Thursday, February 18, 2010

What does K/9 Measure

You will sometimes see the statistic k/9 (strikeouts/nine innings) called a pitcher's "strikeout rate". People will talk about this as "how often" a pitcher strikes batters out. The idea is that innings are a measure of opportunities a pitcher has to strike out batters and the strikeouts are how often they are successful. But, in fact, innings measure how many outs a pitcher got whether he had the opportunity to strike out 10 batters or 3. So k/9 actually measures the percentage of outs a pitcher gets  that are by strikeout. This makes the idea of it as "strikeout rate"  a little odd. We don't, for instance, divide a batter's home runs by their hits to determine their 'home run rate".

Does this matter? I think it does. Because there are two things that effect k/9. It goes up when a pitcher strikes a batter out and it goes down when they induce an out. A high k/9 may be caused by the ability to strike batters out or by the inability to induce batters to make feeble hits. The average ball in play  goes for a hit (minus home runs) about 30% of the time - for statheads the major league BABIP is about .300. If you are below average in that regard, as Randy Johnson was, you will need to strike out a higher percentage of the batters you face to be successful. But Johnson is a rarity, most successful pitchers are above average at inducing outs and the best pitchers are usually good at both. If you have a pitcher with a high K/9 you need to look at the hits they are giving up to determine whether that reflects a great strikeout artist or just an inability to get people out any other way.

7 comments:

NoDak Twins Fan said...

Hopefully all of the Twins pitchers have a good K/9 rating this season. Well... hopefully they are getting outs. Leave the K/9 ratings to the bullpen guys.

Anonymous said...

That's why looking at only one statistic (or looking only at statistics) doesn't give you the whole story. Everybody uses different stats and relies on personal observation to different degrees when they evaluate skill or predict future performance. You seem to be the only blogger fighting for the "believe what your eyes tell you, not the spreadsheets" argument.

So are you saying that K/9 is a metric that is useless in isolation so it is useless, or are you saying it's important to use it within the context of additional metrics?

TT said...

It doesn't matter what you use if you misinterpret its meaning.

David said...

What you point to is easily correctible by looking at total innings pitched and/or total strikeouts. Let's say pitcher X has a 12.0 K/9 rate but hardly ever gets past the third inning and ends up with only 60 innings pitched all year. No "stathead" in their right mind is going to say, "Wow, what a great pitcher." Rather, the "stathead" would look at the 60 IP and miniscule 80 K total, and conclude, "Huh, this guy has some ability to miss bats; the problem comes when he doesn't miss bats."

No one should ever rely on one statistic in isolation, but should rather understand what other statistics can clarify the meaning of the particular stat being used. In this case, the other statistics are pretty basic: mere strikeout total and total innings pitched will tell you whether the 12.0 K/9 rate is the mark of an elite pitcher or the mark of a guy who is simply a feast or famine flame thrower.

TT said...

David -

I am not sure I understand your point. Total IP tells you how many outs a pitcher got. Total K's tells you how many strikeouts he got. That is how you calculate k/9, but I am not sure how knowing the totals helps you understand its meaning.

"Huh, this guy has some ability to miss bats; the problem comes when he doesn't miss bats."

And that would be the wrong conclusion. Every pitcher has some ability to miss bats, but K/9 doesn't measure that ability. If a pitcher induces a lot of groundball outs, his k/9 will be lower regardless of his ability to miss bats.

That was the point of the article. Not that K/9 is useless, but that it is easily misunderstood and therefore misused.

David said...

My point is that strikeouts require more pitches (on average) than do other forms of outs. If you have a high strikeout pitcher, chances are they throw more pitches per inning. The problem would naturally be compounded if theoretical pitcher X was incapable of inducing outs any other way. By looking at total innings pitched and total amount of strikeouts, you see if the rate actually means something - i.e., is this an effective strikeout pitcher, or is this just someone who either gets a strikeout or gives up a hit.

And that would be the wrong conclusion. Every pitcher has some ability to miss bats, but K/9 doesn't measure that ability.

Every pitcher has the ability to miss some bats, but K/9 generally indicates that you have "stuff" that is particularly able to miss bats, more so than, say, Nick Blackburn (not that I mean to rag on Blackburn). If pitcher X has a 12.0 K/9, there's no question he's got pitches that are capable of inducing strikeouts fairly regularly; he just needs to either stop walking so many batters or get more help from his D.

If a pitcher induces a lot of groundball outs, his k/9 will be lower regardless of his ability to miss bats.

I don't understand your point. It's not like a pitcher has to have an incredible K/9 to be effective, but obviously his strikeout rate goes down as his other out rates go up...

TT said...

Just to clarify lets use an example. Two pitchers both face 20 batters and strike out 5 of them. One of them pitches six innings (i.e. gets 18 outs - 13 on balls in play). The other pitches 5 innings (i.e gets 15 outs - 10 on balls in play).

The second pitcher has the higher K/9, (5/15 compared to 5/18)but he was no more likely to strike a batter out than the first pitcher.

MLB Twins Updates