Over at Beyond the Box Score, a commenter named Iblemetrician suggested that I reconsider how I calculated the value of switching a ball to a strike.
When I originally calculated the run value of a ball to a strike, I looked at all plate appearances for a given count. That came out to a run value of .161 runs Iblemetrician pointed out that certain counts will have more pitches taken – and even potentially have more pitches on the edge, which might be more likely to be miscalled.
With those suggestions (and after a little bit of programming), I went back and calculated a run value weighting by both called pitches and called pitches on the edges of the strike zone.
Let’s look at the same sort of table as in the first post – this time weighted by called pitches.
|B||S||WOBA||LW||RV Ball||RV Strike||RV B->S||Called Pitches||Weighted RV|
Run Value of Switching a Ball to a Strike – Weighted by Called Pitches
Now let’s look at the same breakdown, but considering called pitches on the edge. These are pitches within 2 ball radii of any side of the strike zone (and within the strike zone on the other dimension). That means we’ll count the pitch if it’s one ball width outside the strike zone on the left and middle of the strike zone in height, but we won’t count it if it’s one ball width outside the strike zone on the left and three balls widths outside the zone high.
|B||S||WOBA||LW||RV Ball||RV Strike||RV B->S||Called Pitches on the Edges||Weighted RV|
Run Value of Switching a Ball to a Strike – Weighted by Called Pitches on the Edges
I’m a little worried about sample size here – at least comparatively. The results back up the expected behavior though. Players are less likely to take a close pitch with two strikes than otherwise – which lowers the relative run value of those events and therefore the overall run value of “mistakes”.