THE BOOK--Playing The Percentages In Baseball

IIRC, Woba is based on linear weights. What if the linear weights go against conventional wisdom like 3b being less valuable than 2b? or if the difference between the two is marginal in the linear weights?

How do isolate the values of these hits from other hits? Do you adjust the values of 2b if the guy on 1b is fast? that can skew the data a bit.

Linear weights is more of a dominion rod than a bow. That’s what I learned.

Always enjoyed the follow lyrics by Suicidal Tendencies ..

Just cause you don’t understand what’s going on don’t mean it don’t make no sense. And just cause you don’t like it, don’t mean it ain’t no good

Grammer be damned.

pm: if you get triple less than double, then you made a calculation error.  Obviously, if your sample size is too small, and the two sets of data are not from the same population, you can get weird results.  That’s why you have to do it the right way.

Very nice explanation on that web site. It is somewhat personal preference whether you want to include ROE, IBB, and SB/CS. It also depends on what you want the stat to represent and what you are using it for.

Using ROE is obvious if you want to reflect a batter getting on base and not making an out regardless of how or why. That is somewhat the case with IBB, although the value is obviously less than a NIBB or a HBP and is somewhat player dependent (which is one reason why leaving it out is not too bad - since the average value of an IBB for any particular player is around equal to the average value of his PA). Also, the ROE has some skill component to it (at least with respect to the batter’s speed, handedness and GB/PA percentage) although not as much as with the other offensive components. Using SB/CS adds something to estimating a player’s overall value, but since it is not a batting stat, it may be better to include it separately. 

wOBA is truly a great overall batting value stat and it is widely becoming THE stat in somewhat advanced circles, as it should be. Total Average (Tav) by BP is exactly the same thing. If it was not developed before or independently from wOBA, as far as I can tell, it is a rip-off…

TAv (True Average), and its relationship to wOBA was discussed here:

http://www.insidethebook.com/ee/index.php/site/comments/eqa_tav_coalescing_around_linear_weights/

And here:

http://www.insidethebook.com/ee/index.php/site/comments/eqa_renamed_tav_true_average/

I forgot about those discussions. And whoops, True Average not Total Average.

The proper discussion about Tav is simple “It is pretty much the same thing as wOBA, only we use slightly different, dynamic weights and it is always park adjusted.”

There should be no discussion about whether it is better or not. End of story.

wOBA is just linear weights as well, so I’m just happy that we don’t have to have these discussions any longer.  We have LWTS and we have BaseRuns, and that’s all we should be talking about. 

I show it as wOBA for the OBP scale, Colin shows it as TAv for the BA scale.  I also have wRC along the Runs scale, and wRC+ along the average=100 index scale.

Same language, different dialect.

Say, when you don’t have ROE, can you use some fraction of AB-H-SO to substitute in the formula?

Yeah, sure.  I mean, you can have a separate rate for LHH, RHH.  You can have one based on SB/(1B+BB) and/or 3B/(2B+3B).  Whatever is reasonable is fine.

You can also just ignore it altogether.