The Holy Grail of Defensive Evaluation?

July 12, 2009
Originally appeared at
Jeff Fletcher

SAN FRANCISCO -- Derek Jeter is the posterboy for all that's wrong about the way we evaluate defense among major leaguers. The "experts," coaches and managers, have rated him among the game's best shortstops by awarding him three Gold Gloves. The statisticians, however, rank him among the game's worst.

Defensive evaluation is an inexact science, for sure.

Not for long, though.

Sportvision, the California-based company that brought you the first-down marker on football telecasts and the K-Zone on baseball telecasts, has unveiled a revolutionary system that could be the Holy Grail of defensive analysis in baseball.

Sportvision showed off its embryonic Fielding f/x system on Saturday in a conference attended by a few dozen physicists, staticians and analysts from major league teams.

"I think we are on the cusp of having data that is going to be more accurate than we've ever had before," Giants analyst Yeshaya Goldfarb told FanHouse. "We'll be able to get a more accurate reading of whether a player is a good defensive player or a bad defensive player."

The best defensive metrics available now, things like Ultimate Zone Rating, are all dependent on humans to track where on the field a play is made, and how hard the ball was hit. Obviously, involving humans in that process makes the numbers inexact and inconsistent.

Sportvision is removing the human element.

For the past couple years, we've been used to seeing their Pitch f/x data. It is used on TV broadcasts to show where a ball crossed the plate.'s Gameday application uses it to show the velocity, movement and location of pitches. Hit f/x data uses the launch angle and velocity of balls off the bat. You'll get a good look at a similar technology during the Home Run Derby on Monday in St. Louis.

That type of data comes from fixed cameras around the ballparks that track the baseball.

Starting a few weeks ago, Sportvision installed four cameras high atop San Francisco's AT&T Park, and they have begun not only tracking the ball, but the players.

"We can compute the trajectory of the ball, the launch angle," said Marv White, Sportvision's chief technology officer. "We can do that for every batted ball. Every thrown ball. We know where the players are. Given all that, you can figure how fast they ran to get to a ball, and where they ran."

White said Sportvision is still experimenting with different types of cameras and locations, but eventually the goal is to have them installed in all 30 major league ballparks. When that happens, probably within the next couple years, they'll be able to produce a range factor for each player, taking into account how far he went from his starting position to reach balls hit at various speeds.

It's not clear how all that data might be incorporated into one easy-to-digest number, since there will be at least three variables: the percentage of balls reached, the distance covered and the time to get there. The important thing is the data will be there, so then the statisticians and analysts can figure out what to do with it.

White said eventually the data will be available to the public. In the short term, Sportvision has to make money off the technology, which probably means selling it to major league clubs for use in scouting.

"It's exciting," Goldfarb said. "The data is never going to be exact, but you are going to take the human element out of it and get actual numbers."