Info on Blog

Friday, April 27, 2007

Adjusted K/BB: My Thoughts

I got into a discussion about K/BB on Lefty's new site (oh, I guess I better change my link to him at some point...).

Here is what I posted, in two comments there, I thought it is interesting enough to share, plus you can poke holes in my idea :^) :

While I totally agree that K/BB is a key statistic to follow, it is going to be skewed a bit for the Giants because Sabean seems to want to corner the market on pitchers with high BB-rate/low BABIP and those pitchers often have low K-rates too.

Cain, Zito, Lowry and Ortiz have been lower than the .300 mean that most pitchers regress to, as someone noted above. Couple that with Zito, Lowry, Morris, and Ortiz's relatively low K-rate and you have a bad K/BB for the team.

Also, one should look at their BB/9 rate to see whether that is very bad or not. It is bad at 3.7 (ideally under 3), but again, low BABIP mitigates against a slightly higher BB/9.

I think the team needs an adjusted K/BB. I'm not sure how to calculate it but for a pitcher with a lower BABIP, you have some formula to "shift" an amount out of his BB into H to make up for the reduced BABIP, and that is his adjusted K/BB.

Doing a extremely rough estimate, a BABIP in the .275 range would result in about a 0.7 drop in H/9. Dropping the BB/9 rate from 3.7 to 3.0 vs. a 5.4 K/9, the K/BB is now 1.8, very close to the ratios Lefty noted.

That's not even accounting for the fact that a walk is not equivalent to giving up a hit, that should result in an adjusted BB/9 that's lower, though I have no idea how to proxy for that.

Trying a better example, using career numbers (Morris), I got a 0.68 drop in H/9 when the BABIP is .275. That's basically the same as above (1.79), but I think makes a stronger case about the ratio of conversion from H's to BB's, though still not accounting for extra-base hit effect.

Career BABIP:

Zito = .269
Cain = .244
Lowry = .289
Morris = .299
Ortiz = .289, but includes horrid last two seasons.

Doing my best to take out 2005 and 2006, Ortiz career BABIP is .284. That works out to about a drop of 0.44. But his K/BB is still horrible, about 1.5 after the adjustment.

Zito's very low BABIP works out to 0.78 less H/9. Career K/9 of 6.87, BB/9 of 3.53, K/BB 1.95. Adjusted BB/9 of 2.75 results in K/BB of 2.50, which is in the elite category (best are over 2.4).
Just thought of a proxy. TangoTiger has worked on expectancy tables and my 2006 The Hardball Times had the value of a single as .465 and the value of a walk as .315 (which they got from Retrosheet.org from work by Tom Ruane), which works out to be 1.48 times more valuable. Assuming you save at least a single and replace with a walk, that changes the team's ratio to 5.4 K/9 and 2.7 BB/9 or a K/BB of 2.0, which is the minimum you would want it to be. Zito's would fall to 6.87 vs. 2.38 or 2.89 K/BB.

Still not perfect, since the pitcher at some points avoids an extra base hit of some sort, but I would need a frequency table for those to calculate. But this is probably close enough, the degree of accuracy improvement probably is enough without adjusting for extra-base hits.

Taking Zito's 2006 season, which was a down year for his BABIP, he had a 6.15 K/9 and 4.03 BB/9 with a .280 BABIP for a 1.53 K/BB. He saved 13 hits which worked out to 0.53 H/9 and an equivalent of 0.78 BB/9. That dropped his BB/9 to 3.25 and his K/BB rose to 1.89 - not ideal but very close to the 2.0 target.

For his 2005 season, which saw a similar ERA but much lower BABIP, he had a 6.74 K/9 and 3.51 BB/9 with a .252 BABIP for a 1.92 K/BB. He saved 30 hits which worked out to 1.19 H/9 and 1.76 BB/9, which dropped his BB/9 to 1.75 and his K/BB jumped to 3.85, which is very good.

I also noticed another interesting factoid by looking at the value of various batting results. An intentional walk is worth just slightly more than half that of a regular walk: .315 vs. .176, which is 56% the worth of a regular walk. Thus when calculating these K/BB, particularly for NL starting pitchers, you need to reduce his BB/9 by the reduce value of an IBB, almost 50% less.

Ooops, maybe it is not worth the trouble. I thought there would be an increase in IBB for NL starters vs. AL starters, but Zito does not seem to have less IBB, he seems to have about the same as Morris or Ortiz, they had maybe one or two more IBB in a season. Not a heck of a difference between the two leagues.

2 comments:

  1. Another question might be what kind of factors lead to a below average BABIP. Are there things these kinds of pitchers have in common?

    ReplyDelete
  2. Here is the URL for Tom Tippett's great study: http://www.diamond-mind.com/articles/ipavg2.htm

    The only category that had lowered H/9 and that he named was "Crafty Lefties" but obviously Ortiz is not lefty or particularly crafty, I don't think. Plus there was the knuckleballers, where he profiled Charlie Hough a bit but he didn't separate them into a category. I don't think he was going for an exhaustive list of all the various types, just the main archtypes that most people know: power pitchers, closers, control freaks, crafty lefties.

    I know that Morris probably would fall under Control Freaks, as his BB/9 is very low, even if his K/9 isn't very high.

    Anyway, I don't think Tippett covered all the pitchers who could keep H/9 low, he just wanted to show that there are pitchers who can control H/9, but that they are in the minority (though significant), the vast majority cannot control H/9.

    ReplyDelete