Strikeouts a measure of skill or borderline irrelevant?

[PVNICK]

In reading the assorted threads I see that the strikeout has a strange relationship with a player's value. If one is a hitter, for the most part it is dismissed as being of trivial significance but no real mark either for or against or in terms of evaluation. The argument goes so and so formula shows a strikeout is -.00x vs another out and beats a GIDP.

When it comes to pitcher's, the strikeout is one of the key factors in many forms of Sabermetric evaluation, along with BB and HRA. It is safe to say that if Wang had exactly the same numbers but with 50-75 K instead of other outs he would be held in much higher regard.

So we have the inherent contradiction. Either the strikeout is virtually worthless as compared to other outs or it isn't. The argument for its importance in measuring a pitcher can be that the K reduces balls in play which makes it impossible for a hit, error, etc. A "contact" pitcher with more balls in play will necessarily have to rely more on his defense and/or luck. I've seen books that claim and cite data showing that a pitcher has no control
over BABIP and it is totally random from season to season.

I'd like to hear what the board has to say and see a cogent argument for this apparent contradiction.

[SABR Matt]

Here is the best way I can think of to explain the difference.

Batters have inherently more control over the result of a batted ball than pitchers do. This can be seen very easily by looking at the spread of BABIP among hitters vs. pitchers (once you correct for that pitcher's defense). The batter gets more information about the impending "play" when he makes his aggressive move toward altering the game than the pitcher does. The pitcher can only control how he releases the ball...he can't react to the batter's reaction to his pitch. IOW, aside from scouting reports on hitter tendencies and gut feelings and the like, the pitcher has no real information on what the batter will do when he throws a certain pitch. As soon as the pitch is released, the pitcher's job is over until the next pitch (barring a comebacker to the mound of course). The batter on the other hand gets to see the pitch, change his approach in order to hit the pitch, and hit the pitch in the general vicinity of where he thinks it is best to hit it. You can see this very visually when you watch Ichiro hit. He'll be watching te pitch as it comes in...imagining that the ball is on some imaginary string and deciding (in split seconds) what kind of swing he should use and where he would like to hit the ball.

Given that batters have more information about each pitch than pitchers do, it makes sense that batters should have more control over BABIP than the pitcher. Given that the batter has more control over the ball in play distribution, a strikeout for the batter is less harmful in the grand scheme of things for the batter because he is actively exerting some influence over the batted ball when he does make contact. For a pitcher, every time he allows a batted ball, he is excepting significant risk, because his ability to induce bad contact is far outstripped by each batter's ability to use contact to produce hits. Wang is a rare bird because his success doesn't come from DIPS categories *OR* the BABIP. His success comes from limiting extra base hits and getting lots and lots of double plays (because of the high GB rate). Most pitchers, however, need to exert control over the game by having success in the DIPS events because, unlike Wang, they have almost no control on what happens if they fail to strike out the batter.

[PVNICK]

That makes sense except if the batter has more control or the pitcher has no control once it is released why the anomaly as to who gets credit and blame for the K. It's a + for a pitcher and a neutral event for a batter (vs. another type of out).

[Ubiquitous]

Strikeouts are only neutral if the batter can prove he can do the other things to make strikeouts neutral. If you got some kid who you called up who is batting .220 with a .250 SLG and his K'ing every other at bat then the strikeout isn't neutral. It is a pretty clear sign that the batter is not able to make contact at this level and needs to adjust.

Now then if you got a guy who is batting .305 with a .570 SLG but K's 18 to 20% of the time is the strikeout really a problem for him? Wouldn't you say that he is making consistent contact with the ball?

[SABR Matt]

Right. The reason we pay special attention to Ks for pitchers is that we can't prove whether a pitcher's BABIP is entirely his own doing...the BABIP for a batter is much more strongly correlated from year to year than the BABIP for a pitcher. If you get a pitcher who has a 3.50 ERA because he induced a .225 BABIP one year (extreme example for emphasis), you can't really say that the next season he's going to again induce a .225 BABIP. If you have a hitter who hits for a .225 average, it's more likely to be the fault of the hitter than any kind of luck/defense factor.

[digglahhh]

I used to find this to be a problem, but I have since decided that it is not, in fact, an inherent contradiction.

The most important point here is we are not using the same way in the two measurements. We are not saying that the pitcher who K fewer batters in less effective in a material sense, we are saying that his success can be less attributed to him, and only him. In one sense, that makes him less of a "known quantity" as fluctuations in skill level of team defense will have a more profound impact on his performance.

Basically, the fewer defensive independent outs (Ks) a pitcher receives, the less we can confidently assert that he is responsible for his own success.

Another way to look at this is that league-wide BABIP is usually in the .300 range, which means the best way to lower overall BA would be to reduce amount of balls in play, lowering the ABs that even have the possibility to turn into a hit

From a hitter's perspective, it is not that a K isn't less valuable in than, say, a groundout. It is just that, over the course of time, the fact that the DP is so much more worse than a K, than a K is than a 6-3 (essentially the same with nobody on), that the OVERALL value of the two events aren't so different.

Obviously, this changes from hitter to hitter. Speedy guys are less likely to turn grounders into DPs, sluggers more likely (and more likely to K)...

From the macro-sense in which this model addresses the game, I don't have a problem with it. In the micro-sense, there are plenty of times when you are watching a game and you know that just putting the bat on the ball is a positive thing (in relation to a K).

There are plenty of situations in which a ground ball is better than a K, but it is never as much better than a K, as a K is to a GDP. The weightier but less frequent GDP negatives basically negate the more frequent but less weighted positive of the 6-3 to the K.

There are very subtle advantages to situations that seem even though, for example exchanging a slow runner for a fast one on an FC. Things like that won't show up in linear weights.

[Tango Tiger]

There are very subtle advantages to situations that seem even though, for example exchanging a slow runner for a fast one on an FC. Things like that won't show up in linear weights.

.03 runs

You're just not using the full version!

[digglahhh]

.03 runs

You're just not using the full version!

Don't be ridiculous, Cousin Larry.

I'm not looking at any version, and the fact that it may appear in a full version doesn't change the fact that it is not measurable. It just means that somebody attempted to measure something that can't be measured. Check my blog for my "measurement" of the number of grains of sand in the Mojave, must be measurable, because I have contrived a value...

So, trading any slower runner for any quicker runner is worth .03 runs?

Ramon Castro for Shawn Green, Ramon Castro for Jose Reyes... obviously the same relative advantage...

I'm talking in the micro sense, one discreet event. That effect is not measurable.

Linear weights are irrelevant in the sense of an individual game, or situation. They can be contradicted over and over without changing their elemental truth on their own terms. This is apples to oranges.

[538280]

I agree, Diggs. I don't understand how you can so easily generalize that "fast for slow" is worth a specific amount of runs. Fast and slow are general terms, they don't give precise measurements on what is coming in and going out. I also agree that the value of linear weights is that they are in the aggregate. They aren't useful for individual games, they only make sense when you realize that they are totally in the average.