BaseRuns - Addendum

Some data and formulas used in my BaseRuns article

1974-1990, all data generated from Retrosheet and Ray Kerby

Events Official Stat lwts_RC lwts Freq A B C D
Single x 0.463 0.460 436518 1 0.726 0 0
Double x 0.753 0.750 103215 1 1.948 0 0
Triple x 1.035 1.033 15962 1 3.134 0 0
HR x 1.402 1.402 53874 0 1.694 0 1
Walk x 0.303 0.303 202435 1 0.052 0 0
IBB x 0.176 0.176 22260 1 -0.483 0 0
HBP x 0.330 0.330 12652 1 0.163 0 0
Error - 0.481 0.478 31270 1 0.799 0 0
Interference - 0.357 0.357 369 1 0.277 0 0
OtherSafe - 0.631 0.631 2947 1 1.433 0 0
Sac x 0.106 -0.090 28251 0.080 0.727 0.920 0
Strikeout x -0.111 -0.269 361333 0 -0.057 1 0
Out - -0.098 -0.265 1367321 0 -0.004 1 0
SB x 0.193 0.193 48729 0 0.813 0 0
CS x -0.282 -0.437 20543 0 -1.188 0 0
Pickoff - -0.120 -0.228 9654 0 -0.507 0 0
PickoffError - -0.082 -0.182 30 0 -0.347 0 0
Balk - 0.250 0.250 4980 0 1.054 0 0
PB - 0.276 0.276 5021 0 1.165 0 0
WP - 0.279 0.278 18371 0 1.174 0 0
DefensiveIndiff - 0.132 0.132 184 0 0.555 0 0
OtherAdvance - -0.252 -0.362 633 0 -1.061 0 0
FoulError - 0 0 1056 0 0.001 0 0
implied outs - -0.451 0 3461 0 -1.488 1 0

A couple of quick points as to why I did certain things. I tried to weigh the merits of not breaking up "official stats" like strikeouts, so that we can apply these measures historically. However, some "K" are also safe plays. It's rare enough that we can ignore this. But then I get the problem with the sac bunt having safe plays as well. This is why you see that annoying ".08" in the basrunner field. I also have a problem with partial innings, and the case where you have runners left on base. So, I introduced "implied outs", which is the number of outs "left in the game", to balance out. The "lwts" values and the "lwts_rc" values are identical except for the outs. However, you can get outs with a single, which is why the two numbers are not exactly the same. Using the Retrosheet scoring system of events, anything that happens following a single is credited to a single. Even a double-steal counts as "1" in the SB field.

Anyway, there's alot of technical things that don't amount to a hill of beans, but that I had to make certain assumptions/decisions so that everything added up. I'm not sure if all the decisions are correct, and I suspect that I may change some of this in the future.

"Freq" is how often that event occurred. If you multiply the "freq" by the "lwts_rc", you get the total number of runs scored. If you multiply "freq" by "lwts", you get zero (or pretty close to it).

The BaseRuns formula follows the true definitions of scoring runs: BaseRunners x scoreRate + HR. BaseRunners is denoted by "A", scoreRate is "B" / "B" + "C", and HR is "D". Again, you can make the case that the CS removes a baserunner, except that sometimes a CS is still safe. And you can also take it from the point of view of "initial baserunners". Again, I'm not 100% sure that these decisions are the best ones.