Evaluating Fielding

From Wiki
Jump to: navigation, search

The effect of a player's fielding has traditionally been somewhat overlooked in the sabermetric arena. This is largely due to the difficulty in accurately measuring fielding value. There are two broad categories of systems for evaluating fielding based around whether the inputs include play-by-play data.

Traditional Systems[edit]

Traditional systems are based around relative basic fielding statistics that have been collected since early in the 20th century, generally putouts and assists. There are numerous systems in this category including:

One major drawback to traditional systems is that they are not able to take into account the number of chances for a given player. This limitation is caused because, until recently, score keeping did not record the location of a batted ball. Once hit location data was recorded, beginning in the mid 1980s, more advanced analysis could occur.

Play-by-play Systems[edit]

The advent of tracking the location of batted balls in the mid 1980s allowed for more complex fielding evaluation systems. Originally, this data was only available to official stats organizations. However, more recently, both Stats, Inc. and Baseball Info Solutions (BIS) have made this data available to the public for a price. Analysts have used this information to develop evaluation systems considered much more accurate than the traditional systems. Examples of play-by-play systems include:

  • STATS (as seen on SI.com)

Reliability of Evaluating Defense[edit]

As is probably apparent from the number of available systems, fielding evaluation at this time is far from perfect. There are many factors that contribute to a level of uncertainty that don't apply to measuring offense:

  • Differences in data collection between Stats and BIS lead to variations in zone definition and hit classification.
  • Ballpark effects are adjusted for differently by different systems (this can also apply to offensive stats if they are adjusted for park, though not necessarily in the same ways).
  • Conversion rates that are accurate at the league level may not hold at the team level.
  • Fielder position is not currently tracked.

Because of these reasons, most analysts believe the best results can be achieved by combining the output of multiple systems. However, you would not combine ZR and UZR for example, since UZR uses everything in ZR, plus more.