When I first became interested in sabermetrics almost ten year ago (thank you, Rob Neyer), one of the major tenets was that fielding just isn’t very important. Maybe that was because current fielding measures were bad (errors, assists, etc.) and there wasn’t anything better. People didn’t know how to handle fielding, so they ignored it.
Ironically, debates about player value in 2007 seem to under-rate fielding. Ryan Braun appears destined to win NL Rookie of the Year, with his poor fielding reputation just a subjective footnote. Joe Mauer and Derek Jeter lost the 2006 MVP award to Justin Morneau, who was barely a better hitter and not even one of the top ten players in the league.
We’ve hit the time of the year when there’s a lot of discussion about player value — awards, team retrospectives, and 2008 projections — so it’s important to get a good handle on fielding. I’m going to talk about the wrong approach to judging fielding, the right approach, specific stats that take the right approach, and the effect of using the right approach on player value. Feel free to skip down if you’re in touch with the basics.
The Wrong Approach
Traditional fielding stats include errors, fielding percentage, assists, putouts, total chances, and maybe double-plays turned. Anything that’s just a raw count of plays made is highly dependent on the number of groundballs given up by the team’s pitchers. There’s a difference between playing behind Barry Zito and Chien Ming Wang. It’s not how many plays you make, but how many you make with what you were given. Therefore, assists, putouts, total chances, and double-plays are all out the window.
Errors and fielding percentage (the second is just errors as a rate stat) are a different beast. Yes, errors are bad. But crappy fielders can actually have very low error-rates. Imagine a fielder who can’t move, but has great hands — say, Luis Castillo circa 2007. He won’t commit many errors, but groundballs will constantly roll by him that would be scooped up by most other players. Is it worth not screwing up 20 plays per season in exchange for not getting to 30 hits? No way. Range is the name of the game, especially for outfielders.
A Better Approach
Any decent fielding metric should compare plays made with plays that could have been made. These days, all the big name stats providers (STATS, BIS, etc) track where each ball is hit, dividing the field up into zones. Stats such as Zone Rating hold each fielder accountable for certain zones. You get credit for each ball hit into your zone that you turn into an out and get docked for each one you don’t, regardless if it’s an error or ball you can’t get to.
The next step is take into account the difficulty of balls hit into each player’s zone of responsibility. If one player happens to have 500 balls hit right at him, he shouldn’t come out looking better than a player who has 500 balls hit to the edge of his zone. Precise zone data is important for these advanced metrics. With it, you can assign different locations a difficulty level based how often balls in each zone are turned into outs by the league as a whole. Then, for a specific fielder, compare how many of the balls hit towards him you’d expect an average fielder to turn into outs with how many he actually turned into outs. This is often called a +/- system, or advanced zone rating system.
And finally, since number-of-plays-made doesn’t mean a whole lot to most people, you can convert to runs prevented. A routine grounder through the legs might cost the team .6 runs on average (the full difference between an out and a single). A difficult grounder down the line is almost always a double, so not making the play only docks the third baseman .1 runs. But if he stretches out to make the play, he gets credited .8 runs (most of the difference between a double and an out). Plays that are neither routine nor impossible are graded on a sliding scale.
The great part about judging fielders on the scale of runs is that you can easily convert to wins (10 runs is about 1 win) and you can combine them with offensive runs. For example, Derek Jeter’s hitting is worth 40 runs and his fielding is -15 runs, so he’s worth 25 runs overall.
So What Stats Are Best?
Because zone data takes a lot of effort to create, it costs money. Teams can pay for it, successful websites can pay for it, and rich people can pay for it. Teams haven’t given away their numbers, but people from the second two categories have. Check out Ultimate Zone Rating (UZR) by Mitchel Lichtman, who used to work for the Cardinals. Check out the Probabilistic Model of Range (PMR) by Dave Pinto, who used to work for ESPN. Check out Justin’s +/- stat based on zone data from The Hardball Times (the same data that goes into John Dewan’s Fielding Bible). None are perfect (due to design flaws, data that’s not precise enough, and differences in the way balls in play are recorded by different stats providers) but the benefits far outweigh the lack of total accuracy.
If you don’t want to bother with the whole advanced fielding metric thing yourself, at least use Defensive Efficiency Ratio (DER), a team-level stat that’s simply outs made divided by balls in play. It’s like batting average from the fielders’ point of view. Good fielding teams turn more balls into outs.
Ok, Why Should I Care?
Simply put, the difference between the best and worst players in the field is a significant chunk of the difference between the best and the worst players at the plate. Good fielding metrics aren’t just more accurate, they help demonstrate the relative value of hitting and fielding.
Take, for example, Ryan Braun and Troy Tulowitzky, whom I discussed yesterday. Compared to replacement level, Tulo’s offense has been worth 35 runs and Braun’s 50 runs. That 15 run difference is worth about a win and a half. But (depending on which fielding metric you use) Tulo is +20 runs compared to average while Braun is -20. That’s 40 runs in Tulo’s favor, making him a better player by 25 runs, or 2.5 wins. Still want to ignore fielding?
The difference between positions is also important. Vlad’s more or less an average rightfielder. But would you play him at shortstop? No way. He’d be a sieve. So just by being able to play shortstop, Miguel Tejada help his team more than if he played right field. How much more? However many runs Vlad would cost your team at shortstop. Tango once came up with these relative values between average fielders at each position:
Pos Runs
CA +10
CF +5
SS +5
2B 0
3B 0
RF -5
LF -5
1B -10
DH -15
The range of fielding value over a full season goes from Manny (about -30 runs — some say more) to Adam Everett (+25 runs) plus a 10 run positional difference. In 2007, ARod’s created 155 runs, an awesome number. If you go down the list 65 runs (the difference between Manny and Everett), you’ll find the likes of Aubrey Huff and Johnny Damon. Would you rather have a guy who hits like ARod and fields like Manny or a guy who hits like Aubrey Huff and fields like Adam Everett? The numbers say it doesn’t really matter.
But that’s why fielding matters. You can turn ARod into Aubrey Huff.
Popularity: 5% [?]
Share This
Sky is a baseball fan and racket sport afficianado living in upstate NY. His favorite color is orange and is just about ready to give up on his life-long dream to become the next Magnus ver Magnuson (World's Strongest Man). His favorite baseball teams are the Yankees and Red Sox, proving that there's hope in the Middle East.