Comments on: Found: Verlander’s Velocity

By: Mike Fast

Mike Fast — Sat, 28 Jun 2008 00:07:07 +0000

Video tracking of the ball is not that hard. It’s probably cheaper (and more accurate?) than a radio transmitter system, and you don’t have to mess with the baseball.

It really is an issue of cost and return, though. Ultimately I believe the video tracking system for the whole field (i.e., HITf/x) is going to happen within the next few years, but the investment they’ve made into PITCHf/x has to show that it can pay for itself first.

By: Scappy

Scappy — Fri, 27 Jun 2008 20:36:46 +0000

I am still surprised that they have not tried putting some type of accelerometer and radio transmitter inside of a baseball yet. The radio transmitter could be used to track position and time of the baseball throughout the game and the accelerometer would help give batted ball information because you would be able to know when the ball hits the ground.

This is major league baseball, they make millions upon millions of dollars, who cares if each ball costs 20 bucks.

By: Mike Fast

Mike Fast — Fri, 27 Jun 2008 13:45:02 +0000

One of the problems with using a radar gun to determine pitch speed is that there are a lot of things moving at different speeds–the pitcher’s arm, the pitcher’s hand, a decelerating baseball–all of which generate radar returns, and you have to reliably pick out which one of those speeds you want to measure. The pitcher’s hand after release is the fastest-moving object, and that may be the source of the 108-mph readings.

There may be some radar experts who can explain it better than I can, but that’s what I understand as the difficulty involved.

By: Scotsw

Scotsw — Fri, 27 Jun 2008 13:14:43 +0000

108???

And these are devices the police rely on? That doesn’t do much for my “confidence level.”

By: Dave Cameron

Dave Cameron — Thu, 26 Jun 2008 17:04:14 +0000

Like Mike said, radar velocities on any given pitch can be hilariously wrong (go to any minor league game, sit behind the scouts, and wait a few pitches – you’ll inevitably see a “108″ pop up), but for the data displayed on fangraphs, I’m sure those obvious errors aren’t counted in the averages. I’d imagine that the error bars around the data displayed here is probably less than 0.5 MPH – over hundreds of pitches, you’ll get a pretty accurate reading of what a guy throws.

By: Mike Fast

Mike Fast — Thu, 26 Jun 2008 16:30:48 +0000

Kent,

PITCHf/x gets pitch speeds within about 1 mph of random measurement error for each individual pitch. There may also be systematic errors of a similar magnitude.

Radar gun measurements are what BIS uses, unless I am mistaken. These are subject to much greater error and are much more operator and condition dependent than PITCHf/x. So for a specific pitch here or there, it wouldn’t surprise me if BIS data is 5 mph off. But once you aggregate over hundreds of pitches the error should go down well below 1 mph, I would guess.

By: Kent Bonham

Kent Bonham — Thu, 26 Jun 2008 13:50:09 +0000

Not specific to this post, Dave, but just something I’ve always wondered: What confidence levels are associated with the velocity readings? Actually, “confidence level” probably has a very specific statistical meaning that I’m probably bastardizing, but…

If BIS logs a fastball at 91.6 MPH, what are the chances the pitch was 91.2 MPH or 92.0 MPH (or whatever)?