Matt Adams has Braves fans dreaming.

The mid-season acquisition was just supposed to be a stopgap for the injured Freddie Freeman, but he quickly proved as more than that. Since his May 21 Braves debut, he’s posted a solid .299 with 12 homers, the most among first basemen, with a wRC+ of 148, good for eighth among first basemen with at least 100 plate appearances over that span.

With that, Braves fans started to wonder if Adams could play anywhere else on the field to allow the Braves to get his potent bat in the same lineup with Freeman, who was bound to return in July or early August.

Of course, the Cardinals wondered the same thing, asked Adams to lose some weight, and proceeded to shove him out in left field at the start of 2017. That experiment lasted six games as Adams managed to accumulate minus-2 defensive runs saved (DRS) and an ultimate zone rating (UZR) that would have equated to around -34.4 over the course of 150 games.

How bad is that? Just one outfielder who was allowed to play at least 1,000 innings in a single season since 2003 ever put up a UZR worse than -34.4 (Brad Hawpe, 2008 Rockies).

That alone was enough to squash any chance of moving Adams elsewhere and most had resigned to the fact that Adams would be another Atlanta player shipped away for prospects at some point.

Then Freeman opened his mouth, volunteered to play third base and ushered in one of the most unique defensive experiments we’ve seen in baseball in a long time.

While Freeman did play third in high school and a whole five games in Rookie ball in 2007, he accumulated three errors over 48.1 innings. Sure, we saw Chipper move to left field to accommodate Vinny Castilla in 2002 and 2003 and we saw Alex Rodriguez move from shortstop to third when he became a member of the Yankees.

But left field is generally a less consequential defensive position and the move from short to third is actually easier in a lot of ways. First base is typically the position you move a player to once their defensive prime begins to pass as a way to keep them in the lineup.

Certainly, it’d be a nice boost to the Braves offense to be able to work both guys in the lineup, but can Freeman actually become a viable third baseman? That’s what I set out to determine.

## Quantifying defense: rPM, DRS, RngR, UZR

Essentially, there are two major methods for quantifying defensive ability — **defensive runs saved** (DRS) and **ultimate zone rating** (UZR).

These statistics can be quite involved, but I’ll provide a quick synopsis in order to give those less familiar enough of an understanding to move forward. Those of you familiar with these statistics can skip down to the next subhead, that’s where the good stuff starts.

DRS is a statistic developed by *The Fielding Bible.* Essentially, DRS combines a lot of smaller statistics that quantify different aspects of playing defense such as ability to play a bunt, range, arm ability, etc. into a simple number. DRS identifies how many runs a player’s defense is worth relative to the average player at their position. DRS is measured so that 0 is an average player at that same position, it does NOT mean that a player is an average defender compared to all players at all positions. So if a first baseman has a DRS of 10, his defense is worth 10 runs more than the average first baseman. Any other questions can be addressed by *The Fielding Bible.*

UZR works similarly. It takes into account different aspects of fielding and spits out a number where 0 is the average at a position and a positive number is number of runs above average, while a negative number is runs below average. Everything you need to know about UZR can be found in the Fangraphs’ Glossary.

Now when it comes to third basemen specifically, there’s two aspects of defense that are critical — arm ability and range. While Statcast is measuring a fielder’s throwing speed and some other potentially valuable defensive data, that is not yet publicly available, so we won’t be able to delve into that. However, range is a statistic that has been quantified both through DRS and UZR, so we can look at that as well.

DRS measures range through a statistic called rPM — **Plus/Minus Runs Saved**. To put it simply, this statistic measures a fielder’s ability to get to a ball and turn that fielded ball into an out relative to the average player at his position.

UZR uses a similar statistic called **Range Runs** (RngR). Like rPM, 0 is the average and it measures how many runs a player is worth based on the balls in play that they do or do not get to relative to the average player at the same position.

## So what does this have to do with Freddie Freeman?

Ok, I’m getting there.

I decided that I wanted to take the four statistics I explained above — rPM and RngR for range measurement and DRS and UZR for overall defensive performance — and see if I can extrapolate how a player performed at first base to get an estimate for how they might perform as a third baseman.

Luckily, DRS and rPM date back to 2003, so there’s actually a pretty decent sample size of players to draw information from. While UZR and RngR go back an additional year, I set 2003 as the cutoff date in order to have a more complete look at defensive measures.

It’s not enough to just identify players who played both positions at some point in their career after 2003. Unlike, say, OPS, these defensive metrics don’t have a very strong year-to-year correlation, meaning that just because a player is a bad defender in one season doesn’t necessarily mean they’ll be bad in the next. Because of this, I decided to identify only players who played both positions within the same season, so outside factors like weight gain/loss, injury, etc. wouldn’t affect the results as much.

I started with a pretty arbitrary threshold of 50 innings at each position, which isn’t a lot but it’s a start. I found more than 200 individuals seasons, which was more than I needed. I decided to weed out players who weren’t in the top 75 percent of innings played at both positions, which gave me 140 observations to work with and thresholds of 89 innings at first base and 108 innings at third base.

The other main issue with the data was that most players favored on position or the other, naturally, so the inning totals were way different. That matters a lot for these four statistics because they’re cumulative and not averages, so the more innings you play, the more chances you have to gain (or lose). To account for this, I used Fangraphs’ method of normalizing to the statistic per 150 games. Except I don’t care about games, I care about innings, so really, each statistic is extrapolated to what it would be over the course of 50 games worth — 1,350 innings.

Using those individual seasons, I decided to build basic linear models using each statistic at first base to predict the same statistic at third base.

### Range

First, let’s take a look at how Freeman’s range might translate to third base. Building a model to predict 3B rPM from 1B rPM, we get this:

*rPM(3B)=−7.19+0.36×rPM(1B)*

First of all, it’s important to note that this model is very statistically significant, which means I can feel good about using this data. Among the most important numbers in that model is the -7.19. This number indicates that a league-average first baseman has a range that’s -7.19 runs below the average third baseman. That’s really bad.

The other number in the model, the 0.36, shows that there is a relationship between range at first and range at third, and that better first baseman have better range at third base. Certainly that’s good, but what it actually indicates is that for every 1 run above average a player gains as a first baseman, he only gains 0.36 runs as a third baseman. This is what we call a diminishing return, because you’re actually losing .64 runs above average for every run a player gains at first base if you instead play them at third base. This is visualized in the plot below, which shows where a player would fall if his third base rPM were exactly the same as his first base rPM. Players who are better at third are in blue while players who are worse are in red.

The difference between the dotted line (1B rPM = 3B rPM) and the line of predicted values shows the diminishing return as a player becomes better at first base. The solid line also shows another important observation — we wouldn’t expect a first baseman to be a league average third baseman unless his rPM per 150 games was around +30. That’s an absurd tradeoff.

Range Runs shows a similar, although less defined relationship, between first base range and third base range:

*RngR(3B)=−3.47+0.16×RngR(1B)*

This model is only statistically significant at about 82 percent significance. In layman’s terms, that means I can only be 82 percent sure that a non-zero relationship exists between third base RngR and first base RngR. For this experiment, 82 percent confidence is confident enough.

This model shows a less drastic 3.5 drop in range runs for an average first baseman and a much less steep payoff for each additional range run accumulated as a first baseman, just 0.16.

### Total defensive performance

So range doesn’t seem to translate well to third base, which doesn’t bode well for Freeman. But maybe he can make up for that in other ways and his total defense won’t be that bad?

Well, don’t count on it. Here’s how we can expect Freeman’s DRS to translate as a third baseman:

*DRS(3B)=−7.27+0.31×DRS(1B)*

Like rPM, we’re highly confident that there’s a relationship between first base DRS and third base DRS. The average first baseman is worth just -7.3 defensive runs saved and each additional defensive run saved only translates to 0.31 runs at third base.

Also like the rPM model, the plot for the DRS data shows that in order to expect a first baseman just to be a league average third baseman, he’d need to put up a DRS around 40 per 150 games.

Finally, the UZR data posts a similarly bleak outlook on the translation to third base:

UZR(3B)=−6.66+0.21×UZR(1B)

The UZR model shows that a league average first baseman will be 6.7 runs below average at third base and an one-UZR increase at first base results in just a 0.21-UZR increase at third base. It would take a first baseman with more than 30 UZR in 150 games to result in just a league average third baseman.

### So… what does that mean for Freddie?

So we’ve established that Freeman certainly isn’t going to be a better third baseman than he is as a first baseman.

But does that mean he’ll be a bad third baseman? As a first baseman, he’s pretty much average. He’s put up a respectable 15 DRS over his career, which equates to about 2.5 defensive runs saved per season.

UZR likes him a little less, pitting him at -0.1 runs below average over the course of his career, essentially a league average player.

In order to put into perspective how his career numbers would be expected to translate to third base based on how he performed in 2017 before the injury, his career best, his career worst, and what he’s averaged over the course of his 7.5-year Major League career.

Also presented is the prediction interval, which essentially just tells us what the worst-case scenario would be based on the observed data and the best case scenario.

rPM | DRS | |||||||

Actual 1B | Predicted 3B | Low-End Pred. | High-end Pred. | Actual 1B | Predicted 3B | Low-End Pred. | High-end Pred. | |

2017 | 8.3 | -4.2 | -39.7 | 31.3 | 0 | -7.3 | -39.0 | 24.5 |

Career best (2016) | 5.7 | -5.2 | -40.6 | 30.4 | 8.6 | -4.6 | -36.4 | 27.1 |

Career worst (2014) | -8.4 | -10.2 | -45.8 | 25.3 | -6.5 | -9.3 | -41.0 | 22.5 |

Career average | 0.3 | -7.1 | -42.5 | 28.4 | 2.5 | -6.5 | -42.8 | 29.8 |

RngR | UZR | |||||||

Actual 1B | Predicted 3B | Low-End Pred. | High-end Pred. | Actual 1B | Predicted 3B | Low-End Pred. | High-end Pred. | |

2017 | -1.7 | -3.8 | -27.2 | 19.7 | 7.0 | -5.2 | -33.2 | 22.8 |

Career best (2015) | 4.1 | -2.8 | -26.3 | 20.7 | 6.2 | -5.3 | -33.3 | 22.7 |

Career worst (2011) | -11.6 | -5.4 | -29.0 | 18.2 | -11.6 | -9.1 | -37.2 | 19.0 |

Career average | -1.6 | -3.7 | -27.2 | 19.8 | 0 | -6.7 | -34.6 | 21.3 |

Now certainly, you’re welcome to look at those best-case scenarios and be optimistic, but the value that is the most important is the prediction itself, and not a single prediction has Freeman as more than around a 5.0 run below average third baseman, at best.

Of course, that’s over the course of a 150-game, full season. If Freeman meets his target return date for the Washington series before the All-Star game, that would give him 79 games before the end of the season. So using these predicted values, this is how we’d expect Freeman to perform at third base over those 79 games:

rPM | DRS | RngR | UZR | |

2017 | -2.2 | -3.8 | -2.0 | -2.7 |

Career best | -2.7 | -2.4 | -1.5 | -2.8 |

Career worst | -5.4 | -4.9 | -2.8 | -4.8 |

Career average | -3.7 | -3.4 | -1.9 | -3.5 |

While those numbers would make Freeman among the worst third basemen in the league, he wouldn’t be the worst in any statistic. But if any fanbase should know what a bad defensive third baseman looks like when it sees one, it’s Braves fans. Over the last three seasons, the Braves have had Adonis Garcia, a noted awful defensive player, manning the hot corner for more than 200 games. So how would Freeman’s expected 3B performance compare to Garcia’s based on their career averages over a 79-game span?

rPM | DRS | RngR | UZR | |

Freeman | -3.7 | -3.4 | -1.9 | -3.5 |

Garcia | -3.6 | -4.9 | 3.0 | -1.3 |

DRS favors Freeman, while UZR drastically favors Garcia. In reality, we should probably expect that Garcia would actually be an ever so slight improvement upon Freeman because of one important caveat — most of the players in this data set were third basemen by trade who shifted over to first base. While the data establishes a clear decrease in defensive ability from first base to third base, it does so among players largely considered capable third basemen already. Because of that, it’d be much more likely that Freeman’s numbers fall between the predicted value and the lower end of the prediction interval.

In short, this is a bad idea.

*Writer’s note: In order to keep this as terse as possible, I excluded some details on the methods. If you’re interested, feel free to reach out to me on Twitter or in the comments and I’ll be more than happy to go into more detail.*