RotoWire Partners

The Saber's Edge: Predicting ERA

Jeff Zimmerman

Zimmerman writes analytics-focused baseball and football articles for RotoWire. He also handles scouting and reporting for and contributes to, and Jeff is also a two time FSWA award winner, including the 2016 Football Writer of the Year award.

Any early-season implosion can ruin a pitcher's ERA for what seems like an entire season. It could be a starter who gave up four 30-mph wind-aided home runs, or a reliever that allowed a three-run bomb in his first inning pitched of the season. Owners will see the 6.00+ ERA and may be ready to move a pitcher who has put them in an early hole. Be the owner who helps them out of their desperate situation by knowing when to buy low. One way to do this is by using an strikeout and walk-based ERA estimator called kwERA.

ERA estimators (FIP, xFIP, SIERA) exist to help put an unexpected ERA into perspective. These estimators can help to find pitchers who should regress to a more normal ERA as the season progresses.

This early in the season, I stay away from the most common ones because of some small sample issues. Instead, I like to use kwERA, which only uses a pitcher's strikeout (K%) and walk percentage (BB%). Besides having a limited number of variables, both of those stats stabilize quickly.

K% and BB% are the first pitching stats to stabilize, with 70 batters faced the value for K% and 170 batters for BB%.

Using K% and BB% are proven to be more predictive of the next season's ERA than other estimators.

I am not looking at season-to-season. Instead, I am looking at an early-season stretch compared to the entire season. First, here is the kwERA equation for 2013:

kwERA = 5.26 -10.9 * K% + 9.8 * BB%

From 2013 pitching data, I looked at all the pitchers with 10 IP in March and April and at least 20 IP for the entire season. I compared their March/April ERA, FIP, xFIP, kwERA, and regressed kwERA (r_kwERA) to full season ERA. Here are the average standard deviations from the final ERA and the various March/April estimator values.

Predictor in March/April: Std Dev from final ERA

ERA: 1.04
FIP: 0.88
xFIP. 0.81
kwERA: .80
r_kwERA: .78

Of all the factors, ERA is the worst future predictor with r_kwERA coming out on top.

Note: I know not everyone will have the time or the desire to calculate r_kwERA all the time, so if given the choice between the commonly available estimators, use xFIP. Its standard deviation is little more than the kwERA values. Additionally, it is an improvement over FIP and ERA.

So, here are all the pitchers rank by r_kwERA (min 5 IP):

A few thoughts:

Everyone knows Felix Hernandez is good, but I am surprised a bit by how well Masahiro Tanaka has pitched. I figured it would take him a while to adjust to the majors. It will be interesting to see how teams adjust to him the second or third time they face him.

Stephen Strasburg's ERA struggles (5.33 ERA vs 2.80 r_kwERA) are centered on his .394 BABIP. His career BABIP of .294 is not considered to be high. A 100-point difference should eventually regress some, but maybe not under .300. One reason for the limitation is that the Nationals' defense looks to below average right now. The entire team is allowing a .322 BABIP (29th in the league) and the defense has a -8.6 UZR (tied for 27th).

I had to look at LOOGY Jake Diekman and his 5.43 difference (8.38 ERA vs 2.95 r_kwERA). In his career he has a 2.13 FIP vs left-handed hitters (LHH) and a 3.71 FIP vs right-handed hitters. So far this season he has faced 13 lefties and 30 righties. What could go wrong?

Ervin Santana is a perfect buy-high candidate. He has a nice and shiny sub-1.00 ERA and his owner may be looking to sell high. Go ahead and buy. He is the ninth-highest rated starting pitcher according to r_kwERA. The move to the National League is nice for his stats with the pitcher's hitting ninth and the unfamiliarity some hitters have with him.

Jesse Chavez just gives me the shivers being a Royals fan. He allowed 2.1 HR/9 during those two dreadful seasons. I think there is some good and bad with Chavez. The Good. He has moved from a two pitch (four-seam/slider) guy, to a kitchen sink pitcher with a four-seam, two-seam, cutter, curve and change (each pitch he throws over 10% of the time). Also, he is carrying a 52% groundball rate to go along with his 27% K% and 5% BB%. I am not sure how long he can keep it up, which brings us to the Bad. He is pounding the strike zone at a career high rate (57% of the time). With more pitches in the zone, hitters have just decided to just not swing at the pitches (his swinging zone% is down from a career 64% mark to 53% this season). Normally, two times a pitcher's swinging-strike rate will equal their K%. Right now Chavez is at a career low 8.5% swinging-strike rate, which projects out to a 17% K%, rather than his current 27%. It will be interesting to see how lineups adjust once they see him again and start swinging more.

As an owner with too many shares of R.A. Dickey, I thought he turned the corner during the second half of 2013. Right now he looks like he is going around in circles. His ERA is inflated a bit, but even if it levels out, his >14% BB% is just killer. It seems players are figuring out the knuckleball. In 2012, 34% of the time hitters swung at pitches out of the strike zone. In 2013, that figure dropped to 31%. This season, it is at 22%. If batters aren't chasing the knuckleball for weak contact or swings and misses, he is not going to be effective. I think his owners should wait for one of those three-game stretches when he puts it all together and then try to move him.

In deeper leagues, where Roenis Elias is in play, move him. People may see a useful starter with a sub-3.25 ERA. Instead, concentrate on the 16% K% and 12% BB%. That combination will quickly get a pitcher out of the league.

Tim Lincecum. Will he ever figure out how to pitch instead of throw? He is striking out more hitters than he has in five years. He is walking fewer hitters than ever, but he's getting rocked for home runs as his velocity has dropped another 1.5 mph this season. He has given up at least one home run in every start this season. His home run rate (2.6 HR/9) is greater than his walk rate (1.7 BB/9). Generally, Lincecum isn't home-run prone with a career 0.7 HR/9 and a value near 1.0 the past two seasons. He has me torn on the correct action. Any starter with a 6.0 K/BB rate should be on everyone's radar, but will the home runs ever stop leaving the yard with a sub-90 mph fastball? I think it would come down to the make-up of my team. If my team is deficient in pitching and doesn't have a clear solution, I may try to acquire him and hope the home runs stop. If my pitching is set and I don't want to roster the volatility, I may move him.

All data as collected from FanGraphs before the games on 4/22/14.