Picking up where I left off on the last post...

This post is a continuation of the game score analysis I posted a few days ago.

In essence I looked at the average game scores of each starting pitcher this season (minus Liriano) as a measure of their dominance, using the standard deviation of their samples as a measure of their consistency. Scott Baker, for example, posted an average game score of almost 55 with a standard deviation of around 10, which is fairly consistent. I then calculated a "modified" standard deviation by throwing out each pitcher's best and worst game score to arrive at a more accurate measure of their consistency.

MJesser, in the comments, had the great idea of looking at each pitcher's quartile results. Basically you look at the best (and worst) 25% of their results to get a benchmark of what you can reasonable expect as their best (and worst performance). I haven't had time to run these numbers yet on any league leaders or current/prior Johan seasons for comparison, but I will try to.

Here are the results:

Format: best case: ave GS, SD (that is, level of consistency)/modified ave GS with best performance omitted, mod. SD with best result omitted
Also, with SD's, looking at historically great (and bad) seasons, it looks like a score above 12 is rather erratic, between 6-12 is average, and below 6 is very consistent.

Scott Baker

    * Best case: ave. GS of 67.33, SD of 6.5/m.ave. GS of 63, m.SD of 6
    * Worst case: ave. GS of 44, SD of 9.5/m.ave. GS of 50, m.SD of 1

Kevin Slowey

    * Best case: ave. GS of 79, SD of 9.1/m.ave. GS of 72.33, m.SD of 4.16
    * Worst case: ave. GS of 32, SD of 19/m.ave. GS of 44, m.SD of 2

Nick Blackburn

    * Best case: ave. GS of 69, SD of 4.1/m.ave. GS of 66.25, m.SD of 3.3
    * Worst case: ave. GS of 28.5, SD of 8.1/m.ave. GS of 33.75, m.SD of 5.37

Glen Perkins

    •    Best Case: ave. GS of 59.33, SD of 3.21/m.ave. GS of 56.3, m.SD of 2.08
    •    Worst case: ave. GS of 29.66, SD of 12/m.ave. GS of 44.66, m.SD of 1.5

Boof

    •    Best Case: ave. GS of 64, SD of 6.92/m.ave. GS of 59, m.SD of 8.66
    •    Worst case: ave. GS of 18.33, SD of 6.65/m.ave. GS of 27.66, m.SD of 10

Livan

    •    Best Case: ave. GS of 63.75, SD of 3.4/m.ave. GS of 61, m.SD of 4.24
    •    Worst case: ave. GS of 15, SD of 5.6/m.ave. GS of 19.25, m.SD of 4.92

Some pretty interesting results. Using the modified results (omitting their best and worst performances), Baker can be expected to be no worse than average on any given day. However, his ceiling is significantly less than that of Slowey, although Slowey is a bit more unpredictable, and when he's bad he can occasionally be really bad. Blackburn, so far, has demonstrated a slightly higher ceiling--and more consistently--than Baker, but when he's bad he's pretty bad. Perkins still doesn't show that much potential to be anything more than mediocre at best, but his low end performances don't tend to be all that bad either.

That leaves Boof and Livan.
Both show potential to be pretty good when they're good, but Boof is pretty erratic on the upside. On the downside prospects look pretty grim. When Boof is bad, he is very bad and unpredictably so. Livan is even worse, and more predictably. As much as he eats innings and gives you the occassional solid performance, Livan is (according to this analysis) just too much of a predictable negative risk to be trotted out every fifth day.

One last little tidbit. I've been looking at the Leverage Index for each pitcher as well, and so far Blackburn has pitched most consistently in high leverage situations among the starting pitchers. Slowey, on the other hand, has pitched in the least intense situations overall. The other four are all about the same. When compared with their Win Probability Added percentages, however, a clear trend appears: Baker, Slowey, and Blackburn are solid pitchers who consistently give us a chance to win. Boof and Livan both need to go (well, now that Boof's in the pen Livan's the only one who needs to leave). Perkins, up to this point, hasn't shown the stuff to be considered more than the #5 starter, and should be upgraded if possible. Whether that means Liriano, Duensing, or someone else I do not know.