Post by ericmvan on Oct 8, 2013 19:05:15 GMT -5
So, the first question about Hellickson's season is: can we combine the inconsistent last four games with the inconsistent first 14, to simplify things? With one important exception, it sure looks like we can. When you split the good games from the bad games in each stretch, you get two pairs of lines which clearly pair up in terms of K rates, BABIP, and GB%, even though two of the lines are just two starts each (details available upon request; I started boring myself typing them!).
So here's the inconsistent Hellickson, first his 12 bad games and then the line for the 6 good games (two in back-to-back pairs, two isolated). (HRC = HR / Contact, xBP = expected BABIP, more about which in a moment).
The two HRC are for the first and second stretches. That's the exception: in his last four starts, he gave up a ton of homers.
Expected BABIP here is not the simple one predicted by the league BABIP rates for each type of ball in play. That gave significant "errors." However, the correlation of K rate to BABIP is well established, and in this case, even though I had just four data points (these two lines and the sustained good and bad stretches), the correlation of the "error" to K rate was .97 (p < .03). IOW, essentially all of the BABIP that's not explained by the ball in play breakdown is explained by its known sensitivity to K rate, which for him is extreme.
So, what do we see here? In his good games, he struck out precisely 50% more batters and became an extreme flyball pitcher to boot. His popup rate nearly tripled, and even including the late-season HR troubles, he cut his HR / OF Fly rate more or less in half, from .233 to .118 (not in the table). He cut his LD% by 23%, and all of this combined to move him from a terrible BABIP pitcher to one with an off-the-charts BABIP skill (I have no doubt that this kind of performance can be real in isolated games, and hence scattered games combined like this; it's not sustainable because batters adjust. If you've watched a lot of baseball, you've seen this happen a lot).
So, you'd expect that his sustained good and bad stretches would resemble these, right? Hah!
In his sustained good stretch, he wasn't an extreme flyball power pitcher with dazzling command; he had just a solid K / BB ratio and got a ton more groundballs. On other hand, in his sustained bad stretch, his GB% much more resembled that of the power pitcher, but his strike zone command was just terrible, with a resulting rise in LD% and BABIP.
I remain entirely agnostic about what all this means, even that it means much at all, other than that he seems to be a guy who tries different approaches with varying success. Remember that he pitches for an organization that might be the single most analytic-intensive in the game, so it's credible that they might be having him make significant alterations.
If I had more energy, I'd look at pitch/fx data, but I actually typed all this up while not only watching the amazing A's / Tigers game, but scoring every pitch! In any case, I hope this gives folks enough to try to identify which version of Hellickson we're getting. If he's pounding the strike zone (crap, I forgot to look at Str%, etc.) and getting easy fly balls and popups, we're in deep trouble, and have to hope that a fly ball or two leaves the yard. Otherwise, we should be fine.
So here's the inconsistent Hellickson, first his 12 bad games and then the line for the 6 good games (two in back-to-back pairs, two isolated). (HRC = HR / Contact, xBP = expected BABIP, more about which in a moment).
What This Is... BFP K% BB% HRC / HRC BABIP xBP GB% OFB% LD% PU%
Inconsistent Bad 291 .165 .065 .056 / .080 .343 .342 .430 .269 .251 .049
Inconsistent Gd. 141 .248 .057 .029 / .071 .191 .205 .316 .347 .194 .143
The two HRC are for the first and second stretches. That's the exception: in his last four starts, he gave up a ton of homers.
Expected BABIP here is not the simple one predicted by the league BABIP rates for each type of ball in play. That gave significant "errors." However, the correlation of K rate to BABIP is well established, and in this case, even though I had just four data points (these two lines and the sustained good and bad stretches), the correlation of the "error" to K rate was .97 (p < .03). IOW, essentially all of the BABIP that's not explained by the ball in play breakdown is explained by its known sensitivity to K rate, which for him is extreme.
So, what do we see here? In his good games, he struck out precisely 50% more batters and became an extreme flyball pitcher to boot. His popup rate nearly tripled, and even including the late-season HR troubles, he cut his HR / OF Fly rate more or less in half, from .233 to .118 (not in the table). He cut his LD% by 23%, and all of this combined to move him from a terrible BABIP pitcher to one with an off-the-charts BABIP skill (I have no doubt that this kind of performance can be real in isolated games, and hence scattered games combined like this; it's not sustainable because batters adjust. If you've watched a lot of baseball, you've seen this happen a lot).
So, you'd expect that his sustained good and bad stretches would resemble these, right? Hah!
What This Is .. BFP K% BB% HRC BABIP xBP GB% OFB% LD% PU%
Good Stretch 169 .219 .071 .025 .274 .256 .492 .208 .192 .108
Bad Stretch 126 .103 .087 .040 .400 .403 .354 .293 .293 .061
Inconsist Gd 141 .248 .057 .041 .191 .205 .316 .347 .194 .143
Good Stretch 169 .219 .071 .025 .274 .256 .492 .208 .192 .108
Inconsist Bad 291 .165 .065 .058 .343 .342 .430 .269 .251 .049
Bad Stretch 126 .103 .087 .040 .400 .403 .354 .293 .293 .061
In his sustained good stretch, he wasn't an extreme flyball power pitcher with dazzling command; he had just a solid K / BB ratio and got a ton more groundballs. On other hand, in his sustained bad stretch, his GB% much more resembled that of the power pitcher, but his strike zone command was just terrible, with a resulting rise in LD% and BABIP.
I remain entirely agnostic about what all this means, even that it means much at all, other than that he seems to be a guy who tries different approaches with varying success. Remember that he pitches for an organization that might be the single most analytic-intensive in the game, so it's credible that they might be having him make significant alterations.
If I had more energy, I'd look at pitch/fx data, but I actually typed all this up while not only watching the amazing A's / Tigers game, but scoring every pitch! In any case, I hope this gives folks enough to try to identify which version of Hellickson we're getting. If he's pounding the strike zone (crap, I forgot to look at Str%, etc.) and getting easy fly balls and popups, we're in deep trouble, and have to hope that a fly ball or two leaves the yard. Otherwise, we should be fine.