Members: Login | Register | Feedback
 
   
 
Playing around with Run Expectancy Matrix
Posted: 24 October 2008 02:56 PM   [ Ignore ]

I’ve been thinking about a play-by-play data based approach to gauge offensive production using the Run Expectancy Matrix I found here (http://www.tangotiger.net/RE9902.html):
Per each plate appearance one compares the state before and after the ball is hit into play, the batter walks or strikes out. Therefore the result of runs scored+runs expected (after)-runs expected (before) is used to measure the offensive production of a player (positive=extra runs a player created compared to an average player, negative=runs a player cost a team compared to an average player).

E.g.
Before: runners on first and second, 1 out = 0.971 runs expected (Player hits a ground rule double)
After: runners on second and third, 1 run scores, still one out= 1.467
So the batter is credited with +1.496 points.

Before: bases loaded 1 out=1.65 runs expected (Player hits into a double play)
After: inning ended=no runs
Batter is charged with -1.65 points

Of course offense is not all about batting so there a few open points:
*) stealing: baserunners earn/lose points based on results of stealing/getting caught
E.g. before: runner on first nobody out 0.953 runs expected to score, (runner gets caught stealing 2nd)
    after: nobody on 1 out 0.297 runs expected to score, runner charged with -0.686 points
*) errors: that#s a tricky one. I’m leaning towards not awarding any points for advances on the basepathes, runs scored..etc based on errors
*) wild pitches: players of the offense are not credited with points

Applied the model to World Series game 1 and got some interesting results:
Jimmy Rollins and BJ Upton both got about -3 points (mostly for hitting into double plays) and Iwamura and Utley lead the pack with about 1.8 points.

Okay so what do you think about that? (If there’s something different I got to apologize I didn’t want to steal someone else’s idea)
Obviously it would be a little complicated to determine and much like the RBI stat a player’s performance depends on the work of his colleagues but IMO an advantage is that it would take GDPs into account which hurt alot.

PS: Please excuse my bad English (that’s not my native tongue)

Posted: 24 October 2008 07:46 PM   [ Ignore ]   [ # 1 ]

Some of this is already done, at www.fangraphs.com.  They have the full regular season results available in terms of both runs and wins added above the average hitter.

Posted: 24 October 2008 09:18 PM   [ Ignore ]   [ # 2 ]

Okay thanks a lot should be the BRAA stat.