Stat-A-Day Challenge #1: UZR
First I will attempt to give a basic overview of what UZR is and what it is used for:
UZR or Ultimate Zone Rating is, in its most basic form, a statistic used to put a run value on a player’s defense. It takes into account several factors on every defensive play, measures them and places a value on them. If you’re wondering who compiles this data, it is done by a group called Baseball Info Solutions. It is designed to give a more accurate account of a player’s defense than the traditional methods like counting errors. It also rewards players for the amount of ground they cover. If player A makes a diving play to catch a ball and player B (hypothetically) catches that same ball without a dive, to the naked eye it would appear player A made the more outstanding play, but UZR would award both players equally. If a defensive player has range, UZR will reward them for it. UZR is not blown away by your unnecessary diving plays Mr. McLouth! In general, UZR provides an objective account of a player’s defensive value. It’s not perfect, but it’s probably the best defensive statistic out there.
What data is used to calculate UZR?
Baseball Info Solutions uses a series of measurable data to calculate UZR:
Batted Ball Types - All batted balls are placed in the following categories: ground balls, bunt ground balls (Only used for 1st, 2nd and 3rd Basemen), outfield line drives and outfield fly balls. Infield fly balls (Fly balls within 180 feet of home plate) are ignored because they are almost always caught and when they are dropped it is almost always a matter of confusion on the field between fielders. Also on infield fly balls certain fielders would be able to hog more of the fly balls and thus increase their UZR. Infield line drives (Line drives within 180 feet of home plate) are ignored because they usually caught and only dropped as a result of a mix-up.
GIDP - There is no special distinction for a double play. They are treated the same as a single out. GIDP’s are taken into account in another calculation as above or below average based on the speed and location of the ground ball and the number of double plays turned per that type of ground ball.
Arm - As in an outfielders arm. These values are calculated separately from “regular” UZR values based on the speed and location of the given batted ball and how often the runner advances an extra base (Advances), doesn’t advance and extra base (Holds), or is thrown out trying to advance an extra base (Kills). Park factors are also part of the equation. So playing left field in Fenway Park is treated differently than playing left field in PETCO Park.
Base Runner/Outs Adjustments - UZR assumes the following: With a runner on 1st and less than 2 outs the SS and 2B are playing at double play depth which is shallower and closer to second base, with a runner on 1st and no one on 2nd it is assumed the 1B is holding the runner and is therefore close to the first base bag, and with runners on 1st and 2nd and less than 2 out the third baseman is playing in anticipating a bunt.
Other Infield Positioning - UZR categorizes left and right handed batters separately and also categorizes the batters into one of two speed categories based on a seperate speed calculation. Players are considered either above-average speed or below-average speed. Using this data UZR assumes that defenders are usually playing a bit more shallow and are more hurried with a speedy runner at the plate. If there was a shift on for a certain hitter and it is deemed to have affected the play at all by the person recording the data then the play is ignored in the data collection.
Outfield Positioning and Wall Balls - On balls hit in the air to the outfield the batter is grouped into one of two categories: batters with less than average power and batters with greater than average power (Just like they were divided for speed). They are categorized based on the average distance of their average distance of their balls hit in the air. If a ball hits the outfield wall, the person collecting the data determines whether or not the ball was catchable. If it was, then it is treated like any other batted ball. If it wasn’t then the data for that batted ball is ignored.
Park by Park Adjustments - Every park is subject to infield and outfield park adjustments. The length and type of grass of the infield is taken into account, as are is the altitude and average temperature for the outfield. In addition, the outfield zones are divided into short and deep zones in left, right and centre field for a total of six zones. Each zone is the subject to the same park adjustments. A park, like Fenway for example, has separate adjustment values for each zone. Deep left field in Fenway has a rating of 0.5 (with 1.0 being MLB average) for the difficulty of making a catch in deep left field. The number represents the overall catch rate of that area of the field. So the 0.5 in deep left field at Fenway suggests that it is more rare to make a catch in deep right field. Adjustments are made to these park factors each time a material change occurs to the field or if the team moves.
So what does the average player’s UZR look like?
The average UZR at each position should be 0.0 or thereabouts. If you added up all the UZR’s of the players at one position the total should be 0.0, but there may be a slight variance due to rounding errors. This is because the UZR is measured against the average player for that position and year so there should be an equal total on either side of the break even mark.
How accurate is UZR?
Whenever comparing data for baseball it is important to accumulate enough data to justify labelling something as a trend. With defensive metrics it is always suggested that you use three years of data to draw conclusions on a player’s fielding ability. This creates a bit of a problem because of the demand for data that tells us how good a player is now. It is a slight downfall of defensive metrics and UZR in particular, but overall the positives of UZR outweigh the negatives in my opinion. One complex and thorough statistic is better than none.
So if a player has a high UZR one year it does suggest he had a good year that year in comparison to the league average, but it doesn’t necessarily earn that player the title of elite defender. If they do it over a period of 3+ years, then we can begin to consider them an elite or above-average defender.
How do I compare players’ defense with UZR?
As I mentioned earlier, UZR is an accumulative statistic. So if Player A plays 140 games and Player B plays 80 games, Player A will have had a significantly greater opportunity to accumulate a higher UZR. Due to this, the best way to compare two players’ UZRs is to use a modified version of UZR/150. This averages out the players UZR per 150 games (Duh!) to make it more equal and therefore more comparable.
Conclusions: What’s left to say?
Well, I think I’ve summed up UZR about as well as I can in this short piece, but I’ll end this blog with a bit of a case study (or whatever you want to call it).
Let’s take a look at the 2010 Gold Glove winner’s UZR and then rearrange the awards based on the highest UZR’s for each position respectively. From there you can draw your own conclusions on UZR. It is definitely interesting to look at.
AL Gold Glovers (UZR Gold Glovers in Brackets)
1B - Mark Teixeira, UZR: -2.9 (Daric Barton, UZR: 12.1)
2B - Robinson Cano, UZR: -0.6 (Mark Ellis, UZR: 9.9)
SS - Derek Jeter, UZR: -4.7 (Alexei Ramirez, UZR: 10.8)
3B - Evan Longoria, UZR: 11.1 (Kevin Kouzmanoff, UZR: 16.1)
OF - Franklin Gutierrez, UZR: 7.3 (Brett Gardner, UZR: 21.9*)
OF - Carl Crawford, UZR: 18.5 (Crawford)
OF - Ichiro Suzuki, UZR: 15.6 (Suzuki)
NL Gold Glovers (UZR Gold Glovers in Brackets)
1B - Albert Pujols, UZR: 1.5 (Ike Davis, UZR: 10.1)
2B - Brandon Phillips, UZR: 9.7 (Chase Utley, UZR: 10.3)
SS - Troy Tulowitzki, UZR: 7.1 (Brenden Ryan, UZR: 11.5)
3B - Scott Rolen, UZR: 10.6 (Chase Headley, UZR: 16.5)
OF - Carlos Gonzalez, UZR: -2.7 (Andres Torres, UZR: 21.2)
OF - Shane Victorino, UZR: 2.6 (Jay Bruce, UZR: 20.2)
OF - Michael Bourn, UZR: 17.6 (Bourn)
*Highest in MLB
(Remember pitchers and catchers are not included in UZR)
So obviously UZR and the naked eyes of those who vote on the Gold Glove Awards have differing opinions as only three of the fourteen recipients are agreed upon. Just something to think about.
Hope you enjoyed segment #1 of my Stat-A-Day challenge. Tomorrow I will tackle a more simplistic, but still valuable statistic: BABIP.