Graphical Game Scores
Posted by Andy on June 30, 2010
Here we examine a graphical look at Game Scores.
We've talked a lot lately about Game Scores. I decided to look at them graphically. This isn't necessarily a great idea--I just got curious to see what it would look like.
First, if you need a refresher on how Game Scores are calculated, click here. It's the second paragraph on the page.
I picked a couple of games at random, both from Sunday. First is the A's/Pirates matchup featuring Gio Gonzalez and Ross Ohlendorf on the mound. The other one is the Red Sox/Giants game with Tim Lincecum vs. Jon Lester.
I went through each line in the play-by-play accounts and assigned each pitcher a score for the outcome of the at-bat. For example a flyball out would be +1, a single would be -1, and a strikeout would be +2 (that's +1 for the out and +1 for the strikeout.) A solo homer is -6, the sum of -2 for the hit and -4 for the earned run.
This graph shows Game Score vs Inning. Data points between 0 and 1 occur during the first inning, and the value at 1 inning was the Game Score at the conclusion of the first inning. (I don't think I did this exactly correctly in all cases--bear with me since this was a quick attempt to see if this graphing method id worth anything.) When the Game Score stops changing, the pitcher was out of the game. (Only Lester completed his game while Lincecum got knocked out after the 3rd inning.) I have also added credit of +2 for any inning completed after the 4th inning, as per the standard Game Score calculation.
The Lester and Lincecum graphs don't tell us too much we don't already know. Lincecum got bombed and Lester was excellent. The Gonzalez and Ohlendorf lines are more interesting. Ohlendorf started stronger, allowing fewer baserunners early. He hit a rough patch during the 5th inning (shown on this graph between innings 4 and 5) but he stayed in. He finished with a strong 6th inning and was then lifted from the game. Gonzalez meanwhile started more slowly than Ohlendorf but passed him in the 5th. His lead was short-lived, however, when he allowed a solo homer leading off the 6th, dropping his score by -6 and ending his day. (The graph doesn't reflect this exactly--it shows the -6 drop coming happening right at inning 6 instead of just after 6.)
I believe Graphical Game Scores might be quite useful for analyzing when pitchers were left in too long in games. I would think that a lot of graphs end like Gonzalez's, with one or more earned runs right at the end causing a sharp downturn. This is an indication (a 20-20 hindsight one, of course) that the pitcher should have been removed from the game earlier.
Looking at the lines for Ohlendorf and Lester, they both overcame bumps in the rO.D. Gonzalez did not. We could count these bumps (easy to do with a simple derivative calculation for each line) and assign points to pitchers for 'bumps overcome' or to managers for leaving pitchers in too long...
Anyway, don't read too much into this post. Just a little brainstorming out loud!
June 30th, 2010 at 10:30 am
It might be better to stop graphing the score once the pitcher leaves, it may show a good vs bad performance more easily
June 30th, 2010 at 11:52 am
i think it's fairly clear that the horizontal portion is when the pitcher has left.
June 30th, 2010 at 2:02 pm
This is neat. Especially because it shows while Lester wasn't at his best early, he went into cruise control in later innings
June 30th, 2010 at 8:36 pm
Forgive me, but while I like the idea of seeing game score graphically, I'm not sure I see the purpose of retrospective examinations of game score to indicate when a pitcher should have been pulled. Follow my logic here:
Game score does, by definition, indicate the performance of a pitcher over his outing. Conversely, it makes sense that a live game score would indicate his performance at the time. Further, one could look at a game score, look at its lowest point, and say "right before that is when he should have come out." Which is obvious, considering that bad things make game scores go south, while good things make it go up. One could note, as in the case with Ohlendorf above that taking him out during the 5th would have been a mistake, as he managed to improve later in the game. Also, an obvious fact from the game itself and ONLY helpful because he stayed in. It's possible that Lincecum could have improved too--we'll never know.
I just question the value of interpreting a game score as a means of showing live pitching performance when a game score is designed to be a rough measurement of live pitching performance. It'd be like tracking batting average over each at bat and using it to show slump trends. The lack of hits pretty much already does that.
My thoughts. That being said, I like the look of the graphics!
July 1st, 2010 at 2:59 am
Where are the Game Score Stats for the pitchers ?
July 1st, 2010 at 11:34 am
While Michael is right -- this invites second guessing -- if you looked at a lot of them certain patterns might emerge that would indicate a future drop in performance. Kind of like stock market chart patterns -- maybe a head and shoulders formation would mean pull the pitcher NOW.
July 1st, 2010 at 12:59 pm
Yeah, as I said in the original post, they are useful only in hindsight. I would not suggest using them as a live statistic to decide when to take the pitcher out. They would only be useful after the fact to note when mistakes were made and possibly make better future decisions.