Different Way to Look at K-rate
-
- Major League Regular
- Posts: 63
- Joined: March 13th, 2010, 11:24 am
Different Way to Look at K-rate
The other day when Randy Wells gave up 5 runs without recording an out, I got to thinking about how this isn't reflected in his K-rate. If all 6 batters he faced hit the ball slightly differently, he'd have 0 Ks in 2 IP, but as it was, he could go and K the next guy and have 1 K in 0.1 IP (27 K/9). Hmmm. Would looking at Ks per AB be more beneficial in terms of determining true skill? Or K/40 AB to mirror the K/9 IP thing. Should a guy that induces a lot of grounders be down-graded in the K/9 department due to the increased amount of outs on balls in play, not to mention increased GIDP which each count as 0 K in 2/3 of an IP? Both of these hurt his K/9 without really reflecting his strikeout skill which I believe is the underlying factor we're looking for when using K/9. And it follows that BB per BFP (excluding IBB in the numerator and denominator) or something similar would be more useful than BB/9. The main issue I have, I suppose, is using IP as the denominator for these stats when it can mean so many different things. I'm just thinking out loud, feel free to point out the faults in this loose theory of mine. Just some food for thought I suppose, and since I'm thinking about this in 2010 I'm sure it's been analyzed to death for the past 10 years, and I've just never read anything about it.
- viper
- Hall of Famer
- Posts: 1480
- Joined: December 31st, 2008, 11:32 pm
- Preferred Style: Currently in an AL-only league with the Bill James Technical RCA as the single hitting category and ERA as the single pitching category.
- Contact:
Re: Different Way to Look at K-rate
K/BF does seem like a better indicator than K/9. The same could be said for BB/BF instead of BB/9.
The avalanche has started. It is too late for the pebbles to vote. -- Ambassador Kosh
Mike Ladd
Buffy, the Umpire Slayer
Mike Ladd
Buffy, the Umpire Slayer
Re: Different Way to Look at K-rate
Our projection process does exactly what you outline - we use K/batter faced and do everything as a percentage of plate appearances. There is absolutely no doubt that from a projection standpoint that K/BF is a better metric than K/9.
I think the real reason for the use of K/9 is it's an easy number to calculate and it rolls off the tongue and it prints easy on the page and is easy to understand. Plus, it's pretty close to as effective as K/BF over the long run.
Your double play example is a good one - but think of it another way.
Pitcher 1 goes 6 innings allowing 6 hits, walking 1 and striking out 4.
Pitcher 2 goes 6 innings allowing 10 hits, walking 4, and striking out 5.
Pitcher 1 K/9 is 6.0, K/BF is approximately .16
Pitcher 2 K/9 is 7.5, K/BF is approximately .16
Over the long haul, assuming BABIP stabilizes, using K/9 and K/BB will get you very close. But from a projection standpoint per batter statistics are absolutely better.
I think the real reason for the use of K/9 is it's an easy number to calculate and it rolls off the tongue and it prints easy on the page and is easy to understand. Plus, it's pretty close to as effective as K/BF over the long run.
Your double play example is a good one - but think of it another way.
Pitcher 1 goes 6 innings allowing 6 hits, walking 1 and striking out 4.
Pitcher 2 goes 6 innings allowing 10 hits, walking 4, and striking out 5.
Pitcher 1 K/9 is 6.0, K/BF is approximately .16
Pitcher 2 K/9 is 7.5, K/BF is approximately .16
Over the long haul, assuming BABIP stabilizes, using K/9 and K/BB will get you very close. But from a projection standpoint per batter statistics are absolutely better.
-
- Major League Regular
- Posts: 63
- Joined: March 13th, 2010, 11:24 am
Re: Different Way to Look at K-rate
Yeah I realize it's only really useful for smaller sample sizes, such as early in the season. But it would take a lot of guesswork out of the equation when finding SPs that are on a hot streak and who to snag off waivers. This is probably something I'll create a few spreadsheets with and then ultimately forget about it as it's too much work to constantly maintain.
Just for fun I had a look at what SP K/9 are the most deceiving on both ends of the spectrum. I subtracted HBP and IBB from BFP and multiplied by 38 to mimic K/9 IP. Ubaldo Jimenez and Adam Wainwright both had about 1 K higher on their equivalent scale (9.0 from 7.9 and 9.6 from 8.8, respectively) and Bud Norris, Charlie Morton, John Maine, Justin Masterson and Rich Harden took the biggest hits with each losing about 1 point off of their K/9. But as these K/9 are normalized, the BB/9 are also moved in the same direction. So the net effect isn't helpful if you simply look at K/BB I suppose.
Just for fun I had a look at what SP K/9 are the most deceiving on both ends of the spectrum. I subtracted HBP and IBB from BFP and multiplied by 38 to mimic K/9 IP. Ubaldo Jimenez and Adam Wainwright both had about 1 K higher on their equivalent scale (9.0 from 7.9 and 9.6 from 8.8, respectively) and Bud Norris, Charlie Morton, John Maine, Justin Masterson and Rich Harden took the biggest hits with each losing about 1 point off of their K/9. But as these K/9 are normalized, the BB/9 are also moved in the same direction. So the net effect isn't helpful if you simply look at K/BB I suppose.