TZ / TZL
While UZR and DRS are both based off of data collected by Baseball Info Solutions (BIS), Total Zone (TZ) is the lone defensive statistic calculated exclusively using play-by-play data available from Retrosheet. Invented by Sean Smith, it’s calculated in a variety of ways depending upon how much data is available in that specific year (details can be found here), but since it only requires play-by-play data, TZ scores can be calculated for any player in baseball history. As a result, TZ is used in historical WAR scores on Baseball-Reference.com and FanGraphs (pre-2002).
Total Zone with Location Data (TZL) is an improved version of TZ that Sean Smith developed in 2010. You can read about all its details here, but in short, it uses Gameday hit location data to make its calculations more accurate.
Context:
Like UZR and DRS, TotalZone and TotalZone with Location Data are both presented as “Runs Saved”. League-average is zero, while positive scores represent above-average fielding and negative scores denote below-average fielding.
| Defensive Ability | TZ/TZL |
|---|---|
| Gold Glove Caliber | +15 |
| Great | +10 |
| Above Average | +5 |
| Average | 0 |
| Below Average | -5 |
| Poor | -10 |
| Awful | -15 |
Things to Remember:
● TZ and TZL are both good metrics, although UZR and DRS are still considered the more accurate fielding metrics. However, UZR and DRS can only be calculated for modern-day players due to technology constraints, so TZ is the best historical fielding statistic available. If you want to compare a modern day player’s fielding with a player from the 1940s, TZ would be the statistic to use.
Links for Further Reading:
TotalZone Data – Baseball Reference
Updates to TotalZone Using MLB Gameday Hit Locations – Baseball Projections
Steve is the editor-in-chief of DRaysBay and the keeper of the FanGraphs Library. You can follow him on Twitter at @steveslow.
I adore your wp theme, exactly where did you get a hold of it from?
I still have absolutely no idea, if TZ based on play by play data, how it can still be calculated all the way back to 1871. Can anybody fill me in?
I think that for times before 1953 it looks at more primitive stats like errors.
I believe that’s an incorrect statement. Seems like TZ was based on pbp data back to 1950, and then Sean uses a different method that uses PO, A, E prior to 1950.