from Amazon
Thursday, June 18, 2009
Win-equivalent SB
Dan converts SB, CS, PK into an equivalent-line, depending on the game state they occurred. Basically, it works out as the following: newSB = actualSB * LI . And then he scales it by the run environment. Basically.
I disagree with his using of rolling averages. The run environment of 1969 has nothing at all to do with the run environment of 1968. (Is that one reason Camps shows up so well?) It should strictly be based on the runs per game. It should be fairly straightforward to come up with a function that determines the win value for SB and CS given the runs per game, similar to what I did for the run value for major hitting events.
Also, the reason for the better run value for the PK compared to the CS is that PK do not always lead to outs (neither do CS for that matter). I think 25% of PK keeps the runner on base. It’s one of those silly accounting rules that confuses matters. Ideally, we’d have a PKsafe and PKout, to distinguish between runners who were “picked off” by the pitcher but managed to get to 2B anyway.