THE BOOK--Playing The Percentages In Baseball

Ok, I did a double-check, and they are roughly comparable.  Though I probably should update my link with fresher data.

I’ve always been confused by the connection between boLI and LI.

My assumption is that overall leverage index is indicated by two things: base/out leverage index and score/inning leverage index - and that there should be a well-defined formula relating the two to get to the overall LI.

The implication being that if you looked at the same score/inning state with two different base/out states, you’d be able to calculate the same score/inning leverage with some manipulations of the overall LI and the boLI.

I’ve tried both linear and multiplicative models and I haven’t been able to make it work.  Unfortunately I’m not sure I still have my scratch pads where I was playing with this, but I could easily reproduce them.

What am I missing?

You definitely cannot do what you are hoping to do.

What Dan is saying is this: you have isLI (inning/score Leverage Index) and boLI (base/out Leverage Index), and that you should be able to multiply the two to get isboLI (or Leverage Index as I have formulated it).

But, that requires a condition that each boLI is independent of the inning/score.  And, that can’t possibly be true.  I think you can easily focus on some bottom of the 9th scenario to show how untrue this is.

If each boLI isn’t independent of the inning/score, how can you apply them generically?

I think that’s the main crux of my internal problem.

Right, you should only apply boLI if you believe that the inning/score does not impact the batter/pitcher approach to that base/out state.

This is the same thing as say showing us how a batter does with runners on or bases empty as two categories, which explicitly creates a barrier between the two, when in fact, some bases empty situations may have an approach closer to the traditional runners on situations, and some runners on situations may have an approach closer to the traditional bases empty situations.

Now, to do what you are really wanting to do, you want to take my Leverage Index, and come up with a better, more specific, process to convert it into “independent” boLI and isLI.

So, for example, let’s say boLI says it’s 1.5, and isLI says it’s 1.3, and LI says it’s 3.0, then 1.5x1.3 won’t add it up (since that gives you close to 2.0).  There’s still yet another 1.5 of “interdependence LI” unaccounted for.  You could say take each of boLI and isLI and multiply by the square root of 1.5 and apply the interdependence LI that way.

Thansk for giving me something to think about!