Wednesday, March 14, 2012
The Components of ERA
By Tristan H. Cockcroft
ERA, or Earned Run Average, is one of the most instantly recognizable, frequently cited measures of a pitcher's performance.
The problem: It's also one of the most misleading measures for forecasting future success
and forecasting future success is the name of our game.
Sure, ERA accounts for two of the most critical skills of a pitcher -- recording outs and preventing runs -- but many external factors impact those departments, often skewing the results. Even the Wikipedia page for Earned Run Average hints at its misleading nature, noting that "because of the dependence of ERA on factors over which a pitcher has little control, forecasting future ERAs on the basis of the past ERAs of a given pitcher is not very reliable and can be improved if analysts rely on other performance indicators such as strike out rates and walk rates."
That's not to say ERA is entirely without value. For one thing, it remains one of the five primary Rotisserie categories. But before you blindly assume that a pitcher's low, past-season ERA -- or even his three-year ERA -- portends future success in the category, be sure to consider its context. This is something we do every season when formulating our projections, and it helps explain how a pitcher like Ryan Vogelsong, who had the sixth-best ERA in baseball in 2011 (2.71), can be projected to have an ERA a run and a half higher (4.21) in 2012.
Instead of taking Vogelsong's 2.71 -- or similarly successful or unfortunate pitchers' ERAs -- at face value, consider that any or all of the following factors might have had a strong influence on the final numbers:
In a literal sense, of course unearned runs have no impact on ERA; earned runs are the ones included in the calculation. But unearned runs are the product of faulty defense, a result of an error or passed ball, and while a pitcher doesn't get charged for many of the runs he allows after such a mistake, for evaluation purposes maybe he should. (At the same time, maybe he should get credit for the outs he'd have gotten when an error occurred instead.) Remember, recording outs is the most critical task for a pitcher, and just because his defense lets him down, his skills should be strong enough that he's capable of converting four -- sometimes more -- outs in a single inning when necessary. Those pitchers who allow themselves to be unraveled frequently by errors might have flawed skills, which could easily be exposed even when backed by a flawless defense.
These were the leaders in unearned runs in 2011, the chart including what percentage of the team's unearned runs that pitcher allowed, as well as the number of times opposing batters reached base via an error.
Garza's statistics are the most notable. His number of unearned runs matched his major league-leading number of batters who reached on errors (17 apiece), and his unearned runs total amounted to more than a quarter of the Chicago Cubs' entire team total of 66. Yes, the Cubs had one of the game's worst defenses a year ago, but as you can read here, they should improve in that regard this season. That Garza wasn't unraveled further by his defense last year is somewhat remarkable; it's a testament to how much hidden value he has in 2012 drafts.
Conversely, Moseley's numbers are troubling, considering he surrendered the seventh-most unearned runs (15) despite pitching for one of the game's 10 best defenses (by any measure) last season. Only eight baserunners reached via error against the right-hander in 2011, yet a quick scroll through his game logs reveals two instances in which he allowed three unearned runs after two outs had already been recorded in the inning. He had a 4.09 ERA in his final 14 starts last season, and coupled with his tendency to struggle recording "fourth outs," one might expect him to struggle in the ERA department this season.
This isn't exactly the same as the previous section, unearned runs, because unearned runs only occur as a direct result of an error or passed ball. Defense, however, is about a heck of a lot more than merely errors and passed balls. Range -- a better measure of defense's ability to convert balls in play into outs -- is every bit as important and arguably much, much more so. A defense can be accurate with the plays it makes, but that doesn't mean it's getting to the tougher plays, which, when made, result in greater team success.
Perhaps there's no greater example of that than last year's Philadelphia Phillies; they committed the second-fewest errors in baseball (74), but in terms of both UZR (Ultimate Zone Rating) and Defensive Runs Saved, they didn't crack the top 10 (they were 18th and 27th).
The Phillies' winter losses, however, might actually be their gains in the defensive department. Their worst defender from 2011, left fielder Raul Ibanez (minus-16 Defensive Runs Saved, minus-18.9 UZR), is now with the New York Yankees, and their second-worst defender, first baseman Ryan Howard (minus-9 DRS, minus-4.8 UZR), will miss a significant chunk of the season recovering from an Achilles' injury. With even slightly improved defense from either of those two positions, Phillies pitchers might benefit in terms of outs they might not have gotten last season, and those can have a profound impact upon a pitcher's ERA.
It's for that reason there are countless statistical tools out there to measure the impact of defense upon a pitcher's ERA. Two of my favorites are FIP (Fielder Independent Pitching), and xFIP (Expected FIP), both of which evaluate a pitcher's performance regardless of how his fielders performed behind him. The difference between the two: Whereas FIP includes a pitcher's home runs allowed in the formula -- (HR*13+(BB+HBP-IBB)*3-K*2)/IP, plus a league factor designed to put it on an ERA scale -- xFIP "normalizes" home runs, calculating instead based on the league average of home runs allowed per fly ball.
Pitchers whose FIP or xFIP vary greatly from their ERA might have experienced one of two things: Particularly strong or poor defense, or particularly good or bad luck. That's why, when evaluating ERAs, it's smart to take FIP/xFIP into account.
A caveat, however: Some pitchers do tend to be historically strong in FIP/xFIP despite it never resulting in a low ERA. Ricky Nolasco is a prime example. His 3.57 FIP ranked 30th and xFIP 14th among 135 qualified starters (per FanGraphs) over the past three seasons combined; his ERA during that span, however, ranks 18th-worst. It's smart, therefore, to consider a pitcher's FIP/xFIP history.
These were the 2011 leaders in both FIP and xFIP:
Greinke's xFIP dominance pops right off the page, doesn't it? If there's any fact that should temper 2012 expectations for the right-hander, it's that he has a history of success in both the FIP and xFIP categories, having ranked 20th and 21st in those departments in 2011. In addition, his Milwaukee Brewers signed Aramis Ramirez, potentially diminishing their defense, something that might result to another significant split between Greinke's ERA and FIP performance. Still, even on a team that was only so-so defensively in 2011, Greinke managed a 2.59 ERA after the All-Star break, so his rebound potential remains immense. We've projected him for a 3.39 ERA
which is perhaps even a bit conservative.
The FIP "whipping boy" entering 2012, meanwhile, is Tampa Bay Rays sophomore Hellickson, who had a whopping 4.44 FIP and 4.72 xFIP despite a 2.95 ERA. Hellickson has his critics, and if he repeats what was a surprisingly low 5.57 strikeouts-per-nine innings ratio last season, by all rights he might have a difficult time keeping his ERA much beneath 4.00. We've projected him for a 3.24 ERA this season accounting for some regression in the category, but that's also forecasting an increase in his strikeout rate, bringing his number closer to those during his minor league career. But unless Hellickson manages to approach a K-per-inning rate, his prospects of a second straight sub-three-ERA season are practically nil.
Also called "Strand Rate," this measures a pitcher's ability to prevent runners he allows to reach base from scoring. Along with BABIP and home run/fly ball percentage, Left-on-base Percentage is one of the more popular measures of a pitcher's luck. A typical, league-average strand rate is generally about 70-72 percent; those who have numbers significantly higher might have gotten a few lucky bounces, while those with numbers significantly lower might have been struck by terrible luck.
Not to say that strand rates are always a sign of good or bad luck. Brandon Morrow, for example, has a 67.2 left-on-base percentage the past two seasons combined, sixth-worst among 86 qualifiers. But there's an explanation: He has afforded only .219/.306/.339 triple-slash rates with the bases empty, a time during which he'd be pitching from the windup, compared to .273/.348/.459 with men on, when he'd be pitching from the stretch.
These were the leaders and trailers in strand rate in 2011:
There are both Garza and Greinke, again, on the "unlucky" side of the ledger, though they share the same space with the aforementioned Morrow. Garza is the most peculiar inclusion on the list, however. His 69.7 percent rate represented easily his lowest of any of his four full big league seasons, was substantially beneath his 75.5 percent number of 2010 and is noticeably lower than his 72.9 career rate in the category. Numbers like that have a tendency to even out over time, and to put Garza's into perspective, had he stranded runners at merely his career (72.9 percent) rather than 2011 (69.7 percent) rate, he'd have surrendered seven fewer runs
meaning his ERA would've been 3.00, not 3.32. Having a stronger defense behind him, as discussed above, can only help.
The aforementioned Hellickson and Vogelsong, meanwhile, fall on the "luckier" side of the ledger. Hellickson's left-on-base percentage provides another rationale behind his higher projected 2012 ERA, but in the case of Vogelsong, his projected ERA suffers considerably more because he lacks a skill Hellickson might yet develop: He's much less likely to boost his strikeout rate. Vogelsong is 34, after all. And considering the San Francisco Giants won't be doing the veteran any favors on the days they trot out Ryan Theriot at shortstop, there's a good chance Vogelsong's out-of-nowhere year might indeed be termed a "career year."
|Looking at Ryan Vogelsong's underlying statistics, owners shouldn't expect the 34-year-old to repeat his 2011 success.|
One of the least discussed, and infrequently measured, factors that impacts ERA, a starting pitcher's bullpen support can be critical to his success. Fantasy owners might see a starting pitcher's final stat line in the box score on any given night, but what they might not see was that the pitcher departed with two outs and the bases loaded, only to see his reliever surrender a grand slam in the very next at-bat. Three of those runs are charged to the starter, a significant hit to his ERA. Even a small handful of extra inherited runners allowed to score by the bullpen over the course of a given season can taint an otherwise quality performance by a starting pitcher, so always consider whether such a starter might have been let down by the men behind him when evaluating his ERA.
Baseball-Reference.com tallies "Bequeathed Runners," or runners that a starter leaves on base when departing; the chart below shows 2011's leaders in terms of percentage allowed to score, minimum 10 bequeathed runners:
Only one pitcher in baseball had his bullpen allow more of his bequeathed runners to score than Romero (11), and that was Pineiro (12). To think, Romero still managed a 2.92 ERA, meaning that if Toronto Blue Jays relievers had allowed half -- OK, let's say six, or half plus one -- of those bequeathed runners to cross the plate, his ERA would've been 2.72, which would've ranked him seventh in the majors. Heck, had Romero received comparable bullpen support to Blackburn, his ERA would've been 2.56. Don't underestimate the significance of that, because the 2011 Blue Jays bullpen had the majors' 10th-worst ERA (3.88), 12th-worst WHIP (1.33) and allowed the sixth-most homers (53). And since then, they've retooled by adding both Sergio Santos and Francisco Cordero.
Two New York Yankees right-handers, Garcia and Nova, meanwhile, ranked among the most fortunate starters when it came to bullpen support. Between them, Yankees relievers allowed only two of their 39 bequeathed runners to score, both off Nova. Yes, the Yankees' bullpen remains stout, with Mariano Rivera, David Robertson and Rafael Soriano at the back end, but neither Garcia nor Nova is guaranteed a rotation spot, let alone as much good fortune in the left-on-base category. It's for that reason we've projected both for ERA increases in 2012: Nova to 3.91, Garcia to 4.57.
This one's obvious. A pitcher's ERA is impacted by the venue he calls home, as bandbox ballparks lead to untimely home runs that inflate that number, while those who pitch in spacious venues can afford the occasional mistake without taking as much of a hit. Fantasy owners have long taken this into account, but for a specific measure of a ballpark's impact on ERA, Adjusted ERA (or ERA+) can be a handy evaluation tool. What it does is calculates a pitcher's ERA assuming a neutral ballpark; anything significantly over 100 is considered good, beneath 100 bad.
These were the 2011 leaders and trailers in ERA+:
The top five pitchers in terms of ERA and ERA+ were the same last season, though that's to be expected; the No. 5 finisher in ERA, Weaver (2.41), finished with an ERA more than a quarter-run lower than the No. 6 finisher, Vogelsong (2.71). In other words, the top five was an especially exclusive group in 2011.
At No. 6 on the ERA+ list, however, is a pitcher who escaped a hitting-friendly venue for a pitching-friendly one: Wilson. Though his 2.94 ERA ranked 25th, his 152 ERA+ demonstrates how valuable he was when ballpark effects are taken out of the equation. That explains his projected 3.07 ERA, which is actually lower than the 3.14 he had during his two seasons as a Texas Rangers starter.
Conversely, a pitcher such as Jeremy Guthrie might be in trouble as a result of his winter change of teams. Guthrie, whose 4.33 ERA was already below-average, had a 95 ERA+, 65th among 93 qualified starters. Now he's moving to Coors Field, meaning he might be little more than a matchups type in road games. Understandably, we have him projected for a 4.84 ERA.
Not that any one of these five influences should be expected to radically shift a pitcher's ERA in one direction or another, but they can, which is why instead of looking at that one category on the surface, you need consider what went into it. Remember, even for the most durable starting pitchers, one earned run can mean a 20th of a point in ERA or more; five earned runs can cost a pitcher as much as a quarter of a run in ERA (and often an ERA crown).
That's why, when evaluating pitchers, I place ERA lower on my priority list than categories like WHIP, strikeout and walk rates. That's not to say ERA is devoid of value -- you simply can't take it at face value.
Tristan H. Cockcroft is a fantasy baseball analyst for ESPN.com, a two-time champion of the League of Alternative Baseball Reality (LABR) experts league, and a 2011 FSWA award winner for Best Baseball Article on the Web. You can e-mail him here, or follow him on Twitter @SultanofStat.