Pqs

[dwc]

I find your PQS interesting and the running PQS in the player link helpful, but I have a question. What is the rationale for including K/IP and BB/IP into PQS? The ultimate goal is to prevent the other team from scoring, and K and BB are a means toward that end. That is, it seems to me that K and BB could be evaluated as predictors of quality starts (as ordinarily defined), rather than incorporated into the definition of a quality start. Does this make any sense, or am I missing something?

Didn't know where exactly to post this. Let me know if this sort of question belongs somewhere else.

The simple answer is that K/IP and BB/IP are better predictors of ERA and WHIP than past scoring (ERA). Both the Fundamental Skills section and the Pure Quality Starts blurb in the Pitching Toolbox address this issue: http://www.baseballhq.com/subscribe/...tool_pit.shtml

[Apple Pie]

I watch K/IP and BB/IP almost as much as any other combination of two pitching stats. They tell you what "stuff" a pitcher brings and trends in those two are often good predictors of future performance (all other conditions considered -- park effects, opponent, [hidden] injuries).

[PD@HQ]

I watch K/IP and BB/IP almost as much as any other combination of two pitching stats.

I like these combined as Command: K/bb. It's my favorite all-purpose at-a-glance pitcher rate stat.

[Apple Pie]

Yes, but Command blends the two. I like looking at them separately. If K/IP is going down while BB/9 is also going down Command stays the same. But the declines in the two shows a "thrower" learning how to pitch.

[dwc]

I also like separating K/IP and BB/IP, for the same reasons that Apple Pie gave. Given a 2.0 CMD, I'd prefer 6/3 K/BB ratio. Higher BBs mean too many baserunners, lower Ks mean too many balls in play.

But that wasn't really my question. I separate variables into "predictor" variables and "outcome" variables. So we could use K/BB as a predictor of quality starts, as traditionally defined (6 or more innings, 3 or fewer ERs, period). If a guy begins to have higher K/BB (we can call him Nate Robertson) before he posts many quality starts, K/BB is an effective predictor (as it was in Robertson's case). It seems to me that if you put K/BB as part of the outcome, then it muddies the analysis some.

[ANDY@HQ]

Originally posted by dwc
... So we could use K/BB as a predictor of quality starts, as traditionally defined (6 or more innings, 3 or fewer ERs, period).

DWC,

By my reading of your post, you seem to want to "predict" Quality Starts. I would submit that Pure Quality Starts (PQS) is a better measure than just Quality Starts (QS). Ron has written on this in an article titled "Making Informed Decisions on Starting Pitchers":

http://www.baseballhq.com/subscribe/charts/splogs.shtml

And, if my understanding is correct, I would guess that K/BB is a predictor of the outcome variable PQS (in a statistically significant sense, when doing statistical modeling), even though it is not strictly part of it's defineition.

I hope that helps.

[Tom_T]

What about Ron's research that showed generally poor ERAs for pitches with good command but low k/9 rates?