THE BOOK--Playing The Percentages In Baseball

I’ll tell you what.  His talent level may keep him from being a great pitcher, but when he’s through, someone somewhere should give him a shot as a pitching coach. 

If he could get a staff to buy into his future “system” the results may or may not be phenomenal, but they would be interesting to say the least.

1) I hope he somehow ends up a Detroit Tiger at some point in his career.
2) I’m not sure I could root for this guy anymore ... then he says stuff like this. So, so awesome. What I’d give to be able to talk with him for a few hours.

Why wouldn’t a team set up a PitchFX system in their bullpen or workout facility for this kind of practice? Monitoring bullpen sessions seems like something a team would want to do.

LH,
Some teams do pitch track in the bullpen or during side sessions (not necessarily at the MLB level)

You’d at least think he could get the system turned on pre-game, for a kind of on-field bullpen session. Maybe if he played for the Red Sox and not the Royals. Poor guy.

Why wouldn’t a team set up a PitchFX system in their bullpen or workout facility for this kind of practice? Monitoring bullpen sessions seems like something a team would want to do.

In most cases the teams don’t own the Pitch f/x system, Sportvision does and they provide the trained operators to run it.  It is not portable as it requires fixed camera mountings that have to be calibrated for the system to work correctly.

The Trackman doppler radar system has a portable model and some have been sold to teams to do just what you are suggesting.

I do think that with the $ a MLB team has at its disposal he could work out a way to train with a pitch f/x system instead of trying an occasion pitch in a meaningless game situation. 

What do you think?  Is this likely to be fruitful?  If you were a pitching coach would you have guys train this way (can pitch f/x better gauge the movement on a given pitch than the catcher?)?

If the Royals as a team don’t use the PITCHf/x data that’s already freely at their disposal, why do you think they would pay extra money to install a system in the bullpen?

Brian Bannister probably doesn’t have enough spare cash to purchase a system of his own for training.  Or maybe he does, but I don’t know how the team would feel about him doing that.

What do you think?  Is this likely to be fruitful?  If you were a pitching coach would you have guys train this way (can pitch f/x better gauge the movement on a given pitch than the catcher?)?

As Peter mentions here and has been mentioned elsewhere, there are other more portable systems that are typically used for training.  I would absolutely be trying out such a system if I were working with a pitching coach to help him with his pitchers. 

I know that is at least being done at the NCAA level if not at the professional level.  There seems be more freedom to experiment in the college game.
http://www.youtube.com/watch?v=Efsg1YTvprw

Reading Bannister’s comment gave me an interesting thought.  What if the teams would display the LI on a pitch-by-pitch basis on the scoreboard.  It would probably be especially interesting to see how the color commentary-types would (ab)use it.

While something as well tested and high quality as the pitchf/x system might be a bit cost prohibitive to install in a bullpen, there are still things that could be done even with modest resources.  At one point I came across what looked like a doctoral thesis describing a system very similar to pitchf/x, but using little more than easily obtainable digital video cameras and a laptop to track cricket pitches.  It probably wouldn’t be difficult for a player to pony up some cash for the equipment and a local computer science or math or physics grad student to set it up for him in a relatively ‘repeatable to the slightly above average guy’ way.  It might require putting up a black on white gridded background , but that would be pretty cheap too.

Alan Nathan also has some descriptions of ball tracking apparatuses he’s used in experiments on his page.  None of them look terribly expensive, but he can correct me if I am wrong.  The expense may be in the calibration.

There are other ideas you could do too...especially in the confines of a bullpen.  It might be possible to craft a baseball with an RFID tag embedded just under the skin, and a series of readers on either side of the pen....if the readers are able to make readings at a high enough frequency....I don’t know if they can or not.

If it sounds like I’ve put a bit of thought into this....from time to time, I have...I’m still pitching a bit myself in old-man leagues, and there is nothing I like more than experimenting with my pitches.  I also now have a son, who I hope one day will find baseball, and pitching, as enjoyable as I do.  If that’s the case, you can bet there will be a bullpen in my backyard, and if I can swing it, some kind of pitch tracking to go with it.

Anyway, my point to all of this is, if the Royals, or even Bannister himself, wanted to employ some kind of pitch tracking system in the bullpen for essentially not much money, they probably could.

That’s a couple of interesting ideas, there, Ike.  It doesn’t seem like it would be all that difficult to build a frame to hold a pair of cameras, and putting a background on the ground and the opposite side would sure make the programming easier (especially with it being portable, and hence, not having a consistent background).

But the idea of an RFID is also very intriguing, especially as you could embed multiple sensors about as easily as you could embed one.  That would make tracking the spin straightforward.  And I suspect that you could also capture data at a higher rate with an RFID more easily.

Well, at a higher rate than a film camera certainly.  Something built around digital cameras similar to Casio’s Exilim line would also be interesting.

I honestly don’t know enough about RFID to know if it would work or not.  You may have a difficult time with it getting the spatial resolution you would need with it to be useful...but it still might be doable…

I don’t know either, unfortunately.  Although if it’s purely a resolution issue, you could likely solve it with multiple sensors (right next to each other, for instance).

Hmm… In the middle of writing this, I got an email update from my work about new technical books being available for free reading.  One of them is RFID Technology and Applications.  I think I’ll need to take a look.

I got to thinking about this a little more and I think that right now, even with multiple sensors, you’d have to be pushing the limits of technology. 

Consider measuring the distance to a tag with it.  If you are using passive RFID, an external antenna sends a signal, which gets recieved by the tag and then the tag sends out a different signal.  The total time is then c*|B-A| + D + c*|B-S| where B is the vector position of the baseball, A is the position of the antenna, S is the position of the sensor and D is the delay of the RFID chip, and of course c is the speed of light.  If you have multiple sensors and one broadcasters, then the only difference in time comes from the difference in positions of the sensors.  To resolve then a change on the order of 1cm, you would need at a minimum a sampling rate on your reciever of about 30 billion samples per second...or 30 GSPS...or a sampling frequency of 30 GHz...those are all the same thing. (light travels 30 billion cm in a second.) A quick google search showed me that Analog to Digital Converters, which would convert the received radio signal to a digital signal to be processed by the computer, currently max out in the range of 3GSPS...or at least they did at the two vendors whose sites I looked at.  A factor of 10 too slow. 

So with top line tech (if that is indeed the top of the line tech), you could probably resolve the baseball to within 10 cm, or about 4 inches on each sensor...which isn’t terrible.  But I now think you’d get better results with video.

Hmm… You would need to double that frequency, right?  (of course, factor of ten, factor of twenty? who cares? )

Either way, it does sound like video is a lot better, at least for now.

Can’t help but think you got the math wrong. An ADC isn’t all that different from a camera that can take pictures at the same rate. So, a 30 GHz camera would see the ball move 15 nm every frame.

My understanding is that to resolve a 1 cm change, you need to know how long it takes for the object to move 1 cm and sample at twice that rate. At 100 mph, that would mean a 9 kHz sampling frequency.

It’s been a while so maybe I’m completely wrong, but 30 GHz just sounds way too high.

Ike, one thing that Sportvision relies on with their PITCHf/x system is that each point on the pitch trajectory is not measured independently.  You can assume that the baseball is obeying the laws of physics between points and this allows you to measure the trajectory as a whole much more accurately than you can measure for any given point.

Ike (#10):  The experiment Ike refers to was one I did a few years ago to measure the effect of spin on the flight of a baseball.  I fired balls from a pitching machine and tracked them with a set of 10 IR cameras that tracked a reflective marker on the ball over about 15 ft of flight.  The cameras operated at 700 f/s, if I remember correctly.  And the system was VERY expensive (I only borrowed it), about a $40k system.

I just visited Scientific Baseball (http://scientificbaseball.com/) near Oklahoma City.  It is an indoor batting cage with leased PITCHf/x cameras set up.  The idea is to train pitchers and umpires by providing immediate feedback on pitch location, release speed, movement, release point, etc.  It is not exactly a portable system, since moving the cameras would require a complete re-calibration of the system.  That is not impossible but it is time-consuming. 

PitchSight (referred to by Mike Fast, #8) claims their system is more portable than PITCHf/x.  However, their technology is not terribly different than SV’s and it is not likely to be any more or less portable.  The link Mike provided shows the system installed at Boston College.

Peter (#6) refers to the TrackMan Doppler radar system, for which there is a portable version that is ideal for the bullpen.  Setup is quick and easy.

Sean #16:  I may have gotten my math somewhat wrong...but not terribly, but the RFID triangulation method is not like a camera.  It’s not like a camera that measures a position of an object in a 2D plane.  What you are actually measuring is the time it takes a signal to go from point A to B to C.  So the “object” who’s movement you are directly measuring is not the baseballs, but the radio signals. 

Now, if you wanted to make a measurement of the ball’s position every time it moves a centimeter at 100 mph, then you’d need to send 9000 signals every second, or at 9KHz...thats probably overkill.  But if you are measuring a signals arrival time with a 9KHz sampling rate, then the distance light can travel from one sample to the next is huge.  We’re talking zip-code huge.

You might be able to sample at a lower rate though and use dispersion relations to better estimate the distance from source to receiver based on the amplitude of the incoming signal to better estimate the distance...but that could be somewhat problematic.