Fantasy Index

Fantasy Baseball Index

Pitcher ERA: A look at FIP, xFIP and SIERA

Todd Zola discusses why the various expected ERA metrics are better predictors than actual ERA.

"Past performance is no guarantee of future results." You've no doubt heard this with regards to stocks and investments. Well, it's also true in the realm of fantasy baseball.

As was mentioned last time, for the next several weeks, this space is going to focus on various expected stats or estimators. The driving force is outcomes don’t always reflect the skills exhibited by the player. Most expected stats provide an indication of what should have happened, with should being defined as the average outcome of a gazillion times the event, or series of events occurred.

Today’s topic is expected ERA, which are sometimes referred to as ERA estimators. This is the inaugural subject for a couple of reasons, which are probably related. ERA estimators are one of the oldest examples of expected stats, likely because ERA exhibits more variance than other stats. When fantasy baseball first became a thing, it didn’t take long to discover previous ERA was not a reliable predictor of future ERA.

Statistician Dwight Gill and sportswriter Tad Reeve are at the forefront of expected ERA. Well over 20 years ago they devised a formula converting hits, walks, strikeouts, and homers into an expected ERA:

Expected ERA = (.575 x H/9) + (.94 x HR/9) + (.28 x BB/9) - (.01 x K/9) - Normalization Factor

Empirically, the normalization factor ranged between .27 and .285 and was used to have the aggregate actual ERA match the aggregate expected ERA.

The notion was the inputs for the Gill and Reeve algorithm are all within the pitcher’s control. It’s since been shown pitchers have varying degrees of control over the different skills. Hang in, we’ll get there soon. Based on what was known at the time, the practical application was if a pitcher repeated his skills the following season, a better predictor of the ensuing ERA was the expected mark, and not the actual ERA from the previous season.

Data collection and computing power greatly improved over the years and is still being refined. Currently, there are a handful of publicly available ERA estimators and more proprietary methods. Let’s focus on those in the public domain, for which the origin is known and the metric freely available on Fangraphs.

FIP: Fielding Independent Pitching

FIP is just a fine-tuned version of Gill and Reeve’s formula since it incorporates the same inputs, excluding hits (more on that in a minute).

FIP = ((13 x HR) + 3 x (BB + HBP) – 2 x K))/IP + normalization factor

Like before, the normalization factor equates league ERA and league FIP.

xFIP: Expected Fielding Independent Pitching

Noted sabermetrician Dave Studeman refined FIP to account for the randomness of home runs. We’ll focus on this in more depth in upcoming discussions, but for now it’s best to think of homers in a similar vein to BABIP (batting average on balls in play) where pitchers settle around a league mean. As explained, we are learning pitchers have some degree of control over their BABIP, but it isn’t as much as many perceive. The akin notion for homers is pitchers settle around a mean HR/FB (home run per fly ball) mark. Pitchers surrender more or fewer fly balls, fueling their home run rate, but the HR/FB for each pitcher should be about the same.

xFIP takes this into account by normalizing each pitcher’s HR/FB to league average. It’s not perfect, since pitchers have a small measure of control over their HR/FB, but big picture wise, it gives a better reflection of the basal skills of the hurler.

xFIP = (13 x (#Fly balls x league average HR/FB) + 3 x (BB+HBP) – 2 x K)/IP + normalization factor

HR from FIP is essentially expected HR in xFIP by assuming every pitcher sports the league average HR/FB level.

Neither FIP nor xFIP is perfect, hence the term ERA estimator. The imperfections aren’t necessarily flaws; you just must understand what FIP and xFIP are trying to express.

For example, hits are excluded since it is assumed that all pitchers exhibit similar hit rates. This is the DIPS theory previously discussed, and the notion every pitcher should regress to league mean has been debunked. Not only do fly ball pitchers usually sport a lower BABIP than groundball pitchers, but the quality of a team’s defense, and the park can fuel BABIP.

SIERA: Skill-interactive Earned Run Average

As mentioned, there are a few proprietary forms of expected ERA, mostly addressing that batted ball types have more of an influence on ERA than FIP and xFIP suggest. SIERA is a publicly available version accounting for batted ball types. It is the brainchild of Eric Seidman and Matt Swartz, and was first presented on Baseball Prospectus:

6.145 - 16.986 x (SO/PA) + 11.434 x (BB/PA) - 1.858 x ((GB-FB-PU)/PA) + 7.653 x ((SO/PA)^2) +/- 6.664(((GB-FB-PU)/PA)^2) + 10.130(SO/PA)((GB-FB-PU)/PA) - 5.195(BB/PA)*((GB-FB-PU)/PA)

Key:

  • SO: Strikeout
  • PA: Plate appearance
  • GB: Ground ball
  • FB: Fly ball
  • PU: Pop up

The involved math is beyond the scope of this piece. Those interested in how the sausage is made can click on the above link and/or do a web search on something like “MLB SIERA formula”.

The key for fantasy baseball analysis, irrespective of using FIP, xFIP or SIERA, is they are best utilized to pinpoint players with marked differences between their actual and expected ERA. These players are then put under the microscope to get a better feel for why they over or under produced their basal skill levels. Players with an ERA estimator much lower than actual could be draft targets since their price could reflect some of last season’s bad luck. On the flip side, an ERA much lower than the associated estimators is likely to go up the impending campaign, so this could be a player to avoid since the market may price him based on the artificially low ERA.

In full disclosure, I tend to regress actual ERA to xFIP and SIERA in projections (which are part of the Fantasy Index Baseball Draft Kit). To reiterate, this discussion was intended to be an introduction to ERA estimators. The terms will be referenced in the Draft Kit’s player profiles. Do a search on FIP, xFIP and SIERA for a more intense treatment.

Todd Zola is an award-winning fantasy baseball writer and 2020 inductee into the Fantasy Sports Writers Hall of Fame. He's the content provider for the 2024 Fantasy Baseball Index Draft Kit, available now. To purchase, click HERE.

Fantasy Index