Model pitcher database

To help troubleshoot pitching problem areas, I started a database project to capture frame counts for key intervals on elite pitchers. Using the averages from this database, you can add metrics for your own DD or student to try and identify differences between her and the elite. The tab called "CHECKPOINTS" has several columns that (at least in my mind) pertain to important check points in pitch mechanics.

As needed, I'll continue to update the spreadsheet and upload it to this post. You are welcome to download it and add your own data. I'm also very open to suggestions and changes. If the experts have suggestions for changes to the measurables, I'm all ears. If someone has ideas to improve the format, add charts/graphs, find better ways to summarize, have at it, and I'll add to or update the file.


If you would like for me to add model pitchers to the file, just respond with this information:

Name
Description
Link to thread
Post #

Drive foot lift (right before push / centering)
Arm swing peak
Stride foot actively lifted
Drive foot push
Arm @6 o'clock
Drive foot separate from rubber
Arm @3 o'clock
Arm @12 o'clock
Stride foot plant
End of hip drift (frame before it locks or changes to rotation)
Arm @9 o'clock
Arm @6 o'clock
Ball hit or at home plate


Below is the latest database as of 1/20/2018:




Below is a sample evaluation showing how data can be used to evaluate a pitcher:

Can you explain how you are doing frame counts?

Sure.

1. Find a standard speed GIF that I'd like to analyze and save it onto my computer.
2. Open the GIF in virtual dub (probably any GIF viewer will work)
3. Click frame by frame through the GIF and record key check points that are listed in the "Raw" tab

Check points
*Generally speaking I look for the first frame of movement for each of the following items and then record the preceding frame. It's by no means perfect, but I'd bet I'm within ±1 frame of most events.

Drive foot lift (right before push / centering)
- I look for the first movement of the drive foot to center or step before it plants
- This one has a lot of variance, but I kept it anyway. Not sure that it will serve a purpose for analysis...

Arm swing peak
- Watch arm go back. Record the last frame before it starts to move forward

Stride foot actively lifted
- Watch for the first frame that the stride foot is actively lifted
- Try not to avoid picking frames where the stride heel is pulled off the ground because of tilt
- We are waiting for the pitcher to actively lift the leg and record the frame right before it happens

Drive foot push
- Looking for the moment the drive foot is forcefully planted into the rubber

Arm @6 o'clock
- First 6 o'clock down swing
- For non-backswing pitchers, the last frame before the arm moves beyond 6 o'clock

Drive foot separate from rubber
- Looking for full disconnect from the rubber
- Can be hard to see but you can look for the drive heel to kick up. It's usually a noticeable event. Record frame just before.

Arm @3 o'clock

Arm @12 o'clock

Stride foot plant
- Looking for the stride foot to touch and plant into the ground. Record previous frame.

End of hip drift (frame before it locks or changes to rotation)
- This one is tricky. Looking for last frame of hips sliding forward before the plant leg locks it down.
- Sometime the plant leg stops the motion dead
- Other times, the front hip will stop and rotation will occur. For now, I count that as a stop and record the frame before

Arm @9 o'clock
Arm @6 o'clock

Ball hit or at home plate
- This is a little bit of a SWAG but wanted to keep it just because I could.
- If you come up with an average for point of release off of the rubber like (43' - 6' of stride =37' ball travel), then you can calculate pitch speed

Improvement ideas:
- Add a frame rate column so slow motion videos can be analyzed (if you know or can derive the frame rate)
- Consider adding a macro to convert from frames to time (1 frame = 1/30 of a second)
- Maybe add a pitch speed estimate column
- Add userforms to enter data in a more controlled way
- Long shot, but possibly add a GIF viewer in excel to count frames in the application

If someone want to see some test numbers, post a GIF and I'll add the data.

Really interesting I think you have some interesting data there. It is very interesting how the elite pitchers numbers are so close. Other things are places where you could basically have still pictures as reference points like foot separate at 3 oclock, drive foot plant at 9 oclock.

Statistically speaking you probably should not list the same pitcher twice as a particular style in the motion might skew results, to be most accurate if you have more than one result from a pitcher average them together then maybe put an asterisk with a note that this line is the average of several different measurements of the same pitcher.

jryan15 said:

If someone want to see some test numbers, post a GIF and I'll add the data.

Click to expand...

How about Sydney Littlejohn? From the side and behind...same pitch.

http://www.discussfastpitch.com/softball-pitching/5208-model-pitchers-23.html#post306484

CP

djcarter1966 said:

Other things are places where you could basically have still pictures as reference points like foot separate at 3 oclock, drive foot plant at 9 oclock.

Click to expand...

Yes, got foot separate at 3 oclock on the spreadsheet. I noticed the same thing with plant and 9 oclock. Might add that one back in...

djcarter1966 said:

Statistically speaking you probably should not list the same pitcher twice as a particular style in the motion might skew results, to be most accurate if you have more than one result from a pitcher average them together then maybe put an asterisk with a note that this line is the average of several different measurements of the same pitcher.

Click to expand...

That's a good idea. Right now the data set is pretty small. Another factor is what pitch they are throwing... Most of the model pitcher clips aren't marked, and I'm not great at reading pitch types. In the raw file, the location of each clip is noted. If anyone wants to list what pitch each one is throwing, I'll add a column for that.

If the data set could grow enough, it'd be cool to have a drop down to choose between pitch types to compare multiple pitchers or have a pitcher drop down to compare her mechanics by pitch... Auto-filter would work great for this, but we'd loose the average column. Too bad there's not an auto summarize option .

chrispots said:

How about Sydney Littlejohn? From the side and behind...same pitch.

http://www.discussfastpitch.com/softball-pitching/5208-model-pitchers-23.html#post306484

CP

Click to expand...

Sorry, no luck on her. Both clips have some kind of frame rate adjustment causing them to have extra frames with no motion. I ran into a few other clips like that in the model pitchers thread that I couldn't use.

Also the rear view starts without a good way to distinguish if frame 1 is the start of the drive foot push and that's a key measurable for drive.

Hmmmm.... I'm not a gif guy, but I know from my limited experience, I can manipulate the frames in a gif... so how would you know that you are comparing apples to apples? What if the raw video was recorded at a different frame rate?

FrozenRope said:

Hmmmm.... I'm not a gif guy, but I know from my limited experience, I can manipulate the frames in a gif... so how would you know that you are comparing apples to apples? What if the raw video was recorded at a different frame rate?

Click to expand...

I don't have much experience hacking GIF's, but I understand it can happen. This being an analysis tool, I don't expect there to be much incentive for someone to forge a clip. I'd expect it more from hitting than pitching.

The clips I've used have been pretty much all from the model pitchers folder. A few had did have modified frames rates due to slow motion conversions or whatever. They were all pretty obvious just like the LittleJohn clips.

I'll have to keep an eye out for added or dropped frames I guess. If you have any examples of a bad GIF, let me know, and I'll test it out.