Nearly a decade ago Voros McCracken shook the sabermetric landscape with the simple theory that much of what we thought of pitching is actually more about the defenders behind him than the pitcher itself. The seminal piece on Baseball Prospectus can be boiled down to one, oft-quoted line:
There is little if any difference among major-league pitchers in their ability to prevent hits on balls hit in the field of play.
Basically, McCracken found that pitchers really only control a few key things like strikeouts, walks, and home runs. If you just think about it, it does make sense. Should Justin Verlander be rewarded because Adam Everett is a vacuum at shortstop? Should Verlander be punished because Carlos Guillen was manning that spot behind him back in 2006? No, he shouldn't.
After the jump, I'll give you all the reasons you need to to not use ERA as a stick against which to measure pitchers again...
What is Fielding Independent Pitching?
Using the things we've come to accept that are in the pitcher's control, it builds an ERA replicator. The number that comes out is how a pitcher has actually pitched and is a much better predictor of future success than a pitchers ERA.
How does it work?
The formula applies weights to each of the strikeouts, walks, and homers that reflect their run value. You then add a constant (usually around 3.2, but it changes from year-to-year) that puts it on the ERA scale. It is a great stat to see how a pitcher is actually performing, regardless of what the pitchers actual ERA looks like.
Also, given that FIP is simple, there are chances that luck can skew things -- namely in the home runs column. But, that is where Expected FIP (xFIP) comes in. It normalizes the home runs a pitcher allows to the league average because there isn't a lot of evidence that a pitcher controls his home run rate entirely. He can control it, but like Batting Average on Balls In Play, it can range from year-to-year. xFIP attempts to correct for that.
Flaws in the system
Whether you want to consider this a flaw or not, it doesn't track that actual number of runs scored on a particular pitcher's watch. Whether we like it or not, runs score while Verlander is on the mound. Some are his fault, some aren't.
Why you should use it
Alex Remington's recap on FIP said it best:
Basically, the fundamental difference between old-school stats and new-school stats is that old-school stats measure what happened at the surface level — batting average, earned run average, wins. New-school stats try to measure each player's contribution to those surface stats, while filtering out the contributions of their teammates and the random fluctuations of chance.
That's what makes FIP better than ERA. Want to get an idea of whether a pitcher is awful? Check his FIP. If his FIP is lower than his actual ERA, it's indicating that he's pitched better than his ERA, and that his defense or luck might not be on his side.
Hey, if it's good enough for Zach Greinke or for Max Scherzer, isn't it good enough for us?
Resources
Fangraphs, as always, is the first stop for advanced metrics of this nature. Alex Remington's piece I linked to above was a nice jumping off point, as well.