The influence of batters’ expectations on pitch perception

Hi everyone. This is my first post on The Hardball Times Live and I’m really excited to have to opportunity to post here. I hope you enjoy my work. You can find more of my original research at my personal site.

What we’re going to look at today is the relationship between percentage of pitches on a certain count and batting eye by pitch type. You can read more about batting eye at my site but to summarize – it is a measure of how good the batter is at judging whether a pitch is in the strike zone.

Batters are obviously much less concerned with taking a strike when they have less than two strikes on them – and may choose not to swing at pitches which they know are strikes in these counts. Therefore looking at batting eye on most counts is an imperfect measure. For this reason we are going to look only at two strike counts for this analysis.

Unfortunately this gives us a mere four data points per pitch type, which means that the correlations I’m going to show you are based on a very limited number of points. On the plus side each point represents a few thousand pitches. Additionally, while this data is for the 2008 season, the numbers look very similar for 2009. I have also done a bit of follow-up analysis which will allow me to look at more of the points at the same time.

Let’s take a moment and think about what we might see in this analysis. I came up with a couple of different theories.

-You might suspect these measures are completely unrelated. After all a curveball is a curveball regardless of count and the batter should be consistent in their ability to tell if it’s going to be a ball or a strike.
-Perhaps pitchers can outsmart batters and throw pitches which the batters are not expecting.

Since I’m writing the article you can pretty much rule out the first theory and while some pitchers might be able keep batters off balance it doesn’t seem to be the case when we average over the league.

You might now conclude that the two measures are directly related – that the more frequently a type of pitch is thrown on a specific count the more a batter will be expecting that pitch type and the better they will read it. You’d be right – with one exception. Check it out.

We see that this strong positive correlation between percentage of pitches and batting eye exists for three of the four main pitch types. Batters seem to be better at reading fastballs, curveballs and sliders on counts where those pitches are more likely to be thrown. But what’s up with the change-up?

The change-up is generally thrown as a complement to a fastball. After seeing this data I thought that correlating batting eye on change-ups with percentage of fastballs thrown might be interesting. Here we see a pretty strong negative correlation.

image

I want to include one caveat here. Batting eye on change-ups also correlates strongly with percentage of sliders and curveballs thrown. In fact these correlations are a bit stronger. Comparing change-ups to fastballs makes the most sense to me intuitively, but the batter could just be looking for more off-speed pitches which might help them pick up change-ups.

Overall I think that there is ample evidence to support the following conclusions:

-Batters are better at judging whether a pitch will be a ball or strike on counts where they expect that type of pitch to be thrown.
-Batters do a good job of anticipating what percentage of each pitch type will be thrown on a specific count.

The one exception to these rules is the change-up, which batters seem to be better at picking up when they expect less fastballs (and more off-speed pitches.)

I’m really looking forward to delving deeper into baseball and the tendencies and strategies behind it. I hope you’ll enjoy reading my posts as much as I’ll enjoy writing them.


15 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments
Mike Fast
14 years ago

Excellent work, Craig!  Welcome to THT.

Dave Studeman
14 years ago

Wow, that is very cool, Craig.  Thanks for the post.

Nick Steiner
14 years ago

This is over my head Craig.  I’ll read this more in detail later tonight, so I can hopefully be on the same page as you!

Craig Glaser
14 years ago

It’s kind of technical and I’ve done a writeup at my site which I link to in the article which should help a lot.

If anyone has any questions please feel free to email me.  I’m going to try to write up a little ‘101’ series as I get the time and post it on my site because it seems like a lot of people want to understand at a deeper level.

Pizza Cutter
14 years ago

Craig, next stop: batting eye on sequence of pitches?  For example, batting eye on CU that follow FB?

Craig Glaser
14 years ago

I definitely have sequence of pitches on my radar.  I think my next article will be about selectivity (hopefully next week) but I’m definitely planning to look at sequence

Guy
14 years ago

Another way to look at this is to assess batting eye separately for pitches in the strikezone and out of the strikezone, at different counts.  What I think you’ll find is that batting eye on strikes improves as it becomes more likely the pitcher throws a strike, i.e. as the count increasingly favors the hitter.  But batting eye on balls out of the strikezone probably does not increase as much, if at all.  As the likelihood of a strike increases, the hitter primes himself to swing more frequently. 

Let’s say pitchers throw a strike 50% of the time at 1-2, and hitters swing 50% of the time.  With zero judgment, hitters would be right 50% of the time, whether pitch was a strike or not.  But if pitchers throw strikes 70% of the time at 3-2, and hitters swing 70% of the time, then hitters will be right 58% of the time and they will be right much more frequently when a strike is thrown. (I’m not saying hitters swing randomly in real life—just that it’s easier to be right when pitchers are forced to throw strikes.)

That may be all you are measuring here:  that hitters are more likely to swing as the count improves for them (higher number of balls).  Since FA’s are more likely to be strikes, hitters will appear to be more “correct” in their judgment of FAs, but not on breaking balls because they are more likely to be out of the zone.  It may have nothing at all to do with “reading” types of pitches, but just that at 3-2 count pitchers have a big incentive to throw a strike, and hitters—knowing that—are very likely to swing.

Craig Glaser
14 years ago

Hi Guy,

I don’t think we’re that far off from each other in terms of theories here actually.  The one thing you might not realize is that batting eye takes both strikes and balls into consideration and that there’s no way to look at it for one isolated.  It basically looks at your ability to judge and separates it apart from your actual decision making process.

A higher batting eye score allows you to both swing at more strikes and swing at less balls – not one or the other.  If you were trading off it would show up more in the selectivity measure.

I think we’re not too apart theory wise, though.  And I think my follow up about selectivity might fill out some of the gaps.

Guy
14 years ago

Hey Craig:  Thanks for reply.  Can you post an explanation (or link) to your definition of batting eye?  And an explanation of why it couldn’t be calculated for strikes alone (if that isn’t obvious from the definition)?  I’ve spent a bit of time at your site, but couldn’t find a definition.  Thanks.

Craig Glaser
14 years ago

This is probably the best explanation I’ve put up so far: http://sabometrics.com/?p=242 but you probably saw that.

Basically what batting eye is is the difference between your perception of strikes and balls.  That’s why it can’t be calculated for just one or the other. 

I think the fact that there is such a heavy correlation between %age thrown and batting eye is pretty strong evidence for my argument.  Especially when you look at the z-scores and combine all the points.

I am planning to write up more of a 101 version of the theory as it applies to baseball.  Hopefully that’ll help.

Guy
14 years ago

I saw that, but I still don’t see how Eye is calculated.  In any case, one clearly could calculate a kind of error rate separately on strikes and balls.  My guess is you would find that it’s pretty similar for different pitch types at any given count (i.e. a hitter swings at about the same % of FA/strikes as SL/strikes at 3-2). 

The fact that batting eye and selectivity are almost perfectly and negatively correlated is strong evidence for my thesis.  That indicates that what hitters mainly change as the count shifts (and the likelihood of a subsequent strike goes up or down) is their proclivity to swing.  The more likely a strike, the more frequently they swing.  And the more they swing, the WORSE their eye rating becomes—because most pitches at most counts are strikes.  This correlation between strike% and swing% would produce exactly the results you find by pitch type, even if there is in fact no change in hitters’ ability to “read” a particular pitch type.  So at a minimum, I think you need to control for this before reaching conclusions about pitch types.

I also wonder whether there is much utility in the batting eye metric as you’ve constructed it.  It has a hugely negative correlation with OPS by count.  Is that true by hitter as well?  It appears from the data at your site that Pujols’ batting “eye” is no better than Francouer’s.  But he is of course vastly more selective.  Do we know if a high batting eye rating has value for a hitter?

Craig Glaser
14 years ago

I think once I post my follow up it will make a little more sense (plus I’m hoping to better explain the actual calculations behind the stats so hopefully that will help more as well.)

There definitely is interplay on a league average level between batting eye and selectivity which does complicate things.  I’m looking for the strongest correlations which make the most sense to me right now but it will be hard to prove causality either way.

I will say that to me, the way it looks right now is that for different types of pitches different factors are most responsible.  For example if I correlate strike % with batting eye for fastballs there is almost no correlation whereas with some other pitches there is a very strong correlation.

So my main advice is to be patient and wait for the next article (which I’m hoping to have up this week at some point) and then we can continue to discuss and explore the possibilities.

Thanks for the comments though it’s great to have feedback and input!

Craig Glaser
14 years ago

One additional point is that you are not penalized for swinging more.  In fact swinging more at strikes actually improves your batting eye rating.  It is the difference between % swinging at strikes and % swinging at balls which determines the rating.  Not sure if that was clear or not.

Guy
14 years ago

Craig:
Looking forward to your further explanations. 

You may want to look at his interesting analysis by Dave Allen, which shows batter wOBA based on whether a hitter swings/takes a first pitch strike/ball:  http://www.fangraphs.com/blogs/index.php/a-last-look-at-first-pitch-aggressiveness/.  What this shows is that the two types of “error” are not equivalent:  swinging at a ball is worse than taking a strike.  And this is also true for successes:  taking a ball is more valuable than swinging at a strike. (Of course, these values will vary a lot by count.) 

This pattern is why your selectivity metric is associated with good hitting:  successfully taking pitches out of the strikezone is hugely valuable compared to swinging.  But the difference between swinging and taking on a pitch in the strikezone is much smaller.  If your batting eye metric treats both forms of success, and both forms of failure, the same—as it sounds like it does—then it’s not clear that it measures something very meaningful.

Best Fake Yeezys
6 years ago

A fixture at Chicago Bulls games for Michael Jordan’s career in the Windy City, Schmidt acquired many rare pieces of memorabilia from his career, including these game-worn “Cardinal” Air Jordan 7s from early in the 1992-93 season