CI based on the number of throws in each sample assuming the baseline error rate (i.e. assuming there is no effect and sampling the same number of throws).

_SD_____ #ofSD ____ 95% CI

.00234 ___ 1.88 ___ .0104-.0197 ___ ≤10 G in a season
.00263 ___ 2.17 ___ .0098-.0202 ___ ≤10 G in a season, ≤50 G previous experience
.00312 ___ 2.12 ___ .0107-.0232 ___ No Past Experience

One thing to note is that the last group doesn’t have that much overlap with the first two (~70% of the players in the “no past experience” sample are not in the other two), so the different samples showing the effect in the same direction is stronger evidence of an effect than any of the samples by themselves.

Jetsy Extrano
7 years ago

MGL, do you think individual 1Bs are learning to field the position up through age 30, or are we seeing a selection effect? I imagine that the guy placed at 1B at age 22 is probably big and slow, inherently weak fielder, might move to DH by 30. But looking at age 30 1Bs, some are guys who at age 22 were shortstops, and may still be pretty good athletes. These are selected examples but it seems like some of this effect would exist.

I feel like I’ve seen delta-method aging curves for fielding by position, but can’t find them.

MGL
7 years ago
Reply to  Jetsy Extrano

I ALWAYS use the delta method for aging and things like that, so my fielding aging curves use the delta method.

MGL
7 years ago
Reply to  MGL

That being said, you still may have selection bias even when using the delta method because the delta method treats the entire pool as if it were one player. For example, say you had a few 20-25 year olds start out at first base. And say that they started to get worse at age 28 or so. Now, say you had a bunch of players starting to play 1B at 28 or so and they got better because they were just learning the position. If you combine the delta data it might appear as if the peak age were 30 or more but that might only be because the new first baseman at 28 are skewing or dominating the data. Make sense?

james Wilson
7 years ago

Highly rated infields who lose their slick fist baseman suddenly aren’t so highly rated. When infielders know their throws can’t be bad enough not to be picked or snagged, they also play with more confidence.

!B places the least demands on talent, but application is another issue. 3B places fewer demands upon talent than SS–by far–but great short stops are not often great third basemen, and vica-versa especially so.