Acronyms like ERA, WHIP and WAR can tell you how good a pitcher has been, but xFIP is one of the best abbreviations when it comes to predicting how good a pitcher will be for the rest of the season.
Unraveling the Mystery of xFIP
You might not know exactly how ERA is calculated or fully understand the various scenarios in which a relief pitcher can be credited with a save. But if you watch baseball more than once in a blue moon, you likely have a general idea how those traditional statistics work.
Baseball stats go way deeper than that these days, though, and “expected fielding independent pitching” (known in shorthand as xFIP), might be a completely foreign concept to the spreadsheet-averse subset of MLB fans.
Give us just a few minutes, though, and you’ll learn plenty about one of the most telling pitching statistics that FanGraphs has to offer. If you’ve already read our guide to WHIP and wanted to learn about a stat that’s both tougher to calculate and better at predicting the future, this guide to xFIP should be right up your alley.
What is xFIP in Baseball?
Expected fielding independent pitching is admittedly word salad. In oversimplified terms, though, it is an estimate of what a pitcher’s earned run average (ERA) should be if he both had league-average fielding behind him and allowed home runs on fly balls at the league-average rate.
We’ll get into the main difference between FIP and xFIP momentarily, but the general theory behind xFIP is that pitchers have minimal control over the rate at which fly balls go for home runs. As such, their HR/fly ball ratio should, with enough innings pitched, eventually regress to that season’s league average, which has hovered around 12-13% for a decade.
This isn’t always the case over the course of a few games or even an entire season, and one kind of extreme example of that from the early 2020s is Sonny Gray.
Gray finished 2021 with an uncommonly high HR/FB ratio of 16.5% and had a 4.19 ERA to show for it. Two years later, he allowed home runs on just 5.2% of fly balls, which was both one of the lowest ratios in more than a decade and largely responsible for a 2.79 ERA that nearly led the American League that year.
While he had wildly different marks in ERA, Gray’s xFIP in 2021 was 3.66 and his xFIP in 2023 was 3.64, suggesting he was more or less the same pitcher in each season, just with significantly different luck on fly balls.
xFIP vs. FIP: Key Differences
Before we make you go cross-eyed with a few mathematical formulas, let’s touch on the big difference between FIP and xFIP, which is the HR/FB ratio mentioned above.
While FIP uses the number of home runs actually allowed by the pitcher (or team), that portion of the xFIP formula is replaced with the product of fly balls allowed and the league average HR/FB ratio — or the number of home runs the pitcher (or team) should have allowed.
In other words, if a pitcher’s HR/FB ratio is the same as the league average, his FIP and his xFIP will be identical. This was the case for Aaron Nola in 2021 and for Luis Castillo in 2023.
As far as the applicable difference between the two is concerned, FIP is an approximation of what the pitcher’s ERA should have been up to that point, while xFIP is an approximation of what his ERA will be moving forward, assuming strikeout, walk and HBP rates remain the same and the only real difference is regression to the league-average rate of home runs per fly ball.
The xFIP Formula and How to Use It
Apologies in advance if this awakens any high school algebra PTSD, but here is the math behind FIP and xFIP.
The formula for FIP is: [(13A + 3B - 2C) / D] + a FIP Constant
A is home runs allowed, B is the sum of walks allowed and batters hit by pitch, C is strikeouts and D is innings pitched.
The formula for xFIP is exactly the same as FIP, except instead of home runs allowed, the A in the formula is expected home runs allowed, which is fly balls allowed multiplied by the league-wide HR/FB ratio.
The FIP Constant is how ERA, FIP and xFIP end up being the same at a league wide level for each season. Contrary to the name, though, it isn’t actually a constant. It changes slightly each season (and within each season) and is calculated via its own similar formula.
The formula for FIP Constant is: E - [(13A + 3B - 2C) / D]
E is league-wide ERA and A, B, C, and D are the league-wide versions of the same statistics from the FIP formula. (The FIP Constant usually lands in the 3.10 to 3.25 range.)
Who’s ready for a practice problem?
In 2021, the FIP constant was 3.17. NL Cy Young winner Corbin Burnes pitched 167.0 innings with 234 strikeouts, allowing seven home runs with 34 walks and six hit batters. He also allowed 115 fly balls with a league-wide HR/FB ratio of 13.6 percent that season, for an expected HR total of 15.64.
Burnes’ FIP for that season was: [(13*7) + 3*(34+6) - (2*234)]/167 + 3.17, which simplifies to 1.63.
Burnes’ xFIP for that season was: [(13*15.64) + 3*(34+6) - (2*234)]/167 + 3.17, which simplifies to 2.30.
His actual ERA was 2.43. But even though he led the majors in ERA that season, both his FIP and xFIP suggest his ERA should have been even lower.
One key thing to note before we move on is that xFIP makes no effort to account for either park factors or a pitcher’s history when estimating how many home runs he should have allowed. As with all statistics, context is key.
Generally speaking, though, if it’s early August and a pitcher’s xFIP is considerably higher than his ERA, the next two months could be rough. On the flip side, pitchers with an xFIP much lower than their ERA could be gearing up for a strong finish.
Much like ERA, an xFIP below 3.00 is mighty impressive, 4.00 is roughly the league average and anything north of 5.00 is dreadful.
Top 10 xFIP Leaders
While FIP is calculable all the way back to the stone age of baseball—Old Hoss Radbourn had a career FIP of 3.17 from 1881-1891—xFIP only dates back to 2002, which is when fly ball and ground ball data started being reliably tracked.
In those more than two decades of MLB history, though, there have been more than 200 pitchers who logged at least 1,000 innings pitched.
Through the start of the 2024 campaign, here were the 10 career xFIP leaders (min 1,000 IP):
- Jacob deGrom, 2.83
- Chris Sale, 2.97
- Clayton Kershaw, 3.00
- Stephen Strasburg, 3.00
- Curt Schilling, 3.18
- Gerrit Cole, 3.19
- Randy Johnson, 3.23
- Roy Halladay, 3.23
- Brandon Webb 3.30
- Corey Kluber, 3.31
Notably, most of Schilling’s and Johnson’s careers played out prior to 2002. And while all-time greats Roger Clemens, Pedro Martinez and Greg Maddux did each meet the 1,000 IP minimum for 2002 and beyond, none of them ranked top 10 in xFIP in those latter stages of their career. As a result, take the leaderboard with a grain of salt.
The 10 best xFIP marks in a single season by a qualified pitcher (at least 1 IP per team game) are:
- Shane Bieber, 2.04 in 2020
- Clayton Kershaw, 2.08 in 2014
- Clayton Kershaw, 2.09 in 2015
- Curt Schilling, 2.20 in 2002
- Corbin Burnes, 2.30 in 2021
- Jacob deGrom, 2.46 in 2020
- Gerrit Cole, 2.48 in 2019
- Randy Johnson, 2.48 in 2002
- Felix Hernandez, 2.51 in 2014
- Corey Kluber, 2.52 in 2017
Two big takeaways from that list are 1) peak Kershaw was so good and 2) it’s a crime that the 2002 Arizona Diamondbacks were unable to win even one game in the postseason with Schilling and Johnson both so dominant.
xFIP in Fantasy Baseball
Try as we might to find the secret sauce of a forecasting formula, no predictive metric can perfectly project how a pitcher will fare in the future.
But xFIP does better than most, and you can use it to your advantage in fantasy baseball.
It’s not particularly useful when it comes to draft prep, as there are plenty of other much more complex forecasting models that go into all of those preseason player rankings. Midseason, however, is where it can be huge.
Take Yusei Kikuchi for example. When the Houston Astros acquired him from the Toronto Blue Jays at the 2024 trade deadline, he had a 4.75 ERA, but a 3.44 xFIP. And promptly upon arrival in the Lone Star State, he turned a major corner, posting a 3.19 ERA in his first eight starts, each of which resulted in a key win for the Astros.
You don’t need to wait until the end of July to start looking for buy-low candidates, though. In mid-May 2024, Houston’s Hunter Brown had a 7.71 ERA, but a 4.20 xFIP that suggested better times were coming. He posted a 2.02 ERA over the course of his next 19 starts.
In mid-June of that same season, Minnesota’s Pablo Lopez had a 5.63 ERA and a 3.60 xFIP. He had a 2.32 ERA in his next 14 starts. It’s unlikely anyone actually dropped him, but you absolutely could have swung a buy-low trade at that point. Alternatively, if DFS is more your speed, his ROI for those first couple starts after that turning point was likely off the charts.
It’s not always that simple, but checking the FIP and/or xFIP is way better than just relying on a pitcher’s ERA to hold steady.
Frequently Asked Questions
Is FIP or xFIP Better?
Depends on what you’re looking for.
They’re both good and end up being pretty similar, especially over a large sample size. For instance, in 2,840.1 career innings pitched, John Lackey had a 3.92 ERA, a 3.94 FIP and a 3.99 xFIP.
FIP is slightly more volatile, since it’s rooted in the home runs allowed by the pitcher as opposed to the rate of home runs allowed by the entire league in that season. But that does mean it does a much better job of conveying whether a pitcher did a better or worse job of keeping balls in the yard.
For full-season review purposes, FIP is better, but for predictive purposes, xFIP reigns supreme.
How Accurate is xFIP?
Between the 2022 and 2023 seasons, there were 191 cases of a pitcher throwing at least 50 innings prior to July 1 and at least 50 innings from July 1 onward. Slightly more than one third (65) of those pitchers had a second-half ERA within half a run of their first half xFIP.
That probably doesn’t sound all that accurate, but when xFIP tells us to expect a change, it’s usually correct.
Of those 191 pitchers, 93 ended the first half with a difference of at least half a run between his ERA and xFIP, suggesting some degree of imminent regression in one direction or the other. And of those 93, 74 (nearly 80 percent) did have a second half ERA that regressed in the direction that xFIP expected.
Yes, there were some major misses, most notably Taj Bradley in 2023. He finished the first half with a 4.75 ERA and a 3.05 xFIP, suggesting his ERA should be 1.70 runs better in the second half. Instead, it went 1.70 runs in the opposite direction to 6.45.
For the most part, though, xFIP is a solid predictor of what’s to come.
Bring Your xFIP xPertise to Sleeper
Now that you know a bit more about one of the primary statistics used in forecasting pitching, you might want to start applying that new-found knowledge in the world of fantasy baseball. If so, give Sleeper a try, either by signing up on the website or downloading the app. The fantasy offerings are plentiful, especially with the season-long potential of playing DFS with Sleeper Picks.
