THE BOOK--Playing The Percentages In Baseball

We are in for an explosion of data. With ESPN’s system, I suspect we’re about to find out that quite a few parks have numbers on the OF walls labeled wrong. I’d want to see how their numbers compare to HitTracker’s first, before we give them any sort of credibility.

The Trackman radar device has great possibilities.  It was originally developed for tracking golf balls
(see http://trackman.dk/).  A few years ago, the company (ISG) worked with Rawlings to adapt the product for baseball.  As a result, Rawlings has one of these and they call it the Sci-Fly (see http://rawlingsscifly.blogspot.com/2008/03/introducing-sci-fly-system.html).  Through my friends at Rawlings I have actually used the Trackman on several occasions and have developed a pretty good understanding of how it works, how well it works, and possible limitations.  The one thing that I don’t know very well is how accurate it can measure the trajectory of a pitched baseball.  The company claims about half the ball diameter (about 1.5 inches), which is comparable to pitchf/x (about 1 inch).  But, as far as I know that claim has never been tested.

Advantages of Trackman over pitchf/x and hitf/x:
Trackman is completely portable and takes only 15 minutes or so to set up.  Trackman measures the full trajectory of the pitched and batted ball.  Trackman measures the spin on the ball.  Note that Trackman does not measure the spin axis, which is inferred from the trajectory itself (just like in pitchf/x).

Alan, I would appreciate if you would look over my explanation of the Trackman hardware (at the first link in Tango’s post) and correct any inaccuracies I may have communicated there. 

Much of what I wrote was taken either from the conversation that you, Marv, Ken, and I had about the system at lunch last year or from your presentation which I link in my post.

I loved the article in the NY Times.  I thought it was very well written and great positive exposure for what’s going on in our sport.

I do find it a bit ironic that the graphic they lead with (Cunningham’s range) is the one for which I had to do the most seat-of-my-pants guesstimation based on the data I had available, whereas, for example, I was extremely impressed/excited to be able to calculate the real-time speed of the baserunner throughout his path.

Mike:  Your explanation is mostly right.  The phase information from the two pair of detectors determines the angle of the ball, not the angular velocity.  If the position of the ball (relative to the radar location) is known at some point in time, then the radial velocity tells you the radius of the sphere where the ball must lie a short time later.  The two angles tell you where exactly on the sphere the ball lies.  The radius the the two angles then give the exact position. Through this “bootstrap” method, the full trajectory is mapped out.

Also, I don’t think the spin determined by the system depends on seam orientation.

Thanks, Alan.

Maybe my terminology wasn’t precise enough.  What I meant by angular velocity was not the angular velocity of the baseball relative to its own center but rather the angular velocity of the baseball relative to the radar emitter/detector, which I think is consistent with how you describe it.  The velocity is determined in a polar coordinate system with the radar at the origin.  The radial component of the velocity is measured from the Doppler shift.  The two angular components of the velocity in this coordinate system are measured from the phase shifts between the pairs of detectors.  Correct?

Re spin, I thought you had mentioned that seam orientation could produce a doubling of the measured frequency, if the radar saw four seams on each rotation instead of just two seams on each rotation.  Two-seam vs. four-seam orientation may be a bit confusing way to explain this since those terms have meanings for pitch grip orientations, and that’s not exactly what I’m talking about here.  I’m talking about the orientation of the batted ball spin axis relative to the radar and how many seams this exposes to the radar on each rotation of the baseball.

I want to be clear, is it true that Trackman is a different system from what SportsVision will be unveiling at the conference tomorrow (as detailed in the NY Times article)?

Trackman is based on Denmark:
Site

Sportvision is in SF

My question is then what does this technology mean to the future of Hit Tracker Online?

Data is nothing without someone to analyze it.  I don’t know what it means for HitTrackerOnline, but I think it means that Greg will have his hands full even more now.

I added an UPDATE link at the end of the original post to point to Sky’s article as well.

Mike:  sorry to nitpik, but it really is the angle that is measured and not the angular velocity.  Look at slide 36 in this link:
http://webusers.npl.illinois.edu/~a-nathan/pob/ppt/BowdoinApril08.ppt, which explains how the phase difference in the two detectors is proportional to the sine of the angle.  The argument is essential the same one used in basic physics courses to explain the two-slit intereference pattern in optics.

You might be right about the spin (I forgot that I had discussed the seam orientation with you).  It certainly is true that in my own check of Trackman, comparing its determination of the spin with what I can directly measure with high-speed video, I found that the Trackman was either in perfect agreement with the video or exactly a factor of two higher.  I think I was speculating that it might be due to whether the radar sees 1 or two seams per revolution.  I will have to ask the ISG people about that to be sure.

Alan, thanks for nitpicking, and I think I understand now what you are saying.

For all the work I have done, MGL has done, and the people that followed in our footprints, we are going the way of the dinosaur.

Also, this was a terrific idea here

http://groups.google.com/group/alt.sports.baseball.sd-padres/browse_thread/thread/536fdbfce9f81807/8bdadae78d7ca7f3?hl=en&lnk=gst&q=Johnson+defense+positioning#8bdadae78d7ca7f3
We’ve had a similar discussion.  Good to see it moving forward.

I just got off the phone with Eric, who will be attending the conference tomorrow.

We have made comments about how the teams have some stuff which is so much better, but Eric says he’s talked to some people and they seem amazed at the things we ‘hobbyists’ (my term) come up with.

We will have a flood of new data, hopefully filling in all the necessary holes. In one way, we will try to fit the new data into our existing paradigms (how can UZR be improved with location data that’s more objective?). On the other hand, hopefully some of the new data will inspire new thinking.

The teams have data, but in many ways it’s still those of us who love and understand baseball who do the critical thining. So Chris, I do not think we have gone the way of the dinosaur. Just time to write new algorithms. I’m excited.

Chris/16 was marked for moderation and is now open.

Sportsvision and three other companies are expanding this over to football.

http://sports.espn.go.com/nfl/news/story?id=4891097

As well as being able to tell if the ball crossed the goal-line or if the receiver caught the ball inbounds,

“It will be able to tell you who was the fastest player without the ball,” he said, “once they catch the ball, how does his speed change? Is it higher? Which wide receivers are really accelerating off the line of scrimmage?”