A Proposal To Revise The ERA Statistic

 

The way Earned Run Average is calculated is unfair for pitchers.  This is a proposal – although not perfect – that strives to make the metric more accurate. 

ERA, as defined by Baseball Prospectus is as follows:

Earned runs, divided by innings pitched, multiplied by nine.

The resulting number gives an indication of how many runs, on average, a pitcher will allow over the course of 9 innings pitched – a full game.  It gets dicey when relief pitchers are thrown into the mix.  If a starting pitcher departs the game with runners left on base and a reliever allows those runners to score, all of those runs are charged to the starting pitcher.  It doesn’t seem entirely fair, does it?

There have been efforts to better measure ERA.  The first is ERA+.  From The Hardball Times:

ERA+ measured against the league average, and adjusted for ballpark factors. An ERA+ over 100 is better than average, less than 100 is below average. The specific formula divides the league ERA by the pitcher’s ERA (and adjusts for ballpark). So an ERA+ of 125, for instance, means that the league ERA was 25% higher than the pitcher’s ERA (which means that the pitcher’s ERA was 80% of the league ERA). Careful with those ratios.

Although an improvement on ERA, since ERA+ is able to judge a pitcher’s effectiveness in relation to his peers, it still fails to address the inherent inequality in how ERA is calculated.  Another ERA-based stat is FIP (Fielding Independent Pitching).  Again, from The Hardball Times:

a measure of all those things for which a pitcher is specifically responsible. The formula is (HR*13+(BB+HBP-IBB)*3-K*2)/IP, plus a league-specific factor (usually around 3.2) to round out the number to an equivalent ERA number. FIP helps you understand how well a pitcher pitched, regardless of how well his fielders fielded.

This is nice… but doesn’t really deal with the runs a pitcher allows.  It is an “equivalent” statistic.  Kind of like aspartame to sugar.  They both aim to do the same thing but you can tell the difference.  As well, FIP not necessarily based in reality.  By that I mean that FIP calculates how well a pitcher should have pitched, now how well he actually did pitch.  For example, Brandon Morrow’s FIP is less than Ricky Romero’s.  That said,  I know who I want starting game 1 of a playoff series (hint: it’s not Brandon Morrow).  That is, after all, why we play the games. An aside:  Morrow has a better WAR (3.2) than Ricky Romero (2.8).  Lol wut.

My proposal is to revise how the ERA statistic is calculated by dividing runs into quarters based on where a relief pitcher’s inherited runners are stationed.  For example:  if a starter leaves the game with a runner on third base and the reliever allows that runner to score, the starting pitcher is responsible for 0.75 runs and the reliever is charged with 0.25.  If the runner is on 2B, the responsibility is divided 50/50: 0.50 runs for the starter, 0.50 runs for the reliever.  If the runner is on 1B, 0.75 runs are charged to the reliever and 0.25 to the starter.

If there are multiple men on base, the same formula applies.  For example, if a starter leaves the game with a runner on 3B and a runner on 2B and the reliever allows both to score, the 2 runs are divided up as such: 1.25 to the starting pitcher (0.75 for the runner on 3B + 0.50 for the runner on 2B) and 0.75 to the relief pitcher (0.25 for the runner on 3B + 0.50 for the runner on 2B).

Earned Run Averages are artificially high for starting pitchers and (sometimes, depending on the role) artificially low for relievers.  This would balance the scales.  That said, this presents some issues.  Historical ERAs would have to be re-calculated and I’m not sure there is a way to do that without having very detailed game data.  I’m sure there are much smarter people than me who would be able to take on the challenge and succeed.

What do you think?  Is this a good plan? Too simplistic? Or am I missing the mark completely?  Please let me know in the comments section below.

has written for Mopupduty.com since 2006. Follow Callum on Twitter, LinkedIn and Instagram (@callumhughson)

  • gabriel

    It’s both too simplistic and too complicated. Too simplistic since it doesn’t take into account the number of outs. Allowing a runner on 2nd to score with two outs is a great deal different than if you’ve inherited him with none out. Then there’s the problem of multiple relievers- say a situational lefty comes in with a man on first, gives up a single to move the runner to third, and the 2nd reliever allows him to score – how do you divide that responsibility?

    It’s too complex insofar as it changes ERA from being a fairly easy-to-understand stat (albeit imperfect) to a very complex one, while the principal inequities remain (the inability of errors to accurately reflect defensive responsibility for runs scored).

    • http://mopupduty.com/index.php/about-mop-up-duty/authors_callum_hughson/ Callum Hughson

      I admit the number of outs and situational relievers is not something I took into consideration, and I should have.

      That said, I’m not sure how I would address it while keeping in the spirit of trying to remedy the inequalites inherent in the ERA statistic. Do you have any ideas?

      • gabriel

        The simplest measure is to attribute to the starter the run expectancy of the runners he leaves. Then give the relievers the change in run expectancy over their innings. ERA stays the same for innings pitched with just one pitcher, and becomes a run expectency matrix for other innings.

        • http://mopupduty.com/index.php/about-mop-up-duty/authors_callum_hughson/ Callum Hughson

          Excellent. Does this exist in any format? If not, why not? It is brilliant.

          Also, does run expectancy change from year-to-year? I imagine run expectancy would be different in 2010 than it would in 1995 and 1965.

        • http://www.chriskol.com Chris K

          Something when run expectancy would be awesome.

          Reading this at 3am and all my mind can come up with is the starter is responsible for pitchers who are already in scoring position (as it seems hardly fair to blame a reliever for one hit). This would have interesting effects if someone was on 1B when a home run is hit…but I think given that the reliever just gave up an HR, it might be fair to charge him the inherited runner.

          Simply, yes. Stupid, maybe. But this is all I got at 3am haha.

        • http://mopupduty.com/index.php/about-mop-up-duty/authors_callum_hughson/ Callum Hughson

          If we wanted to get really fine with it, we could base a stat based on an individual pitcher’s run expectancy. As in, if Roy Halladay allows 2 runners on with none out, he is less likely than Jo-Jo Reyes to allow those runs to score in the same situation. Therefore Halladay shouldn’t be penalized with a “league average” run expectancy.

          Another idea would be to base run expectancy on a pitcher’s track record in each inning. i.e. a pitcher is x% likely to allow runners to score in the 3rd inning and x% likely to allow runners score in the 7th. Depending on which inning the pitcher is pulled in, he is charged with y amount of earned runs based on the likelihood he would have given up those runs based on his track record in that inning in similar situations.

  • mike in boston

    good article, and i like the sharing runs approach.

    Some starters benefit from great bullpens, or great managers, and some starters have lousy luck. There should be a way of leveling that out. My preferred solution is Relief Independent Pitching, or RIP.

    like FIP, you could have an ERA that is adjusted for the league average when it comes to stranding inherited runners.

  • http://mopupduty.com matthias

    I hate being that guy but BP has a stat along these lines. It’s their FAIR-RA Start (and leads into their FRA statistic).

    From Baseball Prospectus (FAIR_RA_START):

    Fair run average for innings by a starting pitcher. “Fair runs” differ from traditional assignment of runs in that a pitcher who leaves the game is charged with the expected run value of any bequeathed runners left on base, whether or not they eventually score.

    2011 leaders: https://www.baseballprospectus.com/sortable/index.php?cid=1045911

    • http://mopupduty.com/index.php/about-mop-up-duty/authors_callum_hughson/ Callum Hughson

      Out of curiosity, what is the expected run value of bequeathed runners left on base?

      I like it, but again my personal bias is to be “uncomfortable” with stats that calculate what should happen instead of what actually happens.

      Does this expected value take into consideration the number of outs in the inning when the pitcher departs, like Gabriel mentioned above?

      Looking at the leaderboard (with a decent minimum innings pitched filter) it is skewed towards closers since it is rare that they are pulled with runners left on base.

      Increase the filter to 100IP and you’ve got all the studs… and a few outliers. Cory Luebke? 5th best in all of baseball? Madison Bumgarner #8? Interesting to see that FIP darling Brandon McCarthy is #7 on the list.

      Ted Lilly clocks in right before CJ Wilson. Hmmmmm that can’t be right. The ace of the Blue Jays staff, Ricky Romero, checks in at #85. Behind such pitchers as Jason Marquis, Freddy Garcia, Livan Hernandez, Jeff Karstens and Javier Vazquez. Perhaps FRA is subject to managerial bias – some pitchers leave the game following the completion of an inning, while others are victims of a manager’s twitchy trigger finger once they appear to get into trouble.

      • http://mopupduty.com matthias

        Upon further review, this FRA isn’t based upon inherited runners. It’s straight up FRA (estimated runs based upon BIP types).

        One thing to notice re: managerial bias is ‘stud’ pitchers getting the chance to finish things out. You’ll notice the majority of the top 10 (minus Luebke, who spent half the season in the bullpen) have fewer BEQ rnrs than pitchers further down the list. Romero has 22 BEQrnrs, a guy like McCarthy has 12.

  • http://www.passion4baseball.blogspot.com WilliaminMaine

    While we’re at it, can we stop giving the pitcher an unearned run from his own error? If the pitcher boots the ball or throws it away, he certainly “earned” that run, didn’t he?

    • http://mopupduty.com/index.php/about-mop-up-duty/authors_callum_hughson/ Callum Hughson

      Yes yes yes!

  • jimmyright

    how about making those runners unearned runs to the starting pitcher when he leaves the game, and thus not making them affect his ERA? then you could make Inherited Runners Stranded a standard stat for relievers to get an idea of how clutch they are

  • http://deathbytrolley.wordpress.com Ron Brown

    I LOVE this idea. I’m also really enjoying this blog. I’ve been an MLB fan for years, and over this time have been more stats and dollars focused than most other baseball fans I had ever spoken to. Having just finished reading Moneyball – also watched the movie this weekend – I’m totally fiending on Sabermetrics!

    Keep up the good work!

  • http://deathbytrolley.wordpress.com Ron Brown

    What do you think of this idea: changing WHIP to include HBP. Whether it is by hit, walk or a hit batter, the pitcher’s pitching has led to a runner on base. I don’t see why HBP would be left out…

  • Matt

    What if the starter leaves with a runner on second, a first reliever comes in and the next batter reaches on a fielder’s choice putting the runner out at third, and then a second reliever comes in. If the second reliever allows the runner to score, do we charge 0.5 runs to the starter, -0.25 runs to the first reliever, and 0.75 runs to the second reliever. This could make negative ERA’s possible!