Defense Independent Game Score
By Kyle Goings
A New Metric for Starting Pitching
A Brief History of Game Score
Tom Tango would update James' game score formula by lowering the "base," or starting score from 50 to 40, and by applying a penalty for homers allowed. Tango's score, known as Version 2, incorporates many of the same statistics as the original version, including hits & runs allowed. These scores can be found in the game logs at FanGraphs, and are adjusted so the average for both leagues is always 50. This is a major advantage over the Baseball Reference scores.
DIGS: The What & Why
Like Tango's game score, DIGS is adjusted so the average is 50 for both leagues. This is done by adjusting the base score, and can be calculated to fit any single season. DIGS, unlike any previous game score, is also park-adjusted. Obviously, a starter who tosses 7 innings, logs 7 strikeouts, and allows 1 HR in Coors Field should get more credit for his performance than a starter who posts the same line at Oracle Park. Now, using FanGraphs park factors, these two starts can be directly compared.
I created DIGS to get closer to a "true talent" score for single starts, but the metric does a lot more than that. Using what I call Advantage (similar to B-R's and FanGraphs' runs above average metric), I can convert DIGS into a WAR score. Like the base score, Advantage is adjusted so league WAR totals closely match those posted for each season on FanGraphs. And since the DIGS formula closely resembles the FIP formula, individual WAR scores are often very close to their fWAR counterparts. In fact, I compared WARs for over 125 starters in 2019, and the correlation (r^2) was .922. That's a very reliable trend!
September 2019: What do the numbers say?
These results were encouraging for me. Being really good (lots of Ks, few walks and batted balls) or really bad (few Ks, lots of walks, homers, and batted balls) isn't the norm in MLB. Most pitchers land somewhere near the league average, so you know the names on the extreme ends of the list really earned those scores. As you might imagine, Justin Verlander's no-hitter on September 1st, a 106 DIGS, was the highest score of the month, and likely one of the highest of the year.
September stats can be funky, especially for starting pitchers, but the top performers definitely separated themselves from the competition. Fifteen players earned at least 1 WAR, while only two eclipsed the 2 WAR mark. They were Gerrit Cole (2.4 WAR) & Jack Flaherty (2.1), who remain two of the hottest pitchers on the planet. You can see the best of the rest, as well as the month's worst, in the images below. Note that DIGS+, like other "plus stats" is league-adjusted so that 100 is average.
September DIGS: WAR Leaders
September DIGS: Bottom 15
September DIGS: DIGS+ Leaders
What's Next?
In the meantime, I've been using DIGS to track individual game logs, which is very time-consuming, as well as team profiles, Cy Young trackers, and 5-year averages. According to DIGS, Only 10 pitchers have been worth 20 WAR in the StatCast Era (2015-2019). Those pitchers are Max Scherzer, Chris Sale, Jacob deGrom, Justin Verlander, Gerrit Cole, Clayton Kershaw, Corey Kluber, Stephen Strasburg, Zack Greinke, and Carlos Carrasco. Of those elite starters, only Scherzer & Sale have eclipsed 30 WAR.
Can we use DIGS to find the next elite starter? Maybe we already have. According to FanGraphs, Cole & deGrom were the most valuable pitchers in their respective leagues, each just nudging past the 7 WAR threshold in 2019. But fWAR doesn't factor in balls in play. Through the lens of DIGS, Cole, with far more Ks and far fewer BIP, was easily the best pitcher in baseball at 8.8 WAR. That number is up from 6.5 WAR last year. Can the pending free agent continue his dominance next season?
My goal has always been to make game score a relevant measurement of pitcher performance, and I feel like this is my biggest step forward to date. There is still much work to be done, but I will share as I go. I hope this has been informative, and encourage you to reach out to me with questions. I'm on Twitter, @CoveChatter, and have created a Gmail account, spgamescore@gmail.com. Thanks for reading, and here's to an offseason of number-crunching!