Statistics, at their best, describe, explain, and illustrate what is being observed. They provide a platform of understanding and can be used to logically set expectations. Earned Run Average does none of those things; at least not in today’s game.
Yes, back at the turn of the century starting pitchers routinely finished games. So a fan buying a ticket to see Walter Johnson in 1910 could say he gave up a little over a run each game and that anybody scoring 2 or more got to him. Most importantly for that era, the gamblers knew where to place their bets. But that was over a century ago.
The days of pitchers completing nine innings in one outing is over. The occurrence so rare as to render ERA a generally useless statistic with none of the properties listed above.
And, in an ever crowded sports and entertainment landscape, it creates a slight barrier to the casual fan enjoyment of the game.
I can flip on a basketball game and clearly see that Lebron James is averaging 25.9 points, 7.4 rebounds, and 10.6 assists per game and this his usual game time is just under 27 minutes. I can use this information to understand he’s a good player and that if he gets more than 26 points he’s having a good game and fewer than 26 points is an off night.
Notice the stats don’t say how many points, assists, and rebounds he would get if he hypothetically played all 48 minutes. Nor are the statistics of the bench guy who averages 12 minutes a game.
Actual production is listed and the statistics are easy to digest, paint a picture of what I’m watching, and give me the basis to both predict how a player might perform and analyze that performance.
Now, let’s take a look at baseball. If I want to peruse the statistics of the top free agent pitchers on the market, I can clearly see that Gerrit Cole gave up 2.5 earned runs for every 9 innings he pitched. I can also see that he pitched 9 innings in a game exactly zero times this year and has a whopping 2 complete games over a 7 year career and 192 games.
So, about 1% of the time, evaluating Gerrit Cole over 9 full innings actually describes what he is likely to do: give up 2.5 runs. About 1% of the time I can say he had a normal game by going the full 9. About 1% of the time his primary statistic actually tells me if he had a good game, a bad game, or something in between. 1%.
For relievers, that number is 0%. Yet we still show ERA as the first statistic on the screen with a lefty specialist comes in to face his 1 or 2 batters to end an inning. It might take him 16 appearances to cobble together 9 innings, but his ERA is front and center for all to see.
This statistic tells me nothing about the performance of a reliever. It give me no description of his talent, no idea what to expect, nor any way of evaluating if I’m seeing a good, bad, or normal appearance from him. ERA is absolutely worthless in this case.
Now let’s consider explaining ERA to a new fan compared to using an NBA style model: 2.5 is how many runs Gerrit Cole would give up over 9 innings if he ever went 9 innings, which he almost never does. Lebron James scores about 26 points per game and plays about 27 minutes. The better way is pretty evident.
It is time for baseball to switch to a more accessible model for pitching statistics. Baseball-Reference.com and the like should definitely keep ERA around for traditionalists, but I’d love to flip in my TV and see something meaningful, like this:
Gerrit Cole 1.79 runs per game, 9.88 k’s per game, 1.45 bb per game, 6.43 innings per game.
Now, as a fan, I know Cole typically pitched to just about the seventh inning stretch on most nights. He walked a guy or two but struck out over one guy per inning. I know that if the Angels got 2 runs off him, they did a little better than most.
This line tells me a lot of things ERA does not, all of them useful and without the need to do long division in my head.
Let’s follow this up with a reliever. Last year Ty Buttrey appeared in 72 games for the Angels and pitched a total of 72.1 innings. In short, it took him 9 appearances on average to accrue the 9 innings used as a baseline for ERA.
Ty Buttrey: 0.44 runs per game, 1.17 k per game, 0.32 bb per game, 1 inning per game.
This tells me that Ty gave up a run about every other game, struck out a guy pretty much every game, and walked a guy every third game. But, most importantly, it told me all of his statistics were small because his average game was a single inning.
WHIP is probably the best statistic for relievers, but Victor Rojas saying “Ty Buttrey entering the game He gives up a little less than half a run per appearance and strikes out one hitter per inning” sounds more accurate, more descriptive than giving me the average number of runs Ty gave up over a 9 appearance set (3.98).
I love baseball and its traditions. I understand ERA and why it has been used. But I’ve had to explain it to enough people to know that it is simply no longer viable, almost to the point of being obtuse.
The next person I’m going to really explain the game to is my son. When he’s a child and can easily look up how many points the next Laker great scores in a night, but needs daddy to do the math to tell him whether Griffin Canning has good or bad statistics, there’s a problem.
Let’s move to the modern era of the game. We have WAR, OPS+, wRC+, and more acronyms than a military base to describe the offensive side of the game yet baseball clings to a metric from a bygone century to describe pitching.
It is time to ditch ERA as a primary statistic and move on to game averages. They are not only more descriptive, but easier to digest for the casual and young fan; exactly the audience baseball is struggling to convert.