Opening Day 2018 was, for the most part, a resounding success. Freddie Freeman and Ozzie Albies homered, and we will long remember Nick Markakis‘ walk-off, game ending home run. Another bright spot in the game was starting pitcher Julio Teheran, an encouraging performance considering how his first year in SunTrust Park went (it went poorly).
Through 5 innings, Teheran had allowed a lone run. In the 6th, Cesar Hernandez led off with a solo home run. That’s life with Teheran at this point, but pitching in the 6th with a 2 on the scoreboard means you’ve done a fine job as a starter. Teheran got the next two batters out but then ran into trouble. He hit Rhys Hoskins with a pitch, and Hoskins then stole 2nd. He walked Aaron Altherr, to give the Phillies 2 runners with 2 outs.
At this point, Brian Snitker turned to Rex Brothers. This move naturally maddened fans. Brothers was unreliable last season and struggled through much of the spring. He was one of the last additions to the MLB roster, and by most accounts, the #3 lefty in the bullpen (after AJ Minter and Sam Freeman). Luckily for Snitker, this post isn’t going to focus on the reasoning (or lack thereof) of bringing in your worst relievers in the most important moments of the game. I’m just setting the stage here.
Brothers walked JP Crawford to load the bases. He then walked Maikel Franco to bring home Philadelphia’s 3rd run of the game. Snitker replaced Brothers with Dan Winkler, who allowed a 2-run single to Andrew Knapp and then struck out Aaron Nola to end the inning.
A discussion on the Outfield Fly Rule Facebook Group got me thinking about the way we assign credit (or blame) for runs. Julio Teheran, as a result of the way this inning played out, currently has a 6.35 ERA. That 6.35 doesn’t really do a great job of reflecting the job Teheran did on the mound Thursday. With that driving me, I searched for a better way.
No FIP’n Way
FIP, which stands for Fielding Independent Pitching, is an excellent resource. Our own Dan Horton recently wrote about it in layman’s terms as part of his sabermetrics 101 series. I suggest you check it out. FIP generally does a better job than ERA in telling you just how good a pitcher really is, and it is notably more predictive of future ERA than ERA itself. And if you’re into FIP, Baseball Prospectus’ DRA (Deserved Run Average) is like FIP on steroids. It’s even more predictive, and it accounts for all kinds of factors.
Julio Teheran’s FIP is 6.25, mostly because he issued more free passes (4) than strikeouts (3) and surrendered a home run. While I said FIP is predictive, single game FIP isn’t, so don’t worry. While FIP generally does a good job over a long haul telling you about a pitcher’s quality, it doesn’t tell you so much about exactly what happened. FIP can’t really help in terms of illustrating events that happened on the field of play. It’s invaluable in making strategic decisions going forward, but that’s where its value lies.
What I’m looking for is a better way of representing what happened. Something better than ERA, but maybe not as rooted in the peripheral stats which are more predictive than they are descriptive.
The Problem with Mound Runs
One way to distribute run responsibility is the simplest way – the runs that scored when the guy was on the mound. Two runs crossed the plate when Teheran was on the mound, giving him a… Mound Run Average of 3.18. 5 2/3 innings, 2 runs. It’s the easiest to explain to someone, and it’s what most of us as children probably imagined ERA to be until we understood the concept of unearned runs and runners responsible for. It’s the rate that runs score when this guy is pitching.
The problem is that it lets pitchers like Teheran off the hook when they exit a game mid-inning with runners aboard. Teheran does deserve some of the blame for Hoskins and Altherr coming around to score. After all, he put them on base to begin with.
The Problem with Earned Runs
Teherand may have put Hoskins and Altherr on base, but he still isn’t the one who let them come around to score. And ERA treats him like he is. Those runs are completely Teheran’s fault in the ERA calculations. Dan Winkler allowed the single that brought in Altherr, but he now sports a 0.00 ERA against Teheran’s 6.35.
Therein lies the problem with ERA as a descriptive statistic. It is statistically biased against the man who allowed the first base, ignoring the man who allowed the runner to progress the next four.
The Compromise
Here’s how I’ve bridged the two.
When a pitching change happens mid-inning with runners on base, I will apply run expectancy to the equation to balance things out. Run Expectancy sounds like a fancy sabermetric term, but it’s really just quite simple. It’s the average number of runs that score in an inning given the current set of circumstances.
When Hoskins stepped to the plate with nobody on and two out, Philadelphia’s run expectancy for the remainder of the inning was 0.11. When Teheran hit him, that more than doubled to 0.23, meaning that the average MLB team, with 2 outs and a runner on first, will score about 0.23 runs, or once every 4 opportunities. When Hoskins stole second, that improved to 0.33 runs expected. When Altherr walked to put runners at 1st and 2nd, it moved to 0.44 runs expected.
You’re a baseball fan, and this isn’t re-writing the way you understand the game, of course. It’s just putting it into numerical terms. You knew that Philly was much more likely to score after Altherr drew ball four than they were when Hoskins stepped to the plate. Again, this is just to illustrate how we’re going to adjust the runs.
Because Teheran left his reliever with a stressful situation, where the average team gives up 0.44 runs, we’ll dock Teheran for those 0.44 runs. If the reliever gets out of the jam unscathed, he’ll get -0.44 adjusted runs. If he gives up one run, we’ll deduct 0.44, as those are already credited to Teheran. Teheran ends the game with 2.44 adjusted runs allowed (2 mound runs + 0.44 bequeathed runs).
Brothers walked Crawford, increasing the run expectancy to 0.78. He walked Franco, but as that just duplicated the last situation, the expectancy remained 0.78. Remember, it’s not your total inning expectancy, it’s just the expectancy from that moment until the end of the inning. So, Brothers is credited for 1 mound run from walking in Hoskins, 0.78 runs for what he is bequeathing to the next guy, and receives a 0.44 deduction from what Teheran was responsible for. That comes out to 1.34 adjusted runs for Brothers.
Winkler gave up a 2-run single, giving him 2 mound runs onto his tally. But he allowed nothing else, and he came into a situation where Brothers was responsible for 0.78 runs, so Winkler gets 1.22 adjusted runs allowed.
If you’ll notice, the adjusted runs still add up to 5. Each run is accounted for, no more, no less. However, by breaking up the runs in a more deserving manner, we can get run averages that better indicate how each pitched.
Here are the final tallies. Because Brothers didn’t register an out, his run averages are infinite at the moment, I’ve listed how it’ll look IF he retires the next batter he faces:
Earned Runs – Teheran (4), Brothers (1), Winkler (0)
Adjusted Runs – Teheran (2.44), Brothers (1.34), Winkler (1.22)
ERA – Teheran (6.35), Brothers (27.00), Winkler (0.00)
The improved run average – Teheran (3.88), Brothers (36.18), Winkler (8.24)
I just need something good to call it.
Well, it’s like ERA but better. So, Better Run Average? The name should, I suppose, say why it’s a better run average. It’s because it values expectancy, of course. The last thing we need is another hard to remember acronym, so let’s see.
Better Run Average that Values Expectancy.
BRAVE.
If Opening Day is any indication, BRAVE may be a better descriptor of what happened than ERA. I will try to revisit it from time to time and update you as the season goes on. For now, I’m a fan. Maybe I’m biased because of the last half hour I spent on it, but I’m a fan regardless. Sorry, ERA and FIP, you just can’t convince me that Teheran pitched more like a mid-6 pitcher yesterday than a sub-4.
Leave a Reply