On ERA Indicators
I look at the most popular ERA indicator stats and explain how they work and analyze which are most useful
There are four big “ERA Indicator” stats, or at least that’s what I call them.
FIP
xFIP
SIERA
xERA
The purpose of these stats is to show you what ERA to expect in the future for a given pitcher. They are better predictors of future ERA than ERA itself. For an example, here’s Ronel Blanco’s 2024 season:
29 GS
2.80 ERA3.97 xERA
4.17 SIERA
4.15 FIP
4.09 xFIP
If you had to predict his 2025 ERA, you would expect to be much more accurate by predicting something much closer to four than three, given the huge differences in the indicators vs. the actual from 2024.
Now, let’s quickly go over how all four of these generally work.
FIP (“Fielding Independent Pitching”)
The way these all work is to strip out the factors that the pitcher does not control. After the pitcher throws a pitch, the rest of the outcome is out of his hands (literally!). After a ball is in play, the pitcher is depending on his fields to get outs. We don’t want to dock a future ERA prediction on a pitcher because his fielders cost him or saved him a bunch of runs last year. The outcomes that fielders have no say in are:
Strikeouts
Walks
Home Runs
No fielder touches the ball in those three events. So FIP, xFIP, and SIERA focus on these three outcomes to predict future ERA. Those are the basics. Here’s what FanGraphs says about FIP, and it gives the calculation.
xFIP
xFIP is very similar to FIP. The only difference is that it adjusts the home runs allowed based on some luck factors using HR/FB.
We know that there is a bit of randomness in HR/FB year-to-year, and we should expect regression to the mean on that front. So xFIP factors that in and does not penalize or credit pitchers who had outlier results in HR/FB in a given year.
SIERA (Skill Interactive ERA)
The big difference between SIERA and the FIP indicators is that SIERA considers ball-in-play data. While FIP/xFIP ignores non-homer balls in play, SIERA attempts to give correct credit or punishment based on the types of balls in play allowed. Here’s an interesting example from last year:
Jose Soriano: 3.80 FIP, 21% K%, 10% BB% 4.03 SIERA
Joe Ross…….: 3.83 FIP, 21% K%, 9% BB%, 4.38 SIERA
Soriano and Ross had identical profiles in those first three stats, but you see a significant difference in SIERA. Soriano gets the edge mostly due to his 60% GB% (Ross was at 41%). It’s fair to expect Soriano to outperform similar K-BB% pitchers because his balls in play are going for extra-base hits less often because they are usually on the ground.
It turns out that SIERA is the best predictor of future ERA of the bunch, slightly beating out xFIP.
xERA (Expected ERA)
This one comes from Baseball Savant, and it’s the newest of the bunch. It also works much differently. If you know what xwOBA allowed is, you know what xERA is. All xERA is that xwOBA allowed number translated into a number that looks like an ERA.
The way xwOBA works is to generate a wOBA for each ball in play based on the launch profile of the batted ball. So, you hit a ball at 105 miles per hour at 20 degrees - that’s a very high expected wOBA. If you hit a ball at 80 miles per hour at a 0-degree launch angle, that’s a very low xwOBA.
It takes all of those balls in play (and factors in strikeouts and walks) and then gets one number to show you how much damage to the pitcher was expected based on all of that, and then it translates that to something that looks like ERA.
Here is how they all correlate together:
It is interesting to see that the least correlated with same-year ERA are the two that are most correlated with next year’s ERA. The reason for that is that those two take away the actual home run results and adjust them for what we should expect in the future.
Case Study
Let’s look at a pitcher who had one of the most interesting 2024 seasons as far as the ERA indicators go. That man is Reynaldo Lopez.
25 GS
1.99 ERA
3.88 xERA
3.58 SIERA
2.92 FIP
3.44 xFIP
Any time you see an ERA under two, you can be pretty sure there was some good luck involved. You will always find big gaps between actual and predicted here, and that’s the case with Lopez.
While we’re on that topic, here are the lowest marks in each indicator from last season, just to give you an idea of how low these can actually go.
FIP: Chris Sale 2.09
xFIP: Garrett Crochet 2.38
SIERA: Garrett Crochet 2.53
xERA: Paul Skenes 2.50
The number I want to focus on for Lopez is the difference between his FIP and xFIP. For pitchers with at least 20 starts, he had the sixth-largest differential here.
His FIP was fantastic at 2.92, but his xFIP was much worse at 3.44. The reason for that was his 8% HR/FB. The league average (from the way FanGraphs calculates it) is 12%, so he was four points better in HR/FB than average. That would lead us to expect him to give up more homers next year, as that HR/FB is very likely to gravitate towards 12%. xFIP considers that, and FIP does not. And that explains the difference.
Takeaways
It’s useful to look at all of these together. For the sake of simplicity, I tend to stick to SIERA. It’s the most predictive of future ERA, and it does consider ball-in-play profiles, which are in the pitcher’s control to a certain extent. If you don’t like SIERA, I would tell you to use xFIP. I think that HR/FB regression is a very good thing.
If you want the full data, it’s all available on FanGraphs, but I’ve also put it in a Google Sheet that I used for this post here.