Evening all. I hope your Thanksgiving was a good one. The Winter Meetings aren’t too far off, so let’s hope the action around MLB starts to heat up a bit next week. Until then, how about a little more Giants roster talk? I’ll even throw a little bit of my own research into this one.
Caveat: I know that my posts can get a little long-winded at times, and I’ve got a lot I want to discuss here. So I promise to keep keep the exhaustively long paragraphs to a minimum (as best I know how, anyway).
For those who’ve followed this blog regularly over the past couple years, I probably don’t give the impression of a fan/blogger who likes to analyze stats. Very few of my posts are statistically focused, as I would say I tend to focus more on the “human” element of the game.
But that doesn’t mean I don’t love to pour through the numbers. I simply don’t have as much time to post anymore, and when I do have the time, I don’t feel like using it up copying stats from various open website tabs into my post. There are so many reference sites for professional baseball statistics out there that most knowledgable fans can find (and interpret) them on their own. When I’m blogging, I like to focus on the words.
For those reasons, I rarely share much of my statistical analysis on this site. Plus, I’m not Sabermetrically-oriented, so most of my “finds” are probably very surface level compared to the vast baseball research and number tweaking that is shared on the World Wide Web.
Keeping all of that in mind, I wanted to share with you a statistic that I’ve fallen in love with this offseason: the Game Score. You will find Game Score (GS) in almost every recent MLB and MiLB box score, and can easily calculate it for any game – whether professional or amateur – in modern baseball history. And still, it’s not a stat often mentioned, analyzed, or discussed.
I’m not going to spend unnecessary time explaining how GS is calculated (you can find the simple to use formula HERE), but here’s the Reader’s Digest overview: Though not very “Sabermetric,” it was actually created by the king of advanced stats, Bill James in 1988. It’s somewhat like a fantasy baseball idea in that it calculates a score for starting pitchers in every game played, based on their IP, K, H, BB and runs allowed (both R and ER). Each pitcher begins the game with 50 points, and the score is adjusted from there depending on performance. 114 is a perfect score (perfect game with 27 K).
Though GS is rarely mentioned outside of single game analysis (after Max Scherzer logged a 104 during his late-season no-hitter, for example), it can actually be quite practical for studying a pitcher’s season and career performances as well.
Think about it: though there are now sites that track every pitch thrown by every pitcher, in every game of every season (and gobs of insightful information that comes from such tracking), stats like IP, ERA, H/9, K/9, and BB/9 are still very much a part of pitcher analysis. Game Score tracks all of those things, and actually combines them into one number that can directly be compared to other pitching performances.
Game Score is a counting stat, so pitchers who throw more innings, strike out more hitters, and limit their walks are obviously going to look better through the lens of GS… but unless those same pitchers are giving up gobs of hits and homeruns, isn’t that generally the model for a successful pitcher anyway? My point is, while I’m sure a counting stat that measures basic categories and uses a simple calculation to give a score for a single pitching performance may be seen as inadequate in this day of endless number crunching and overstimulation of advanced metrics, I find GS very convenient for dissecting a pitcher’s total body of work.
I’m trying not to ramble here, I promise. So let’s move on. For reference, you can find the top individual Game Scores from 2015 HERE, as well as some additional reading on the topic HERE and HERE. Both of those authors made their own arguments for using GS as a way to measure more than just a single game performance.
I found the SABR.org piece (published in the Summer 2010 Baseball Research Journal) especially insightful, as the author showed the correlation between GS performances and team win probability for the 2007 season. Most interesting to me was the proposition of using GS as a way to calculate pitcher wins and losses, though I’m sure the baseline numbers Jeff used might need some tweaking to match the pitching-dominant era baseball has entered into since that article.
Here’s a brief explanation of this proposal: for each game a pitcher starts, he earns a W, L, or ND just as is common practice today. However, the decision is based on the pitcher’s Game Score rather than the team’s outcome. Using the theory presented by the author (which was tied to win% for the entire 2007 season), a GS of 55 or higher would earn the pitcher a win; A GS of 43 or lower would result in a loss; Anything between 44 and 54 would result in a no-decision. In his proposal, the author split added the no-decisions to the pitcher’s W-L record (splitting the ND’s in half, or adding the “extra” to the wins for an odd number of ND’s).
After playing with the numbers (and creating an extensive spreadsheet including every pitcher who started more than 15 games this season), I personally would not include ND in the Game Score W-L record. Including them created for a watering-down of 20-game winners (44 to be exact). But it also made for some pretty cool old-school looking records, with Clayton Kershaw and Zack Greinke winning 29 games apiece, and Jake Arrieta logging 28 wins.
Ultimately, idea here is that you can use Game Score to judge a pitcher’s W-L record solely on his own performance… which is a pretty cool idea, if you ask me. Frankly, the game seems to be headed that way already, as most knowledgable fans pay little to no attention to W-L records anymore anyway (unless there is a late-season surge for 20 wins, which is becoming more and more rare). Pitchers simply aren’t judged on the number of games they win anymore, nor should they be.
I’m sure you’re ready for an example by now, so let’s look at the 2015 Giants rotation, from a Game Score & Game Score W-L perspective.
Note: an “average GS” of 52 is average, 55-60 is very good, and 60+ is elite.
Madison Bumgarner: 32 GP, 22-4 (6 ND), 61.7 GSavg
Chris Heston: 31 GP, 14-10 (7 ND), 51.6 GSavg
Jake Peavy: 19 GP, 9 -4 (6 ND), 53.8 GSavg
Ryan Vogelsong: 22 GP, 9-9 (4 ND), 49.2 GSavg
Tim Hudson: 22 GP, 7-10 (5 ND), 47.7 GSavg
Tim Lincecum: 15 GP, 5-6 (4 ND), 49.7 GSavg
Matt Cain: 11 GP, 4-7 (0 ND), 44.5 GSavg
Mike Leake: 9 GP, 3-4 (2 ND), 51. 8 GSavg
It doesn’t take advanced metrics to tell you the Giants need some major rotation upgrades this winter. Outside of Bumgarner (who was brilliant) and Peavy (who missed 12 starts), everyone else who started multiple games for the Giants in 2015 was below average by a Game Score standpoint.
I am about out of time for tonight, but I do want to compare potential free agent and trade options through the lens of Game Score and GS W-L (while explaining a little more about the research I’ve been doing). For now, know that tops on my wish list is Zack Greinke, the #3 pitcher in MLB this year with an average GS of 67.1 (Kershaw was #1 at 68, Arrieta #2 at 67.2). By Game Score W-L, Greinke was 27-1, with 4 ND. Yeah, I’d say the Giants could use him in the rotation.
We’ll chat again soon. In the meantime, try calculating a pitcher’s Game Score Win-Loss record for yourself. Just pick a pitcher, go to his Baseball-Reference page, click on his “game logs” for an individual season, and sort the chart by “GSc.” Give him a win for every start of 55 or better, a loss for anything 43 or worse, and a no decision for anything between 44 and 54. You can find his average Game Score for each season by clicking on the “More Stats” link from the player’s homepage, and scrolling down until you find the “Starting Pitcher” data. It’s pretty fun to compare historical seasons from some of the game’s greats.
Let me know what you think. As always, thanks for reading.