THE BOOK--Playing The Percentages In Baseball

Tango,

Do the wOBA numbers presented above represent pass-thru counts or payoff counts?

They could only possibly be pass-through counts.

It is extremely rare that you would use at-counts, with Derek’s example linked one of the very few times you would do it.

I think looking at the impact of individual balls/strikes, beyond the average value of +/-.07, also tells an interesting story:
B1 0.033
B2 0.058
B3 0.115
S1 -.062
S2 -.080

Strike 2 is important, but ball 3 is what really shifts the balance of power most dramatically.

They could only possibly be pass-through counts.

Is that because the results make this obvious or just because you never dwell in action counts?  I wasn’t convinced one way or the other, based on the results, but I didn’t give it a lot of thought.

The example that Derek gave is one that I’ve thought about in the past.  Obviously it’s advantageous to pitch ahead in the count.  I was surprised by how little a difference there was in the xBABIP for those who were at the extremes of count management.  It would have been interesting to see how well the count-based xBABIP correlated with actual BABIP.

Sorry if I’m being dense, but I don’t understand your penultimate paragraph, the one beginning with “by the way.” Can you explain it a bit more?
Thanks.

Strike 2 is important, but ball 3 is what really shifts the balance of power most dramatically

This makes alot of sense.  These are the required tipping points for non-BIP outcomes.  You can’t strikeout without first getting to 2 strikes and you can’t walk without 3 balls.

Action count (at counts… I prefer your term actually) are useless unless they are used in a broader pass-through count process.  Taking a ball or strike matters, as we can see above.

weskelton/#4
The correlation is very weak between count-based xBABIP and actual BABIP.  r-squared is 0.002 for pitchers with 300 BIP (2007-2009).

Derek,

Thanks for the reply.  That is a little surprising.  I thought for sure there’d be somethig there.  I guess once you take BB, K and HR out of the equation, and advantage in the count doesn’t mean as much.  I wonder if you do any better trying to correlate a count-based, expected SLGIP?

A big reason count can’t have much affect on BABIP is that balls are mainly put into play on relatively neutral counts, while few BIP come on extreme hitter or pitcher counts.  Just back-of-envelope estimates of 2009 BIP:

0-2 7800
3-1 4300
3-0 233

0-0 19,500
0-1 16,200
1-1 15,900
2-2 14,800

Good point Guy.  You guys can do what Guy did by going to the B-R.com link, look at PA, and subtract out the HR, BB, HB, SO and SH numbers.  ah, what the heck, here I’ll do it:

18901 First Pitch
15776 1-1 Count
15585 0-1 Count
14533 2-2 Count
14119 1-2 Count
12782 1-0 Count
10762 Full Count
9848 2-1 Count

7505 0-2 Count
4492 2-0 Count
4056 3-1 Count
223 3-0 Count

So, you can see that when it would matter most, that’s when it occurs the least.

Thanks, Tango.  Next time, use the AB column and just subtract HR and SO!

Don’t forget about SF.

Good stuff guys. Not surprisingly, neither pitcher nor batter is interested in being in a big disadvantage situation, and well that generally is what separates effective pitchers and not ... The ability to get into “pitcher’s counts” without throwing too many “crushable pitches” (centered).

I know the BABIP by count numbers are not impressive, and we heard the same thing regarding BABIP by pitch location and BABIP by type of pitch thrown, etc.  But all of these “almost” not worth mentioning BABIP factors can be adding up to make a solid dent in the .020 difference we see in some good and bad BABIP.  Of course pitcher handedness, K rates, defense, and luck are bigger factors but should we disregard all of these “little” things that may be adding up to effect BABIP in statistically significant ways altogether?  I am not sure.