This post is a continuation to my first post at The Hardball Times Live which can be read here
While I was pretty happy with the conclusions of that article one point stood out to me as being relatively weak analysis wise. Drawing conclusions from sets of four data points is something I generally try to avoid. But the batting eye and percentages for the different pitches were so different that it wouldn’t have worked to combine them in the correlation.
I decided to check what would happen if I z-scored everything and then compared them. Here is the updated table with the z-scores included.
With this many data points I feel like putting it into graph form will help greatly so here is a plot – with best fit line included – of the 16 pairs of z-scores (note: click for full-size.)
We see that there is a pretty solid correlation here but we’ve also got 2 major outliers. Thinking back to the original article it didn’t make much sense to include all the points when we saw that one type of pitch didn’t follow the same pattern. So I decided to plot the 12 non change-up points.
Now we’re talking! This is an extremely strong correlation which I think adds further support to the conclusion of my original article.
Just wanted to do my due diligence analysis wise. I think this is a nice addendum to the Hardball Times article.
