Leading into the home stretch, let’s discuss some defensive statistics. As a category, these represent recent innovations in baseball, as defense is the most dynamic aspect of the game. Players can change where they are positioned, which affects their ability to get to a ball. As we gather more data — ball speed off the bat and launch angle — we can better analyze defensive ability. But it’s still messy.
Fielding Percentage
As per usual, we’ll start simple. We keep track of a few basic events:
- Fielding Chances: The plays where a particular player is said to have a “chance” to field. Strictly speaking, it’s the sum of the three events below.
- Putouts: Direct action taken by a player to cause a runner to be out. This includes catching a ball in the air, catching a ball and tagging a base for a force, or tagging a runner for an out.1First basemen accumulate a lot of put-outs. In addition, catchers get a putout for every strikeout they are part of.
- Assists: Indirect action taken by a player to cause a runner to be out during a play. Every ground ball to an infielder who then throws to first base results in an assist for the infielder, and a putout for the first baseman.2Multiple assists can be awarded on a single play to everyone involved until the final putout, and assists are even granted if the play ends in an error that results in an out not being recorded.
- Errors: Any bad play that causes an out to not be recorded that otherwise would have. This involves dropping a ball hit in the air, or throwing the ball poorly.
For a long time, the only defensive metric that anyone used was Fielding Percentage (FP), which measures how frequently a player successfully completes their part of a defensive play.
\text{FP} = 1-\frac{\text{Errors}}{\text{Total Chances}}Equivalently:3The numerator in the below equation, Putouts plus Assists, is called chances accepted.
\text{FP} = \frac{\text{Putouts}+\text{Assists}}{\text{Total Chances}}This measure still holds on as something to consider, but in the 1980s some baseball statisticians pointed out that this can unduly reward players with poor range, who tend to make plays when they get to them, but aren’t quick enough to get to as many plays as some others in their position. If you make 100% of your plays but only get to 10 balls, while someone else makes 90% of their plays but gets to 15 balls, that second player had a greater positive impact.
Range Factor
A first small attempt to correct FP as a defensive statistic was the introduction of Range Factor. As the name implies, it’s a proxy for defensive range. If a player is quicker and able to get to more plays, we expect them to have more opportunities on average than a player with a worse range. So, instead of dividing by total chances, we normalize by playing time.
\text{RF} = \frac{\text{Putouts} + \text{Assists}}{\text{Total Innings Played}}The innings played is innings of defense played at the position in question.
Of course this metric is still rather crude, but if you combine it with fielding percentage you get a broader view of a player’s defensive contributions. It is important to note that first basemen and catchers have inflated putout numbers, so range factor is a less effective tool for them. In addition, we should only use range factor to compare players at the same position and, ideally, around the same timeframe. This is far from a normalized, advanced statistic that allows generalized comparisons across positions, ballparks, and generations.
A Comparison
Fielding statistics aren’t available on Stathead, so I’m doing some custom comparisons in this fielding series.
Let’s look at Nolan Arenado, who is almost certainly the top defensive third-baseman of the last ten years, and Trevor Plouffe, a pretty decent player from my youth who played for the Minnesota Twins.
We’ll compare their performance in 2015, where they played a comparable amount of time (157 games at third for Arenado, and 140 games at third for Plouffe.4Plouffe also played a handful of games at first base that season.) During that time, Arenado had 1,362 defensive innings, and Plouffe had 1,217 defensive innings.
Arenado’s fielding percentage was .966, and Plouffe’s was .972. So, we’d consider Plouffe on the surface to be ever so slightly more consistent in his fielding. But, let’s calculate each of their range factors.
With 105 putouts and 385 assists, Arenado’s range factor was
\text{RF}_{A} = \frac{105+385}{1362} \approx .360.So, he had about 0.36 successful fielding plays per inning, or 3.2 per 9-inning game. Plouffe’s was
\text{RF}_{P} = \frac{101+277}{1217} \approx .311.Plouffe got himself .311 successful plays per inning, or about 2.8 per 9-inning game, nearly half a play fewer per game.
Remember, range factor is a measure of successful plays. Despite Arenado making a few more errors (17 against Plouffe’s 11), he did so while playing nearly 150 more innings than Plouffe, and per inning he made more plays successfully. Thus Arenado was a more impactful defensive third baseman, at least by the analysis Range Factor provides us.5We’ll keep exploring this comparison as we walk through other defensive metrics, because third base is my favorite defensive position.
Continue to Day 22 – Ultimate Zone Rating
- 1First basemen accumulate a lot of put-outs. In addition, catchers get a putout for every strikeout they are part of.
- 2Multiple assists can be awarded on a single play to everyone involved until the final putout, and assists are even granted if the play ends in an error that results in an out not being recorded.
- 3The numerator in the below equation, Putouts plus Assists, is called chances accepted.
- 4Plouffe also played a handful of games at first base that season.
- 5We’ll keep exploring this comparison as we walk through other defensive metrics, because third base is my favorite defensive position.