BaseRuns as a rate stat

[Mean Dean]

When expressing BaseRuns as a rate stat, would it be more appropriate to express it per 27 outs, or per plate appearances in a game (i.e., per 38.5 PA, or thereabouts?)

Thanks!

[Tango Tiger]

Total runs should be expressed per n outs, where n is whatever you want.

[Mean Dean]

Okay, thanks. I ask because outs are basically already part of the formula (the C factor), so I thought it might be "double-counting" them. The BsR-per-27-out list certainly seems very OBP-heavy. But I guess that is how the game really works!

[jinaz]

Here's another thread on this issue, with a few more details:
http://www.baseball-fever.com/showthread.php?t=69298

Also, Patriot touched on this issue here:
http://walksaber.blogspot.com/2007/10/replacement-post-brought-to-you-by.html
-j

[Tango Tiger]

The way to remember it is if the out value is around -.10 runs, then you want to do runs per out.

If the out value is around -.28 runs, then you want to do runs per PA.

[SABR Matt]

Bingo Tom. If the run value of an out is -0.1, it means you counted only the negative impact of the out on the probability of scoring runs...if it's more like -0.27, then you counted that AND the opportunity cost of losing one chance to bat (there being a finite number of outs in a game).

Which is why when you compare BsR runs created (or markov linear weight runs created) to some replacement or margin, you need the denominator to be outs or you're missing a piece of the puzzle.