In previous installments, we've covered hitting and pitching statistics. We finish off the series by taking a look at fielding numbers, and also other stats that are often used to show player performance - in particular, WAR (Wins Above Replacement). As before, we may refer to some principles previously discussed, so it's best if you have read the first two parts. We will be just on the other side of the jump...
E = Errors.
The official scorer shall charge an error against any fielder whose misplay (fumble, muff or wild throw) prolongs the time at bat of a batter, prolongs the presence on the bases of a runner or permits a runner to advance one or more bases.
-- MLB Official Rules, 10.12
This is at the core of basic fielding statistics, and is quite simple. A fielder drops the ball - it's an error. They throw it away - it's an error. Fumble it? An error. Count them up for a season, and more errors is bad. However, there are a number of problems. Firstly, they are an objective decision, because they are a subjective decision based on whether the play should have been made "with ordinary effort." It's a vague criteria, subject to interpretation and individual opinion and, potentially, bias.
Witness the Troy Tulowitzki triple, which completed his cycle in August 2009. The outfielder bobbled the ball in the corner, and the cut-off man then threw wildly past the player covering third. It's hard to argue that, if these misplays had not happened, Tulowitzki would "have been put out with ordinary effort." Yet, rather than being scored as a double and an error, the Coors Field scored gave Troy-boy three bases, completing the cycle. [The obvious home-field bias here birthed the SnakePit meme which says, whatever happens, Tulowitizki is awarded a triple.] However, it is important to stress that such blatant homerism is very much the exception.
But it is the case that teams can apply pressure to the scorer, by asking for a review of plays, though there is a certain tension here. A pitcher will generally prefer to see errors than hits, as errors lead to runs being unearned, which keeps his ERA down. The fielder charged with the error may have a different opinion on this, but they are generally seen as secondary: fielding percentage is not as important to them, as ERA is for a hurler. Conversely, batters want everything to be scored a hit. The pressure can be quite intense, and I know of at least one report of a scorer at Fenway never getting to work there again, because he wouldn't reverse a call he made.
There is also the generally accepted rule that a fielder can't be charged with an error if they don't touch the ball. So, make a great diving stop, then spike the throw to first so the 1B can't quite glove it, and it's an error. Let the ball go past you to left-field and it's a hit. The result is the same - a runner on first - but in one case, the fielder gets the blame. If you don't get to many balls, you won't make many errors: put a traffic-cone (or, alternatively, Adam Dunn) in right-field, and it'll probably play close to error-free baseball over the season. Not necessarily a good thing, though.
F% = Fielding Percentage. Over the course of a season, different players will touch the ball a different number of times, so F% attempts to take this into account. It's the number of putouts plus assists, divided by putouts + assists + errors (the number of chances). Much the same problems arise with F% as with errors; while it's true that whether a play is recorded as a hit or an error will also affect, say, batting average, over the course of a year and 500+ PAs, the number of questionable hits is small, likely no more than a handful - so would have minimal impact on batting average, likely less than ten points.
However, a variation in ten points of F%, due to the subjective nature of errors, is a bigger deal because of the much narrower range between the best and worst. Indeed, the gap in F% last year between the best- and worst-performing first-basemen in the majors, among those who qualified, was only nine points. A couple of harsh errors or generous non-errors would alter your standing considerably. There is also a wide variation by position, so a great F% for a third-baseman, would be mediocre for a second-baseman, and downright dreadful at first. The average F% across all positions in the majors last year was .983, but here's how the various spots broke down:
- First-base: .993
- Catcher: .992
- Center-field: .989
- Right-field: .986
- Left-field: .985
- Second-base: .984
- Shortstop: .971
- Pitcher: .954
- Third-base: .951
These have been discussed in some depth on the 'Pit previously, in a three-part series in 2009, informally titled Fielding Metrics Made Easy, Fielding Metrics Made Somewhat Complex and Fielding Metrics Only That Bloke Off Fringe Understands, so I will refer you to those for details. What you need to know about advanced fielding stats is less how they are calculated and more what their limitations are. This more or less applies to all of them, whether it's Rtot, UZR, Rdrs or +/-.
However, a brief overview might be helpful, and UZR - Ultimate Zone Rating - is probably the best one to know. It divides the field up into squares. Based on previous data, if a ball it hit to square #76, has a 40% chance of being a hit, and 60% of becoming an out. So when the ball is hit there, the fielder gets a credit or debit, based on whether they make the play. Add these up over the course of the season, and you can figure out how many runs were saved or cost by that player. This is, of course, incredibly simplistic - Fangraphs has a more in-depth primer if you want the details - but the basic idea is the same for most of the systems.
The main problem is that they all use different methods of calculations, and indeed, rely on different data sources - there is no standard source of information regarding where a batted ball is hit. So you've got metrics which take different data and manipulate it in different way - it's not really a surprise that the results can be different, sometimes strikingly so. Generally, I suggest you remember the story of the blind men and the elephant; and treat each as a piece of the puzzle, rather than the entire picture. What this means is, if one stat shows a player as great, or terrible, then it's less significant than if all of them say the same thing. So the more defensive stats you look at, the better.
[To go some way to addressing this issues, Fangraphs just came up with "Aggregate Defense Ratings", which is a weighted average of the four metrics they show: we'll see how this works out going forward.]
Not everyone is enamored of Wins Above Replacement. Or as its critics say, "WAR, huh. What is it good for? Absolutely nothing." Hey, you knew that was coming, so I figured we might as well get it out of the way quickly. The notion of WAR was largely created by Sean Smith, who was the creator of the CHONE projections and now, apparently, works for an un-named baseball team. The idea was to come up with a single number that would include multiple aspects of a player's value: for a hitter, this would include hitting, base-running, defensive range, defensive arm, ability at avoiding double-plays, etc.
Sean crunched the numbers to calculate a player's contribution to his team, measured in runs above or below what a hypothetical replacement player would contribute. He also added a position adjustment, to account for the fact that the value of an average fielder is not the same everywhere on the diamond. Once you've got a total for runs above and below replacement, you can convert those to Wins Above Replacement - the ratio there varies depending on the era, but is round about ten runs to one WAR.
Before we get into what is a good WAR for a player, an important point. There are two separate versions of WAR, available from Baseball-Reference.com and Fangraphs.com, known as rWAR (or bWAR) and fWAR respectively. They are not the same and produce different results, sometimes quite radically so. For example, in 2010, Chad Qualls was valued by rWAR at -3.0, while bWAR had him down at +0.3. This is because they are based on different pieces of data [specifically, rWAR uses ERA+ as its basis for valuing pitchers, OPS+ for hitters and Total Zone for defense. fWAR uses FIP for pitchers, wOBA for hitters and UZR for defense]
Which you prefer is largely a matter of preference, as both systems have their strengths. I like rWAR for pitchers, because to me, ERA+ is a better measure of a pitcher's actual contribution than FIP, which feels like an idealized and theoretical projection [can you look me in the eye and tell me Qualls was above replacement level last year?]. However, I'd agree that wOBA - and thus fWAR - is better for hitters. But one thing you can't do is mix and match: they use different replacement levels for a baseline, so you need to stick with one. The rest of the way, unless otherwise stated, when I say WAR, I mean rWAR.
If you asked an average Joe or Jane how many wins their team's best player was worth over a replacement player, I think they'd over-estimate it, as the answer may not be much above five WAR. Only 18 position players and a dozen pitchers reached that in 2010, so if you've got more than one on your team, you're doing well. 10 WAR is seriously rare: the last were both in the 2004 NL West: Barry Bonds and Adrian Beltre for the GIants and Dodgers, with the last pitcher, Pedro Martinez for the 2000 Sox. We've never had one of those: for AZ, the best in team history are 7.6 (Luis Gonzalez, 2001) and 8.8 (Randy Johnson, 2002) respectively. Overall, here's how B-R.com ranks WAR:
- 8+ WAR - MVP
- 5+ WAR - All-Star
- 2+ WAR - Starter
- 0-2 WAR - Substitute
- < 0 WAR - Replacement
The "market value" for WAR is considered to be about $4-5 million per Win, so that can be used to gauge the rough expectation from a player contract. For example, Justin Upton signed a deal for $51.25 million over six years, so if we get more than 13 WAR in total from 2010-2015, it has probably been a good investment [and he notched 3.8 WAR last year alone, so that's a good start]. But the range is huge. The best value among position players in the NL last year was Joey Votto, who earned $525K and produced 6.2 WAR; at the other end, Carlos Lee cost the Astros $19 million and was 1.6 wins below replacement level.
As always, please feel free to let rip with questions, comments or criticism on any of the above.