Basic Sabermetrics: Fielding Independent Pitching

Have you ever read an analysis article on the internet, or heard something on TV, and the writer or speaker was using a statistic you didn't really understand? Me too. At our worst, analysts can use stats and acronyms that everyone just doesn't understand, without providing context or an explanation.

As we try to change the conversation about baseball to something a bit more analytical and objective (at least in some arenas), we'd like to try and help those who're new to the field of sabermetrics -- or who need a refresher* -- get a little more information. So we're offering a new index page: All About Sabermetrics.

* - At the risk of being preachy, going back and re-evaluating what you know, or what you think you know, from time to time is a pretty good idea. It keeps me from being wrong as much as I would normally be wrong, which is a lot.

This article is the first in a series of pieces designed to better help our readers understand the complex (and not-so-complex) statistics that we use in our baseball analysis here at the site. You can find a link to our All About Sabermetrics page in the header at the top of the site, under the "Library" heading, as well as in each of these "Basic Sabermetrics" articles. Over the next few months, we intend to populate that library with loads of information on sabermetrics stats and concepts.

Best of all, these documents aren't just one-off resources -- they're designed to be living documents updated when we get new information, more in-depth research and better ways of explaining things. If you have a question, leave it in the comments for this page, and we'll do what we can to answer it. If you have a resource, leave it in the comments, and we'll try to integrate it.

And if you listen to our podcast here at the site, The Shift, we're going to continue to discuss stats in our "Better Know A Stat" segment. On Episode 1, Andrew and Neil discussed Fielding Independent Pitching, and did a great job of it. So, I thought that FIP would be a great starting point for the first of our "Basic Sabermetrics" posts here at the site.

Fielding Independent Pitching, or FIP, has gotten a lot of traction over the last several years as an important tool for analyzing pitcher performance. It's used heavily in writing and analysis, so it's a great candidate to start this series.

What is FIP?

Fielding Independent Pitching -- or FIP -- is a metric designed to give us information about pitcher performance. FIP measures the events that are directly under a pitcher's control: strikeouts, walks, and home runs. From there, a calculation is used to scale these events to a number very similar to the one you'll find for ERA. It is a "rate" statistic, that resembles how many runs a pitcher might give up per nine innings, given these peripherals.

For what it's worth, FIP is often pronounced as a one-syllable word: "fip" -- as in rhymes with "hip" or "dip". Some folks also spell it out in pronunciation as "eff-eye-pee". I'm not sure there's a wrong way to do it.

The formula for FIP was developed by Tom Tango, based on research on Defense-Independent Pitching (DIPs) done by Voros McCracken and published at Baseball Prospectus.

How to Calculate FIP

The formula for FIP is relatively simple, at least as far as sabermetric statistics go. It is calculated as follows:

FIP = ((13*HR) + (3*(BB+HBP)) - (2*K)) / IP + cFIP

The variables are pretty straightforward: HR = home runs allowed, BB = walks allowed, HBP = hit-by-pitches allowed, K = strikeouts, and cFIP = league FIP constant.

The league FIP constant changes from year to year, and is used to put FIP on the same league-wide scale as ERA. For 2013, that constant was 3.048.

And those multipliers that you see associated with homers, walks, HBPs and strikeouts? Those track with the average linear weights of those events. A home run changes the game more than a walk, etc., and those multipliers help reflect that.

Where to Find FIP

FIP is a cornerstone of the pitching statistics provided by FanGraphs. Their basic player pages and leaderboards* show FIP, updated on a daily basis during the season. You can also find it at Baseball Prospectus.

* - The league leader in FIP among qualified starting pitchers in 2013 was Matt Harvey of the New York Mets, with an FIP of 2.00. The worst among qualified starters for that season? Jeremy Guthrie of the Royals.

You'll see FIP used in writing about baseball a lot these days. Here at Beyond the Box Score, it's common to see FIP used when our writers talk about how well a pitcher has performed over the length of a season, or even over a career.

Using FIP in Analysis

So what is FIP used for? Well, most people consider it to be what a pitcher's ERA "should" be ... but that's not exactly right. While FIP is scaled to ERA, its best use is as a simple descriptor of those things that FIP measures, put into an ERA-scaled context.

ERA has other variables "built in" that skew our ability to use it as a measure of a player's true talent level. Not only is a pitcher's defense likely to show up in ERA numbers, but so is luck -- especially over a small sample size. Pitchers tend to have little control over whether a ball in play falls in for a hit, so luck is definitely a part of what shows up in a pitcher's ERA. With FIP, only those items that are considered fully within the purview of the pitcher are recorded, so there's less luck involved.

At the same time, FIP is not a catch-all replacement for ERA. From Tom Tango at The Book Blog:

So, the purpose of FIP is not to dismiss hits allowed and other non-HR contact plays, but simply to break it up into its own component (fielding-independent pitching, FIP). And, the fantastic byproduct of doing that is that even if you dismiss hits allowed and other non-HR contact plays, you STILL get a very similar answer [to ERA].

FIP is actually more predictive of future performance than ERA, so it may be a better indicator of success going forward. At the same time, FIP is not law. There are pitchers who consistently out-performed their FIP over long and illustrious careers. There are also pitchers who under-performed their FIP over long stretches.

The reason or reasons why a pitcher might have consistently under-performed or over-performed compared to their FIP could be many, but typically you think of pitchers (1) who do have some measure of control over their batted-ball profile or ability to give up hits or (2) who are able to change the way they sequence in order to strand runners on base or get favorable outcomes under different game circumstances.

If during the middle of a season, a pitcher's FIP is a full run lower than his ERA, then there's a chance that said pitcher is in line for some positive regression, and that their ERA will improve as a result. But this isn't law. A season is still a relatively small sample, and that regression may not happen -- even over the course of the full season. And at the same time, it's worth examining if that pitcher has a history of FIP staying below ERA over a long sample. Perhaps something else is in play that needs accounting for, and FIP doesn't tell the whole story.

Associated Statistics

FIP-: FIP- is a statistic that scales a pitcher's FIP against the league average. An FIP- of 100 is league-average. An FIP- of 120 is 20% worse than league average. An FIP- of 80 is 20% better than league average.

xFIP: Expected FIP (xFIP), developed by Dave Studeman at The Hardball Times, is a regressed version of FIP that replaces home runs allowed with the home runs they should have allowed. It is also a tool used to predict future pitching performance.

pFIP: Predictive FIP (pFIP), developed by Glenn DuPaul over at The Hardball Times, is another modified FIP metric that seeks to more accurately predict future performance by re-weighting the multipliers for the peripheral events and changing up the constant.

oaFIP: Opponent-Adjusted FIP (oaFIP), developed by Stephen Loftus here at Beyond the Box Score, is yet another modified FIP metric. This one seeks to take into account the quality of opposing hitters to provide an FIP-scaled metric that accounts for degree of difficulty.