Thought Experiment

by Graham MacAree on Sep 30, 2009 12:29 PM PDT

Pitchers A and B strike out 22% of the batters that they face and walk 7% of the batters that they face. They play on teams with DERs of 0.715 and 0.676 respectively, allow no home runs, and no hit batsmen. Can you tell me the K/9 and BB/9 of each?

A: 8.4 K/9, 2.6 BB/9

B: 8.8 K/9, 2.8 BB/9

Let's stop pretending K/9 and BB/9 are defence-independent. No, it's not a big deal, but we should still be looking at these things as a percentage of batters faced rather than depending on outs.

Tweet Comment 43 comments | 1 recs |

Do you like this story?

Comments

This line of thought should be commonplace

As a child, I remember the stupid defense catching foul balls and costing Randy Johnson strikeouts

by Poochie on Sep 30, 2009 12:34 PM PDT reply actions

carrying it to the extreme ...

if a team had no defense at all, strikeouts would be the only outs. Then every pitcher for that team would have a K/9 of 27.

by Steve Nelson on Sep 30, 2009 12:51 PM PDT reply actions

A baserunner could get hit with a batted ball.

by Manzanillos Cup on Sep 30, 2009 12:56 PM PDT up reply actions

However there would also be batters reaching base after wild pitch strikeout

If more batters reached base that way than were eliminated being hit by batted balls, the K/9 would be greater than 27.

by Steve Nelson on Sep 30, 2009 12:59 PM PDT up reply actions

Wild pitch strike outs would be entertaining games of tag.

by Manzanillos Cup on Sep 30, 2009 1:15 PM PDT up reply actions

and then K/9 would be defense independent (and defence independent as well)

by Steve Nelson on Sep 30, 2009 1:34 PM PDT up reply actions

This sounds like an interesting way to play baseball

Yuniesky Betancourt would be a God at it

by seattlebruin on Sep 30, 2009 1:37 PM PDT up reply actions

Not sure I understand this.

I understand that you could have a K/9 of 9 by striking out one batter per inning but have varying strikeout percentages because the batters faced would be different. Say in a 9 inning game, seeing only 27 base runners would be a 33% strikeout rate whereas facing 40 batters in the same 9 inning game would only be a 23% strikeout rate.

What I am trying to figure out is how defense effects K/9.

Fans are typically idiots.

by The Typical Idiot Fan on Sep 30, 2009 1:46 PM PDT reply actions

If a pitcher's defense makes more outs on balls in play, the pitcher has less opportunity to strike batters out

Let’s say we’re playing the Twins and Felix is pitching with two outs. Joe Mauer hits a line bullet to center field.

In Scenario A, Franklin Gutierrez makes a great diving catch, robbing Mauer and ending the inning
In Scenario B, the ball falls in front of Jeremy Reed, and Felix strikes out Jason Kubel to end the inning.

If we assume that all other outcomes are equal besides these two, Felix will always have a higher K/9 in Scenario B since K/9 is based on innings pitched, and we know that innings pitched is partially a function of defense.

by seattlebruin on Sep 30, 2009 1:51 PM PDT up reply actions

A shitty defense allows the pitcher to face more batters per inning, meaning you could have a higher K/9

even though your K% would be the same or lower.

by marc w on Sep 30, 2009 1:51 PM PDT up reply actions

Speculation

More outs resulting in non strikeouts lowers the chance of any of the 3 outs in an inning (or 27 outs in a game) from being strikeouts?

Wouldn’t that also effect strikeout % tho? Since, theoretically, a stronger defense would turn more balls in play into outs, wouldn’t that be fewer batters faced overall and / or fewer batters faced that result in strikeouts? Or is the percentage not effected as strongly as it is for inning based rates?

Fans are typically idiots.

by The Typical Idiot Fan on Sep 30, 2009 1:51 PM PDT up reply actions

This is true as well, but adding to the numerator will always make a bigger difference than adding to the denominator

essentially, adding that extra strikeout in the numerator of K/9 will make a much larger difference than the reduced at-bats for the K% calculation.

by seattlebruin on Sep 30, 2009 1:53 PM PDT up reply actions

Ah, I see what you mean.

by Teej on Sep 30, 2009 1:56 PM PDT up reply actions

K% is not affected by defence

by Graham MacAree on Sep 30, 2009 1:59 PM PDT up reply actions

How is that?

Speculation 2:

Pitcher A faces 30 batters in a 9 inning game and strikes out 4.
Pitcher B faces 30 batters in a 9 inning game and strikes out 6.

The difference was two exact same foul ball plays that pitcher A’s team got but pitcher B’s team dropped, forcing pitcher B to record two more strike outs. Pitcher A records a 20% strikeout rate while Pitcher B has a 13%.

This isn’t theoretically possible?

Fans are typically idiots.

by The Typical Idiot Fan on Sep 30, 2009 2:04 PM PDT up reply actions

Small sample size

you assume over the course of a season that even if the defense is excellent or shitty, the pitcher has ~the same chance to strike out any given batter.

by seattlebruin on Sep 30, 2009 2:05 PM PDT up reply actions

Well of course. I'm merely speaking hypothetically.

Fans are typically idiots.

by The Typical Idiot Fan on Sep 30, 2009 2:06 PM PDT up reply actions

It didn't 'force' pitcher B to do anything

by Graham MacAree on Sep 30, 2009 2:05 PM PDT up reply actions

Okay.. different term...

Allowing Pitcher B another chance to strike those batters out. Whereas Pitcher A had plays made for them on balls in foul territory.

Fans are typically idiots.

by The Typical Idiot Fan on Sep 30, 2009 2:08 PM PDT up reply actions

The extra strikeouts don't come from the defence dropping the ball, they come from the pitcher striking the batter out

Foul balls are a weird non-PA ending defensive chance that I hadn’t really thought about, but the point holds. K% is the measure of skill, K/9 is not

by Graham MacAree on Sep 30, 2009 2:14 PM PDT up reply actions

And if pitcher A and B were equally skilled in getting strikeouts, one would expect an extra quarter of a strikeout for B

by Graham MacAree on Sep 30, 2009 2:16 PM PDT up reply actions

I agree with the conclusion.

I realized that the only situation I could come up with wherein the defense could be a factor was on foul balls in play. Then i started wondering about whether stadiums with larger foul areas would make that situation more likely, since there’s more ground to cover and a greater chance at making plays on foul balls.

Perhaps I’m just thinking about it too much.

Fans are typically idiots.

by The Typical Idiot Fan on Sep 30, 2009 2:17 PM PDT up reply actions

K% is a probability, K/9 is not

by Jeff Sullivan on Sep 30, 2009 2:16 PM PDT up reply actions

Good defense lowers your K/9 because you face fewer batters, since your fielders are making outs.

Put it this way: If you get a ground ball that your shortstop makes the play on, you only have two more possible outs this inning to get a strikeout. If the shortstop muffs the ball, you have three more opportunities.

by Teej on Sep 30, 2009 1:52 PM PDT up reply actions

Expand K/9

The denominator is 27*Outs, or 27*PA*[1-(K%+BIP*DER)]

by Graham MacAree on Sep 30, 2009 1:54 PM PDT up reply actions

Fangraphs has strikeout %.

Fans are typically idiots.

by The Typical Idiot Fan on Sep 30, 2009 2:08 PM PDT up reply actions

Only for batters

by vivaelpujols on Sep 30, 2009 5:43 PM PDT up reply actions

My fault, Fangraphs has K%

by Manzanillos Cup on Sep 30, 2009 2:08 PM PDT up reply actions

David gets asked this a lot

And his answer, for better or worse, is that it doesn’t really matter, so he has no incentive to change it. There’s no difference in projecting future strikeout performance whether you use K/9 or K%, which is what 99% of all applications of the statistic are for.

by davidcameron on Sep 30, 2009 2:35 PM PDT up reply actions

People hate change

by Matthew on Sep 30, 2009 3:09 PM PDT up reply actions

That is one way in which the Hardball Times' K/G is kind of nice

It’s essentially K%, but normalized to a scale similar to K/9. If you started using K/G, most people probably wouldn’t even notice until they saw some discrepancy between K/9 and K/G, at which point it becomes necessary to explain why weighting things per batter makes more sense than weighting things per inning.

by ubelmann on Sep 30, 2009 3:32 PM PDT up reply actions

Why do people like nine innings as the denominator for individual pitcher stats so much?

This has always bugged me about RA and ERA. Nine innings is kind of a convenient arbitrary standard, because nine innings per game and all (which makes sense when you’re measuring team stats), but starting pitchers throw complete games about 3% of time these days. RA is not a good estimation of the number of runs a SP would give up if he had to throw 9 innings per game.

I think it would be cool if a starting pitcher triple slash line caught on – (IP/GS) / (R/GS) / (K/GS). Context is easy for a casual fan to grasp, and gives you a good idea of what you’re likely to see in the box score after the start. It’s not K%/BB%/GB%, but I could go for it as a “mainstream” stat.

by Manzanillos Cup on Sep 30, 2009 4:20 PM PDT up reply actions

It isn't a big deal in most cases and it's more inuitive

by vivaelpujols on Sep 30, 2009 5:42 PM PDT up reply actions

It's only more intuitive because people are indoctrinated into it

by Graham MacAree on Sep 30, 2009 5:55 PM PDT up reply actions

And because it's on the same scale as ERA.

by vivaelpujols on Sep 30, 2009 6:04 PM PDT up reply actions

.

http://www.lookoutlanding.com/2009/9/30/1062442/thought-experiment#22071152

by Aaron Campeau on Oct 1, 2009 1:09 PM PDT up reply actions

I think R/9 is more intuitive than R%

by vivaelpujols on Oct 1, 2009 1:22 PM PDT up reply actions

The other problem is that K/9 and BB/9 are interrelated

Take Zack Greinke and Clayton Kershaw.

9.55 K/9, 0.27 K% — Greinke
9.55 K/9, 0.26 K% — Kershaw

The difference, of course, is that Kershaw walks a lot more guys than Greinke, so he faces more batters per inning, giving him more chances for a strikeout each inning.

Is it a small difference? Sure, it’s probably pretty small, but it’s just one more reason to move to K% (or K% normalized to something else, like The Hardball Times’ K/G.)

by ubelmann on Sep 30, 2009 3:30 PM PDT reply actions