Boyd's World-> Breadcrumbs Back to Omaha-> RBOA -- A New Pitcher's Metric About the author, Boyd Nation

RBOA -- A New Pitcher's Metric

Publication Date: December 11, 2001

Problems with the First Try

A little while back, I took a first shot at finding a smarter metric for college pitchers, something I called AERA, for Adjusted Earned Run Average. As I said at the time, I wasn't all that impressed with the results, and I think I've come up with something better.

The biggest problem with AERA was that it was based on a pitcher's team's entire strength of schedule. That's OK for the hitting equivalent, AOPS, because hitters, especially top hitters, tend to spread their at bats over the entire schedule. For pitchers, though, there's not that even distribution, and that tends to get really concentrated with the top pitchers -- the staff ace on a really good team is going to tend to face only conference opponents for most of the season, which in most cases is a bit tougher than the non-conference opponents.

Runs below Opponent Average

I've now come up with something that I like quite a bit better. Rather than try to deal with total numbers at the beginning, I think looking at individual appearances is probably the place to begin. The metric that I want to use now, therefore, is something I'm calling Runs below Opponent Average (RBOA). It's a simple idea (in fact, it's simple enough that I'm willing to bet I'm not the first to come up with it, but I can't find a record that anyone has gone public with it before).

RBOA is a counting stat, which just means that it's a cumulative measure like home runs or wins rather than a rate stat like ERA or winning percentage. For each appearance (there are problems with relief appearances because of inherited runner problems, but we're mostly dealing with starters here), you take the average number of runs that the opponent scored over the course of the season (I suppose that, if you're doing this midyear, you'd have to use runs per game to date), compute from that how many they would be expected to have scored in the innings that the pitcher was in, and subtract the number of actual runs allowed to get RBOA for that appearance.

Here's an example to make things clearer: Against Houston this year, Mark Prior gave up 3 runs in 8 innings. Houston average 5.61 runs per game this year. In 8 innings against an average pitcher, you'd expect them to score 4.99 runs (5.61 * 8 / 9 = 4.99). Therefore, Prior saved 1.99 runs (4.99 - 3 = 1.99) in that game, for an RBOA for that appearance of 1.99.

Obviously, this stat is a bit more labor-intensive to track, since you need to have individual game results to compute it. I haven't tried to do a complete leader board, but I did do the following legwork to verify that I like the way it shakes out. I took the NCAA top 10 in ERA for last season. I then eliminated the three of them for whom I couldn't find box scores at this point (unfortunately, that eliminated Dewon Brazelton, who I was really curious about) and computed RBOA for the rest:

Pitcher                  Team                      RBOA

Prior, Mark              Southern California       70.70
Justin Pope              Central Florida           52.11
Aaron Heilman            Notre Dame                50.49
Jason Arnold             Central Florida           45.70
Todd Pennington          Southeast Missouri State  32.78
Kyle Johnson             St. Bonaventure           31.54
Joe Engel                Pittsburgh                17.36

Now, if you look back at the AERA listing, I think you'll find this ordering a bit more to your liking -- obviously, matching our preconceived notions isn't the goal of the exercise, but the smell test does have some validity. For those who wonder why there are those who claim that Prior had the best pitching season in college history this year (I'm not one of them, just because I don't have the historical knowledge to feel comfortable with the claim; I certainly can't name a better year), the following sequence of games shows why:

Opponent         IP   R

Washington       9.0  1
Arizona          9.0  0
San Diego State  5.0  2
Arizona State    8.0  1
California       9.0  0
Stanford         9.0  1
UCLA             9.0  0

That's an absolutely phenomenal accomplishment for college baseball in the offensive environment of 2001.

I really like this metric, but I'll go ahead and point out some potential objections to it in the spirit of fairness:

Cross-Platform Compatibility

One really nice thing about RBOA is that I can't see any reason that it shouldn't work at any level of baseball, from Little League to MLB. Obviously, season length (and game length, for that matter) means that you can't really compare totals across levels, but I'm curious about what the numbers would look like for other levels. I'll also try to honor requests to compute other pitchers (especially if you send me the raw appearance data, but if you can't do that still ask); just let me know who you're curious about.

Boyd's World-> Breadcrumbs Back to Omaha-> RBOA -- A New Pitcher's Metric About the author, Boyd Nation