Originally Posted by rdiersin
You can take it to a player level BF. I understand your doubts for using it on team statistics and the jump to players, but Extrapolated runs (XR) is another similar stat that on a team stat basis gives similar correlation with Runs scored as RC does. Xr, or other types like Palmers I believe, is linear, so each players contribution adds to the teams Xr. So you might then say, well it doesn't work on a micro basis. But if you look at all of 2003 and 2004 games logs, you can see the Xr and similar stats perform fairly well with a correlation coefficient of around 0.85 or so to runs scored for each game in those seasons. So, in summary, it does measure player contribution, and does a heck of alot better than rbi's and the like.
A BB example:
For a stat like RC to work at PLAYER level it would have to operate on the assumption that over time all BBs for all players on all teams are equal (similar for doubles, triples etc.) in regards to Run Scoring efforts.
Example: In national league with pitchers hitting your ninth hole will be an out for first 5 innings what 85% of the time? Your 8th hole will be an out ~70-75% of time on average? You only get THREE outs an inning.
I'm sure you know of that Tango table which gives run scoring percentages based on outs and runners on base. That table doesn't take into account quality of hitters.
So right from the start it is easy to see at micro level why a BB to a top of the order hitter who has great speed is FAR more valuable to team run scoring efforts (and would be on high end of Tango table curve) than a BB to a 6 or 7 hitter who doesn't run very well. (and would be on low end of Tango table curve)
If this was just one exception you might say insignificant but the FACT is these situational "tendencies" for lack of better word or "distortion possibilities" are present every time thru the lineup with every player in different ways. They all wash out for lineup as a whole but for a specific player you'll get same distortion potential over and over and over, every time thru lineup and every game.
- - - - - - - - -
So I guess my point is variance/margin of error would increase greatly at PLAYER level because of individual player and situational differences. Throw all players in a bucket fine. But just to grab ONE player out of bucket - he is likely to be overstated or understated and possibly in a predictable way.