A new version of Game Score, (Game Score version 2.0) is now available on the pitcher game log pages. It is listed under the heading GSv2 and is baselined to both season and league.
Thirty years ago, Bill James introduced us to Game Score, which he described as:
…a kind of garbage stat that I present not because it helps us understand anything in particular but because it is fun to play around with
One man’s trash is another man’s treasure. When you look at the original point system Bill devised, it all seemed reasonable enough. Give positive points for outs (innings) and a bonus for strikeouts, and negative points for hits, runs, and walks, with hits having more impact than walks. And start everyone at 50, since on a scale of 0 to 100, 50 is average.
A sidebar to the strikeout: also around thirty years ago, Bill introduced DER, defensive efficiency record, which is outs per ball in play, or the flip side of hits per ball in play, which is the foundation of DIPS. Bill therefore (almost) discovered the concept of DIPS but he didn’t realize it. It took Voros for the saber-world to notice, and for Bill to thank Voros publicly for the discovery. You can see in Game Score how the idea of DIPS was in Bill’s head, by the giving of the bonus point for the strikeout, over and above the regular out. We’ll get back to this in a second.
I think the reason that Bill considered this a “garbage” stat is that it wasn’t developed with a question in mind. It’s a way to organize a pitcher’s stat line so we can list things in an easy to list and understand manner. From that standpoint, it was likely an underdeveloped concept, a presentation that satisfied Bill’s needs at the time.
If you try to use Game Score and understand its components, you will see it breaks down in a few cases. Not enough to throw Game Score into the scrap heap, but just enough that for the stat to graduate from the garbage to the toolshed, it should be refined.
A few years ago, Bill emailed me that when he publishes his ideas, they are now orphans. It’s up to the rest of the world to adopt them… or not. Whether it’s David Smyth using Runs Created as an inspiration to launch Base Runs, or Bill’s one article discussion on comparing Clemens to Mattingly and Rice to Guidry that formed the eventual basis of WAR, Bill has given the world plenty of ideas that have been essentially Open Sourced.
Fixing the Gaps
That’s where I come in. Game Score has never been modified. I love the basic concept of Game Score, its simple presentation, and powerful message. We just need to make sure that it can hold up to scrutiny. Bill used Game Score for an article a couple of years back where he realized he needed to make adjustments for his particular research. You can read more about it in this piece I wrote, but the basic idea that starting everyone at 50 doesn’t work for starting pitchers who get knocked out early in the game for reasons of non-performance. Bill kept the core of Game Score but added adjustments which ended up making it messy. I offered a very clean and simple solution. And its genesis is replacement level: rather than starting everyone at 50, we start everyone at 40. You can read the article to learn more.
The other gap relates to the walk. I noted earlier how Bill gave a bonus point to strikeouts relative to the out, which is actually in keeping with DIPS. But the flip side of that is the walk, and how its value should actually not be half the value of a hit, but equal to the value of a non-HR hit. Now, to be fair, this idea only works if we consider the third gap: the non-use of a HR.
Bill’s original idea was based on using the traditional pitcher line. But if we deviate that in the slightest, and just include the HR, this allows us to better compare the walk and the non-HR hit.
Game Score 2.0
The end result is this simple formula:
(Note: The K is double-counted, 2 points for the out, and 1 extra. The HR is double-counted, 2 for the hit, and the 6 extra.)
It’s pretty straightforward, owing a great deal to Bill James, but shaped by Pete Palmer and Voros McCracken. You can read the link for more background. The three main areas of improvement is how it starts off each start at 40, not 50, how it better handles the walk, and that it uses the HR.
You can also align it to exactly 50 as league average by setting the constant for each year. In 2015, you’d use 38 instead of 40. Here are therefore the 10 best starts of 2015:
109 Max Scherzer 2015-10-03
104 Max Scherzer 2015-06-14
103 Chris Heston 2015-06-09
102 Max Scherzer 2015-06-20
102 Jake Arrieta 2015-08-30
102 Corey Kluber 2015-05-13
101 Clayton Kershaw 2015-09-29
101 Carlos Carrasco 2015-09-25
101 Cole Hamels 2015-07-25
99 Madison Bumgarner 2015-09-12
Game Scores actually have a fairly linear relationship to wins. Obviously, at the most extreme it’ll breakdown, but it does a pretty good job overall to represent a pitcher that averages a Game Score of 65 will win 65% of the time.
David has implemented Game Score on the individual pitcher pages, which is a terrific addition to the site.
I’m a little confused. Scherzer’s 9 IP / 17 K game got 109, but Kershaw’s 9 IP / 15 K in 2014 got 105, 4 fewer, but if K’s are worth 1, I would expect it to get 2 fewer.
You adjust the constant in the formula (40) so that a final score of 50 would be average for that year. As stated in the article, in 2015 the constant was 38 instead of 40. In 2014, the constant would have had to be 36 to give Kershaw a 105 for that game.