The unfortunate fact of most fantasy leagues is that while ERA is a category, it is not very strong as a predictor of future performance. So how can fantasy players better predict ERA for pitchers moving forward if they can not rely on their performance in the category to date?
In addition to my weekly Down On The Farm pieces I will be writing a weekly fantasy piece called Sabermetric Mining, about how to use advanced statistics to aid you in your roster management. I start this week by looking at Defensive Independent Pitching Stats (DIPS), specifically Fielding Independent Pitching (FIP), Expected Fielding Independent Pitching (xFIP), and SIERA (Skills Interactive ERA).
DIPS Theory was originally put forth by Voros McCracken, who posited that pitchers exhibit little control over what happens once they have pitched the ball. Defense, park factors, and dumb luck can all have an impact on what happens when the ball enters the field of play, so a lot of the times the pitcher will be rewarded or penalized in a way that fails to reflect what he may have deserved. As an example, a pitcher who walks three batters and then allows a home run is charged with four earned runs. However, if he allows a home run and then three walks, he is charged with just one. Here, the pitcher has allowed three walks and a home run in both scenarios, but is penalized in very different ways in terms of earned runs against his ERA. DIPS formulae use different methods to try and strip out the impact of luck, defense, and event sequencing in order to provide a better indicator of the pitcher’s performance and, hopefully, a better predictor of their future performance.
It should be noted that none of the statistics discussed claim to be ERA predictors. Rather, they are backward-looking statistics that happen to better predict future ERA than ERA alone does.
FIP – Fielding Independent Pitching – This stat normalizes BABIP (batting average on balls in play) and event sequencing, basing the formula on walks, strikeouts, and home runs allowed. These indicators are then applied to a constant to put the metric in ERA form.
xFIP – Expected Fielding Independent Pitching - This stat takes FIP and then normalizes the pitcher’s home runs allowed based on how many fly balls they allow and the league average home run rate. This basically takes FIP and applies the caveat that pitchers can control the amount of fly balls they surrender, but not how many of them clear the fence.
SIERA – Skill-Interactive ERA SIERA was created as an alternative to FIP and xFIP, arguing that pitchers can control the types of balls in play they allow, even if they can not control the outcome of those balls. Since groundballs and pop ups are more desirable than fly balls or line drives, pitchers who display a specific trend in batted ball data should have their stats adjusted as such. The stat is far more complicated than FIP and xFIP, as it also controls for factors like walks being less damaging the fewer you allow and high strikeout pitchers allowing weaker contact. While slightly more complicated, the theory is straightforward…but is it more effective?
Which Is Best?
When SIERA first came out, Matt Swartz at Fangraphs tested the different DIPS metrics to see which was the best predictor of future ERA. If you are a math person, I suggest checking out the article for the specifics, but allow me to summarize his results as they apply to using these stats to predict the next year’s ERA:
The gap between SIERA and xFIP is relatively small, while the gap between xFIP and FIP is more pronounced. Basically, using xFIP or SIERA will both be helpful, while using FIP will help to a lesser degree.
With that said, when it comes to predicting rest of season ERA, it should be noted that FIP will do better than these tests show. This is because FIP does not normalize for home run rate, which may be in part impacted by ballpark factors. While xFIP will apply a normal home run rate for pitchers, a pitcher pitching in, say, Petco Park may exhibit a low HR/FB for the rest of the season, making FIP slightly more valuable than in end-of-season analysis.
I tend to favor using SIERA when available, and FanGraphs has all three available on their leaderboards.
How To Use
The best way to use these statistics is to look at pitchers who have ERAs that greatly differ from their FIP, xFIP, or SIERA. What gaps between ERA and one of these metrics will tell us is when a pitcher’s ERA is under- or over-stating their actual performance to date. I will go through some examples to illustrate, but basically you would hope to find pitchers with ERAs higher than their DIPS number and identify them as potential buy-low candidates, while finding pitchers with ERAs lower than their DIPS numbers and idenitfying them as potential sell-high candidates.
Again, these metrics do not claim to be ERA predictors, so when I say that a player is a buy-low, it is not because SIERA is predicting a better ERA moving forward, but that SIERA indicates the pitcher has been better than his ERA would suggest. So if a leaguemate sees only a 4.00 ERA but SIERA indicates a pitcher performing at a 3.50 level, you might be able to get a 3.50 ERA-skilled pitcher for the price of a 4.00 one.
Dan Haren – Haren’s ERA is a disappointing 4.59 so far this season, but his peripheral stats indicte that he is still a decent pitcher, though maybe not at the level we have come to expect. Haren has a 3.91 SIERA (4.01 xFIP), based on the fact that he rarely walks batters (2.18 BB/9) and has been somewhat unucky in terms of BABIP (.315). His batted ball data is in line with his career norms, but his HR/FB rate is inexplicably up, causing him to allow a career-high 1.4 HR/9. Based on this, we would anticipate Haren’s ERA to fall more in line with how he has actually performed, creating a buy-low opportunity.
Cliff Lee – The case of Lee is compounded by the fact that he has just two wins, but predicting wins is a fool’s errand so we will focus on his ERA, a respectable 3.73 mark that is significantly higher than what we have come to expect from him. Lee sports a SIERA mark of 3.15 (and an xFIP of 3.12), drawing our attention to the fact that his dominance (8.55 K/9) and control (1.72 BB/9) have not faltered at all. In addition, Lee is generating a career-best 48.1% groundball rate at the expense of line drives. Lee’s batted ball data lines up favorably with his previous performance, indicating he is still pitching like the Cy Young candidate we know, but happens to be suffering from a career-worst 13.0% HR/FB rate. Lee is a great buy-low opportunity, a chance to acquire an ace, albeit without the wins, at the price of a second-tier starter.
Ryan Vogelsong – Vogelsong has been quite the story again this year, sporting a 2.22 ERA after most thought his 2011 mark of 2.71 was a mirage. While I will touch on pitchers who can perhaps consistently outperform their peripherals later, let’s explore the case of Vogelsong anyway as an illustrative example. Vogelsong strikes out few batters (6.59 K/9), is not an elite control pitcher (2.98 BB/9), and is not an extreme ground ball ptcher (43.4%). While he has allowed fewer line drives than last year, his BABIP of .250 is unsustainably low, and he has stranded 84.8% of base runners, well above the league average. Vogelsong has also allowed home runs on just 6.1% of fly balls. As such, SIERA sees Vogelsong as a below-average starter (4.32) and xFIP sees him as more or less unownable (4.46).
James McDonald – While the bloom has come off the rose over his last few outings, McDonald still sports a 3.38 ERA for the year. SIERA (4.03) and xFIP (4.01) agree that McDonald is more of a league-average starter, primarily because his strikeout rate has regressed while his walks have increased. Whatever McDonald had going in May has been lost, and his better-than-average BABIP and LOB% do not appear sustainable based on his batted ball profile.
When To Adjust For Certain Pitchers
Obviously, none of these stats are perfect, and none claim to estimate future ERA with much a degree of certainty. Simply, they are better tools than looking at ERA alone. While there are countless examples of players regressing after outperforming their peripherals, there are also a few examples of pitchers consistently performing outside of DIPS.
Matt Cain is the primary example, having posted an ERA better than his xFIP/SIERA in every single season of his career, leading to a 3.30 career ERA despite a 4.20 career xFIP and a 4.13 career SIERA. Up until this year, Cain had always shown an ability to limit HR/FB%, has posted a low BABIP in every year but one, and has consistently been among the league leaders in LOB%. Possible explanations include a strong ability to pitch out of the stretch, the ability to fool hitters with four above-average pitches, or Matt Cain just being really good. I think at this point you can safely say Cain has an ability to outperform his peripheral stats, but I would not dare claim to know how.
Ricky Nolasco is the counter-point to Cain, a pitcher who has consistently had a higher ERA than you would expect based on his other stats. Over 169 career starts and 16 relief appearances, Nolasco has posted a 4.55 ERA despite a 3.80 SIERA and a 3.76 xFIP. The explanation for a pitcher underperforming their peripherals is an easier narrative than in Cain’s case, as Nolasco might simply make too many mistakes. This is an easier justification than “Cain doesn’t make mistakes, ever,” and as such Nolasco may have been written off entirely by the fantasy community. Specifically, while he has improved his groundball rate annually, he has done so by sacrificing strikeouts, meaning he is no longer even an attractive commodity for strikeout totals.
Hopefully this article has helped explain some of the defense independent pitching stats available to fantasy owners, and how you can use indicators like xFIP and SIERA to add some value at the trade table. Let me know if there are any specific stats you would like to see examined for future pieces.
I’m new here, so come get to know me on Twitter @BlakeMurphyODC.
All stats courtesy of FanGraphs.