It all starts with a pitch. The lonely pitcher sits on his mound and throws a ball towards a hitter. That pitch could turn into any number of things and whatever happens will be pinned on the pitcher. For roto purposes, we’re interested in a few statistics which come out of a pitchers performance on the mound: ERA, WHIP, K’s, Wins and Saves. A pitcher throws a lot of pitches in a year and sample sizes help reduce the influence of luck but it still plays a major factor in anything beyond a strikeout or a walk from a pitcher. Identifying luck and evaluating past statistics helps us better predict a pitcher’s future performance which makes us smarter fantasy baseballers.
ERA is a fickle mistress that is much more cruel than any other statistic because a pitcher can pitch like a champion but his ERA may still stink (see Zack Greinke 2011). If more batted balls become hits than expected or if more flyballs become home runs than expected then the pitcher’s ERA is going to be affected. In Part 1 of my series of posts on how to evaluate pitcher statistics, we’re going to focus on how to evaluate a pitcher’s ERA in a given season.
Earned Run Average
ERA has been constantly poked and prodded by sabermatricians as they have tried to figure out how to better be able to predict a pitcher’s future ERA. Because a perfect solution hasn’t been found, we’ve been left with a cornucopia of choices that, when combined, give us a clearer picture of a pitcher’s performance in a season. In the journey of being able to evaluate ERA, Voros McCracken came out with the DIPS (Defense-Independent Pitching Statistics) theory that set out to “evaluate a pitcher base[d] strictly on the statistics his defense has no ability to affect” which would be strikeouts and walks and home runs. It has developed more over time to include batted ball types (groundballs, flyballs, etc) as well. But based on the initial theory, one potential ERA estimator came into the world eventually courtesy of Tom Tango and it was called FIP.
FIP (Fielding Independent Pitching) only looks at strikeouts, walks and home runs allowed in order to estimate what a pitcher’s ERA should have been in a given time frame. For some, the simplicity of that turns them off but others can get totally behind how easy it is. When the outcome isn’t a HR, BB of K (in other words, the ball is hit into play), the pitcher doesn’t have as much control over the outcome and it becomes dependent on the defense. Balls hit into play are thus ignored and the FIP formula accounts for the three controllable outcomes (HR, BB, K) to generate a stat that looks like ERA but gives an indication on if a pitcher has been better or worse than his actual ERA.
xFIP (Expected FIP) was the next step in ERA estimation as research started to show that perhaps a pitcher doesn’t have complete control over HR totals. Researchers believed that HR rates were not reliable from year-to-year for a pitcher so the standard FIP formula was adjusted to use a standard league-average HR rate instead of the actual rate of HR’s allowed by a pitcher. This one simple change ended up making xFIP one of the most accurate predictors of future ERA.
tERA (True ERA) then came into the picture because there was thought that FIP shouldn’t ignore all batted balls except for homers. So this formula was designed to include batted ball types allowed by the pitcher (groundballs, flyballs, etc) by putting a different run value on each and using them in the formula. While not as predictive as xFIP, it utilizes a bit more data to generate an ERA estimation for a pitcher.
Because three estimations of a pitcher’s correct ERA was not enough, SIERA (Skill Interactive ERA) came along in recent years. As you can tell, sabermatricians still were not satisfied with the metrics of evaluating ERA. In this case, SIERA is an improvement over tERA much like xFIP improved FIP. SIERA tweaked tERA by putting different weighting on the variables such as giving additional weight to strikeouts, walks and groundballs as research started to show that they each had a larger influence on ERA than originally thought.
Luckily, you don’t need to break out your scientific calculators or look for one of my Google Doc spreadsheets to find all of these stats in one place. Fangraphs has these all hosted on their leaderboards under the Advanced tab. Go ahead and sort to your heart’s delight to see which pitchers should be better or worse moving forward than their stupid ERA indicates.
Keep in mind that ERA is influenced greatly by BABIP and HR/FB, which are thought to be out of control of the pitcher himself as well. While the four ERA predictors above are helpful, don’t overlook BABIP and HR/FB when you’re evaluating a pitcher. If they had a really low HR/FB rate suddenly then that would bring down their ERA quite a bit (same goes with BABIP). When a pitcher has a few years of history to analyze, you can more easily spot when a BABIP or HR/FB rate does not seem to fit his career norms. Put special attention to those cases and recognize the effect that they have on a pitcher’s actual ERA.