Creative argument, attempting a rebut vs. SSI's prosecution of the suspect:

I'm not a big fan of using BQS.  For one, it seems like it has typically low rates, I can't seem to find the leader boards, but I can't imagine the lowest total being much lower than 3 or 4, the highest about 20?  And the things that can lead to them, I mostly see; starting pitchers at 6 innings, 3 runs give up a homer, letting a young pitcher work through issues in a blowout (between the Indians being on the right side of a lot of those and Wedge being the kind of guy to let that happen I think that would come up a lot), or a ill timed bad luck (a misplayed flyball leads to a run instead of out #3).  Of course the fact that the Indians were among the league leaders most years is worrisome (except once again I can see him letting pitchers like Cliff Lee and CC Sabathia work through tough spots when the team is up by 5 more often than other managers).

D.A. Detecto, rather taken aback by this proposal, replied offhand:

The diffusion for BQS is easily equal to or higher to that of most stats.  The average is 6-7; the leaders have twice that, the bottom end has half that.

Compare NFL wins.  In a 14-game season, if averag e was 7 wins, the best team had 13 wins and the worst team had 3, would you say that the wins column had too little distribution to separate the teams?

How about if one team won 2, 3, 4 , 3, 3, 2, 4 games over the course of seven seasons?

The defense persists ...

Also, I don't like the comparison of BQS to Football wins.  In football, the max number of wins is 16, in BQS, the max number would be, what, 90?

Seeing this extremely outside-the-box counter,  the D.A. blinks several times ....  shuffles his papers, raps them on the desk, and finally realizes that he is expected to supply a serious response.

Allllll righty then.  What say we clear the water as to Eric Wedge's bullpen management....

.

 === New Statistical Theorem, Dept. ===

Our man Mal proposes a new statistical theorem, one which apparently states,

"Standard deviaton formulae CANNOT be applied to ---- > a sample of a population, IF ---- > the arguments in the sample ---- > consist of numbers all

.... Unless all arguments could not theoretically have been > 20."

Does that about capture the argument against using Wedge's BQS as significant data?  :- )

.

=== Skeptic's Corner, Dept. ===

1.  Could we see the proof for this theorem?  Previous generations of mathemeticians had not been using it.

2.  Any idea why my spreadsheet does go ahead and return a value for =STDEV on Wedge's BQS?  Does SSI purport to overrule the mathemeticians who wrote the =STDEV algorithm?

3.  Perhaps SABRMatt or Kelly want to calculate a (p=0.xx) coefficient for [Wedge's BQS 2004-09] vs [29 teams' BQS 2004-09]?

(Edit to add, I did go ahead run a ZTEST and get p=0.0000007 that Wedge's BQS numbers have been random.)

.

=== Implications of a Blown Quality Start ===

Also called into question was the issue of whether a large BQS total is a good or bad thing, since Bobby Cox has had a year or two with high BQS.

1.  Bobby Cox might have had a bad year managing his bullpens.  Bobby Cox' skill is not the "given" here.

2.  A single blown quality start might very well represent a correct call on a manager's part.

2a.  A long string of far too many BQS's prove -- by definition -- that a manager is routinely turning great starts into slop starts.

Supposing you wanted a simple, Zen-like number by which to analyze a manager's ability to choose the right RP's in spring, and to choose the right RP's with whom to relieve the SP, and to choose the moment at which to remove the SP.

Is there a stat you can suggest that would be superior to the simple BQS stat?

.

=== Simplify Dept. ===

For those not caring to wait for the numbers to come back:

1.  Of course Eric Wedge fouled up a huge number of good starts by his pitchers.  That's a given.

2.  Of course these large BQS totals were driven by non-random causes.  The probability curve will demonstrate Wedge's BQS "skill," just like it proves the New England Patriots' winning "skill."

3.  Eric Wedge's BQS "skill" is corroborated by his Pythag failures and his on-site reputation.

4.  M's fans better get their petition together, and quick.

.

Your friend,

Jeff