Getting under it

Of all the new baseball insights that HITf/x will bring, both the simplest and most exciting to me is being able to tell between a fluke hit and quality contact. One of the great leaps in modern baseball analysis has been trying to look past the wild short-term fluctuations in results to look how a player is really performing at in areas they can control. Until now, as soon as a ball leaves the bat, we have to use huge generalizations like fly ball, line drive, and ground ball rates to guess at how well a batter has succeeded in making solid contact. Soon we should be able not only to distinguish between batters making quality contact and batters getting lucky, but also to diagnose what has changed when something goes wrong with a hitter. Is your favorite slumping slugger because he’s just not hitting the ball as hard due to injury, or is he not squaring up the ball because his eyes are going?

Unfortunately that time is still some way off, as there is only data from this April to play with so far. We can look at who hit the ball the hardest as Mike Fast has done for that month, but we can’t say anything about individuals without previous seasons (or even months) to compare with, and it’s hard to break down the different kinds of balls in play any further than fly ball/line drive/grounder with such a small sample.

One thing we can do, however, is look at what makes for absolutely terrible contact. THT’s own Colin Wyers has been working on how loft time and distance affects how likely a play is to be made on a batted ball, but what I want to look at is what happens when a batter puts absolutely no pressure on the defense at all by hitting an infield pop-up.

Why on earth would you be interested in that, the reader asks? The infield fly is probably the least interesting play to watch, a guaranteed out with no possibility of any action on the field (Luis Castillo on game-ending plays vs. the Yankees being a possible exception). But exactly that makes them the ultimate accomplishment for a pitcher, and the worst fate for a batter – they are like a one-pitch strikeout, with no chance of finding a hole or being beaten out like even the weakest of grounders. If there are pitchers who have a repeatable ability to induce them, that’s a major part of their effectiveness that doesn’t show up on the box score. Similarly, if a batter starts racking them up, then there’s an obvious reason other than just luck why so few of his balls are falling for hits. So let’s look at what angle a batted ball becomes one of these automatic outs:

Angle              DER
0 -> 5            .515
5 -> 10           .378
10 -> 15          .205
15 -> 20          .369
20 -> 25          .564
25 -> 30          .747
30 -> 35          .833
35 -> 40          .916
40 -> 45          .927
45 -> 50          .945
50 -> 55          .976
55 -> 60          .981
60 -> 65          .987


Since we’re just going to focus in on popups, I have ignored how hard the ball left the bat (let alone the spin on it). That shouldn’t help balls hit straight up anyway, but it makes the results for the lower angles unreliable. Still, clearly fewer balls drop in for the batter as the loft increases from about 11 degrees, and batted balls hit at higher than 50 degrees become as automatic an out as there is in baseball. It’s a reasonably common result: Of all batted balls in April, 9.2 percent of them were hit for harmless flies. How harmless? Of the 1309 balls hit at an angle of more than 50 degrees, there were just 16 hits (for a BABIP of .012) along with six errors. As for runners advancing, there were three attempts to advance on a sac fly, and two were caught—truly an unproductive out.

Now that we have a precise definition of kind of hit makes an automatic air out, let’s go back to PITCHf/x and see what kinds of pitches tend to make that kind of hit happen. Breaking things down by pitch type, here’s the chance of contact with each pitch resulting in a pop out:

Pitch Type     Popout %
Fastball       10.5
Change         9.6
Slider         9.4
Curve          5.5
Sinker         4.8

This meshes with previous studies (such as this one by John Walsh) that have shown fastballs get the most fly balls, but it is also interesting to see the changeup not far behind in terms of harmless flies. That would seem to make sense, as in general they are better at inducing poor contact than missing bats.

There are problems with lumping all pitches of the same type together, however, as different pitches are thrown more frequently to different locations. Obviously it’s going to be easier to get under a pitch thrown at the top of the strike zone, and both by design and the nature of the pitch, fewer curves end up there. How much of the fastball’s ability to get a pop up is due to its “rise” (to be technically correct, its lack of sink), and how much is just based on how high it was thrown in the zone? In PITCHf/x terms, a vertical movement value (pfz) of nine is about average. The following chart shows the chance of a pop out coming on a fastball from a diving sinker to an exploding ball.

Pfz        Popout %
<0          4.9
1           3.4
2           3.2
3           4.9
4           6.0
5           4.6
6           7.2
7           9.6
8           9.6
9           11.6
10          13.5
11          14.3
12          16.8
13          14.4
14<         14.0 

This is almost certainly the reason that one of the recent exceptions to DIPS theory is that dominant closers seem to be able to maintain consistently lower BABIP's, unlike the majority of average pitchers. They tend to be fireballers who will get more backspin and (usually) more rise on their pitches, leading to more harmless pop outs.

However, the vertical location of the fastball does not have as much of an effect; while low pitches typically lead to fewer balls hit in the air, there is an insignificant difference in the number of pop flies between a pitch in the middle third of the strike zone and those above that (even out of the zone). Getting a big league hitter to pop up a pitch is more about getting your fastball to rise than throwing it high.

Fastball Location   Popout %
Up                  13.2
Middle              12.9
Down                9.9

In contrast, you can clearly see the effect of hitters getting jammed on inside fastballs and popping them up. After dividing the plate into horizontal thirds this time, you can see that a player is more than twice as likely to pop up an inside fastball than he is one that he has to reach across the plate for:

Fastball Location    Popout %
Inside               13.8
Middle               10.8
Away                  6.2

April results

I've already walked a fine line with sample sizes throughout this article, so I'm not even going to attempt to draw any conclusions about individual players and their pop out rates through the first month of the season. But for the curious, here are the five hitters and pitchers (at least 30 balls in play) who were involved in the most harmless flies through the month of April. (For hitters, you can find a similar stat in THT's statistics section: infield flies per fly ball. Chris Young (struggling through a terrible season with a groin injury) is currently way out front (suggesting he has kept up his pop out pace), while David Ortiz has plunged to 9.9 percent as he has turned his season around.

Top 5 Hitters

Name            Popup %
Carlos Pena     27.3
Chris Young	25.0
Rick Ankiel		25.0
David Ortiz	24.4
Eric Byrnes	23.5

Top 5 Pitchers

Name               Popup %
Johan Santana       28.6
Kevin Slowey        26.8
Koji Uehara         23.7
Carlos Villanueva   23.3
Tim Wakefield       22.2

So what have we learned from all this?

{exp:list_maker}Unlike most balls in play, balls hit at a greater angle than 50 degrees are essentially guaranteed outs.
Rising, or inside fastballs are much more likely to induce those types of pop ups.
Pitching up in the zone is not as important as having a rising fastball and working inside.
We really, really, need more HITf/x data and then things are going to get very fun. {/exp:list_maker}

References & Resources
The information here was provided by Sportvision and’s Gameday for research purposes only. Sportvision and retain all copyright rights.

A Hardball Times Update
Goodbye for now.

Newest Most Voted
Inline Feedbacks
View all comments
Harry Pavlidis
12 years ago


12 years ago

As I was reading, I remembered a game where Tim Wakefield recorded 3 straight outs on pop ups to the catcher and kept wondering – what about the knuckleball?

When I got to the end and saw the top 5 pitchers list, I was not surprised; and without evidence to the contrary, I very much doubt it’s Wake’s blazing fastball that landed him on the list.

So … what about the knuckleball?

Dave Studeman
12 years ago

Harmless promotion: We talk about infield flies a lot in THT’s Batted Ball Report, which is available to those who pre-purchase the THT Annual.

We use BIS’s batted ball data instead of Hitf/x for now, but it’s worth noting that one guy on Jon’s list, Eric Byrnes, has consistently been the infield fly king the last four years.  More than any other major leaguer, Byrnes hits pop flies.

Nick Steiner
12 years ago

Excellent article.  I love the two graphics; very informative and easy to read. 

I have a question.  Since we know that stuff like movement/velocity/location directly affects quality of batted balls, couldn’t we use Pitch f/x to create a more inclusive DIPS? 

There have been a couple of posts at Beyond the Boxscore that use pitch location to determine HR/FB.  By including velocity and movement into the equation, it seems like it’s possible to model expected runs like that.

Dave Studeman
12 years ago

And I agree about the data tables and graphics.  Way to satisfy both audiences, Jon!

12 years ago

Great article! This will only get more interesting as more HITf/x data is released.

Jonathan Hale
12 years ago

Curious: Almost all of Wakefield’s balls in play in April (52/54) were against the knuckleball, and 11/52 of them (21.2%) were +50 degree pop-ups. I believe the 2 main exceptions that have been allowed into DIPS theory these days are closers and knuckleballers, so right – no surprise to see him up there mainly on the strength of that pitch (and even less of a surprise if you’ve seen him when he’s got his good stuff).

12 years ago

Thanks for the additional Wakefield numbers, Jonathan. And the opportunity to say how much I’m enjoying the analysis of the new data coming out.

Jonathan Hale
12 years ago

Nick: Thanks for the feedback. I was actually going to apologize for the gratuitous use of graphs there but I thought they might help.

Peter Jensen has used hit f/x to create something DIPS-like based on how the ball comes off the bat here:

I think I get what you’re suggesting for pitch f/x though, and that’s a very interesting idea. I worry about trying to apply generalizations about pitches (e.g. sliders low get hit for a low BABIP) back to individual pitchers, though, because you lose all context. It might not be such a big deal but it strikes me that two pitchers could have identical fastballs in pitch f/x’s eyes, but yet in reality show vastly different well-hit rates because they are thrown at different rates, at different times, and are combined with different pitches. To take an extreme example, on it’s own merit along, it should predict Wakefield’s fastball to be among the worst pitches in the game when it’s not so bad alongside his knuckler.

Even comparing a pitcher to their past performances (like we do with hitters and BABIP) could be problematic. Say a guy stops throwing his slider so his fastball is getting demolished because hitters are sitting on it – but the new metric calculates his expected runs based on his typical fastball BABIP and concludes that he is getting very unlucky.

Nick Steiner
12 years ago

Good points.  When I think about it, sequencing probably has a huge effect on outcomes.  I guess we’re stuck with tRA for the time being.

Peter Jensen
12 years ago

Very nice article Jonathan.