Before I begin, let me state that I am not a mathematician, nor do I always completly understand some of the statistics on this website.  I do the best I can, but I'm not sure I'd have the ability to test this theory if I wanted to back up my hypothesis.

Graham and Matthew's tRA is a model based around the idea that a ball in play can only have 5 outcomes (LD, GB, OFB, IFB, HR).  After that, the result is beyond the pitcher's control, therefore, a pitcher that posts consistent rates between these five outcomes will, in a park neutral environment with neutral defense, will post a consistent BABIP.

However, another key point that they raise is the regression towards the mean the batted ball results.  To quote the tRA explained page "The order is extremely important, as influencing GB% will have an effect on LD% later, and so on, sometimes causing regression away from the mean in unusual situations." tRA Primer

Part of my understanding of this (again, I could be wrong) is that a pitchers batted ball results will tend to logically distribute around their tendencies.  For example, a groundball pitcher that suddenly posts a high IFB rate, without seeing a raised OFB or HR rate, is due for regression towards his most frequent outcome, a GB, and therefore, the IFB rate is unsustainable and it is likely his OFB and HR rates will rise at the expense of the IFB rate.  (Again, I could me misinterpreting, but this seems logical to me.)

So my question is that is this true about swinging strikes, and if so, how could this be tested?

For example, we can assume that when a batter swings and misses, there are four possible reasons for the missed contact.

1.  The Batter swung over the ball (I'll call his mO, for Missed Over)

2.  The Batter swung under the ball (mU for Missed Under)

3.  The batter swung too early.  (mE for Missed Early)

4.  The batter swung too late (mL for Missed Late)

Just for the sake of my argument, I'm going to use JJ Putz as an example.  I do not have the pitch fx data, but in 2006, JJ Putz posted a 15.0% swinging strike rate from all his pitches.  Lets assume (feel free to find the data) that JJ's splitter also got a 15% SwSr.  My guess (complete guess) is that when batters swing and miss against JJ's splitter, it is because the bottom falls off the pitch, and they swing too high (mO).  Logic would then say that in looking at the SwSr% in depth against JJ's Splitter, an overwhelming percentage would be mO.

Now, the reason I raise this point is simple.  Lets now take Jarrod Washburn for a moment and look at his SwSr rate.  Washburn has (SSS applies)  posted a 7.3 SwStr% so far in 2009, up from 6% in 2008.  A 7.3% rate would be a career high for Washburn if he were to continue it.  However, if we had data available to analyze his career SwSr breakdown, we could evaluate if this success is sustainable, or due for a regression to the mean.

Hypothetically (Read: I'm pulling these numbers out my ass) lets say Washburn has a SwSr distribution of 40% mO, 40% mU, 10% mE and 10% mL.  Now lets assume that so far in 2009, the rate has shifted as such (35% mO, 45% mU, 10% mE and 10% mL)  If this is true, then, as the tRA model suggests, these changes must be reflected somewhere else. 

More batters swinging under Washburn's pitches, and less batters missing over Washburn's pitches would suggest that batters are getting underneath more pitches, and would logically be reflected throughout his batted ball distribution (more FBs, less GBs etc.)  If everything else were to remain the constant, including, for example, the percentage of TOTAL pitches that were mO, (the increased SwSr means this number is less proportionally) it would be logical to conclude that extra missed bats classified as mU, are therefore unsustainable and due for a regression towards the mean.  Otherwise the entire BB distribution would shift.

This example is a gross oversimplification, but I believe the logic is sound.  To the best of my knowledge, there is no system in place to further classify SwSr, and my guess would be that doing so would be difficult.  The classifications, for example, would be arbitrary at best. 

For example, if a batter is fooled on a change up, out on his front foot, and in an attempt to slow down his swing, missed over (mO) on the pitch, but the swing was slowed enough to be properly timed, the strict definition would say this is a mO, when the real reason for the swinging strike was that the batter swung early (mE), and swung over the pitch as he slowed his swing to compensate.  How do you determine what to call this scenario?

To me it seems logical that if a pitcher that misses most bats by having batters swing too high suddenly pitches a 90 pitch game with 15 swinging strikes, but 10 of those strikes were mU, then we can safely assume the game was a SSS anomaly, and not the result of "________ finally figuring it out!!!"

This entire fanpost is just idol minded speculation, but I do think it warrants some discussion.  This is my first statistical based FP here, so hopefully it was interesting.  Looking forward to some discussion, however harsh :)

Cheers

BQueezy