Koji's Split Personality

[aron7awol]

 
Can we find out from other years if he is better in individual games when he throws 75%+ splitters?  Even if true, drawing conclusions from that is dangerous, because maybe he only throws that many splitters when he has a good feel for it.  It could just be that when Koji has a good feel for the splitter, he throws it a lot, and it works really well. Or maybe he only throws that many when he is facing hitters susceptible to that pitch. There are other explanations as well. But it might not be as simple as more = better. 
 
We have a pretty recent (larger) data point, when Koji had 101 Ks, 9 BB, and a 1.74 ERA in 71 innings while throwing 48% splitters.  Do you think this was a suboptimal mix, and he could have been significantly better that year?  Or is it only 2015 Koji, with his diminished fastball, that should be throwing 75% splitters?  Not trying to be snarky, I'm just wondering if 75/25 is the new optimal mix, or if you think he should have always been doing this.

I hope at some point I have the time to compile all of this into a database and analyze it to that level.  I do agree that his feel for the pitch on a particular night is a factor, as are the hitters faced, the count, etc.  However, that's exactly why I'm splitting the data into relatively large buckets; it should eliminate most of that noise.  I'm convinced at this point that Hanigan is the major factor in the pitch mix difference this year.  Also, it seems there are hardly any nights where he doesn't have feel for the splitter, considering the ridiculous lack of walks throughout his career.
 

I do think that was a suboptimal mix in 2013, despite the ridiculous results that year.  The gap between the splitter and fastball was actually the largest that 
year.
 
The diminished fastball is a very interesting part of this analysis for this year.  It's been well established at this point that fastball effectiveness is directly correlated with velocity, so that's just another point for increasing splitter usage this year.  By the way, based on the data so far, I'm thinking the optimal mix is probably even higher than 75/25.  With zero evidence of any kind of drop in splitter effectiveness, and in fact an increase so far instead, I'd be throwing 80%+ at least until I saw something to suggest it's too much.
 
I really think the most important factor in all of this that I've mentioned before but we really haven't talked much about is his ability to make the splitter behave differently.  It's almost like a knuckleball-lite that he has control over.  To me, that's the reason he can throw so many and have it maintain effectiveness.  Like the knuckleball, you can know it's coming, but that doesn't mean you know how it's going to move, and it moves so late that you can't just see it and hit it very well either.