THE BOOK--Playing The Percentages In Baseball

Sweet!  Thank you, Sean (and Tom).

Awesome. Now just add in atmospherics and we’re good to go, heh. Thank you for this tool.

Love this. Would’ve been an interesting thing to throw into a Johnny Damon post for us Tigers bloggers a week or so ago, had I known about it.

Really, really, really excited for the Pitcher Cards.

http://katron.org/projects/baseball/

Thanks.

SParuchuri/2:
I wish. I think it’s for the best that I don’t have access to hitf/x data; otherwise it would’ve taken much, much longer to finish.

Think we could get it so the “Show hits from:” could have a setting for “ALL”, so that we could look at all of Granderson’s 2009 hits over Yankee Stadium?

This is very cool.

Of Jason Bay’s 15 HR at Fenway only 2 leave Citifield.  Uh oh.

Of course the BIG problem with taking batted balls, adjusting them for atmospherics, and then superimposing them onto another ball park is this:

Batters and pitchers, I believe, pitch to the strengths and the weaknesses of the park.  And the batting eye, visibility and the size of the foul territory of a park can significantly affect the character of the batted balls.  The other problem is the atmospheric one.  No one seems to know the real wind patterns inside of any park.  That can be a pretty big problem.  Basically, it is not as easy and clean as you might think.  Trust me, I’ve tried it…

This is a great tool, but I have a feeling that people are going to misinterpret it a lot.  Gameday measures where the ball was eventually fielded, not where it landed, and that makes a huge difference with in play hits that could be superimposed.  If Sean could put a note saying this on the top of the web tool, that would be awesome.

J. Cross #6:

Take this data with a big grain of salt.  The park diagrams are approximate only, and have lots of scaling issues. 

If you want a good example of this, set up to display on Tropicana Field (LF pole 315 feet from home plate), then when you switch to Fenway Park (LF pole 310 feet from home plate), watch the LF corner jump outward about a third of the distance between the bases (I’d say 30 feet, but…)

Another example: I superimposed all Yankee Stadium homers on Citi Field, and only 9 of the 127 home runs hit to RF in Yankee Stadium show as fully inside the fence line on Citi Field (which is of course ridiculous).

FWIW, I ran Bay’s 15 Fenway homers at Citi Field, using average weather for Citi, and got 8 home runs (with 1 of the remaining 7 very close to making it up and over).  He’ll suffer in Citi Field, but on the bright side, in-division he gets to hit in Philadelphia and Miami, which is a lot like Fenway…

Nick/8:

Good point. I’ll be sure to add it in bold letters.

Greg/9:

Well, the parks are internally consistent, but that doesn’t count for much. Some of the park layouts (Progressive comes to mind) don’t completely match Gameday’s image.
[Now that I look at it more closely, there does seem to be some improper scaling. I haven’t checked to make sure it scales the hits at all. I’ll need to do that tomorrow.]

The home run distance might be the least reliable measurement. Certainly nothing compared to what you have. I find most of them are well beyond the wall despite only rarely being that far.

Also, the x,y scale factors sometimes change in the park. For simplicity sake, I just used one scale factor and this worked great (<1% error from the diagrams) on maybe two or three parks. Most would be accurate to only one field and would vary noticeably to the other two. Plus, unless you’re mapping hits from one park onto itself, there are two translations done on the coordinates.

This is a whole lot more useful than the Ballpark Overlay Tool I developed for Hit Tracker HR data.  Of course, like anything, it’s not without limitations, but this is really useful.  I like it a lot.

The two biggest improvements I thought of when I published my tool were (other than more accurately measuring park dimensions) a) integrating all BIP, rather than just HR, into the overlay and b) accomplishing it via web interface.  They’ve done both of those things.

Kudos to them and thanks for posting this.

A couple weeks ago I was looking at ballparks on Google Maps, and it occurred to me that those satellite images might be a good, consistent data source for mapping field dimensions.  In case anyone wants to give it a try…

It would be nice to expand the filter options to include vs. RHP and LHP and batting R or L.

One addition to the accuracy comments is that some dots are superimposed. I checked Varitek’s HRs at Fenway and only saw 8 of the 10 he hit there. I had to go through it by pitcher to find them all (and in one case, Hit tracker showed one ball in the pair to have been hit about 70’ farther).

Newcomer #12:

Satellite images are one of my primary sources for park diagrams, along with blueprints, and photos of the park.  One thing to watch out for, though, is that you have to correct those images for the viewing angle, which is usually not directly overhead.  Here’s a good example: what is otherwise a very nice detailed image of Dodger Stadium, but if you look closely, the LF and RF lines are not perpendicular (this is due to the viewing angle - you can also see that the angle is off by looking at the foul poles)

http://maps.google.com/maps?f=q&source=s_q&hl=en&geocode=&q=los+angeles,+ca&sll=37.0625,-95.677068&sspn=42.310334,70.224609&ie=UTF8&hq=&hnear=Los+Angeles,+California&ll=34.07363,-118.239778&spn=0.000678,0.00158&t=h&z=20

Some of the park diagrams you see on my website are not the latest version I have, because the routine I use to superimpose flight paths on the diagrams requires a rather laborious formatting of the park diagrams that I only do once in a while, but I’m planning to address that once the season starts.  All the data generated with Hit Tracker uses the most up to date, accurate diagrams, though.

Cliff #13:  For balls in play, we know MLB is plotting the fielding location, but I’m not sure if what they post for home run spots is meant to be the landing point, or the impact point - if a ball hits up high like over the Green Monster, these can be quite far apart.  That may be the explanation for the Varitek data, but I’d have to check carefully to be sure…

I’m sure there is data quality issue.  Classic GIGO.

As for Google Maps, many park guys have used those.  Greg wrote an article about it to for THT last year I think.

Greg #14: I should have mentioned (and intended to) that the pair of Varitek HRs I alluded to were down the RF-line in Fenway. I can see where high fences would cause problems for home runs.

There is another problem with plotting the fielding location on balls in play and that also has to do with fences. Balls rebounding off fences may be fielded in a very different spot from which they hit. For example, a double to LF/LCF in Fenway might end up recorded as a double to medium CF.

It would be nice to expand the filter options to include vs. RHP and LHP and batting R or L.

Done.

I also added a diagram for Shea. There’s a slight bug right now because you can select Shea in both selects, but it should only appear in one. That’s soon to be fixed once I find the SQL to do it.

I tried to add one for Yankee Stadium, but Gameday’s diagram is identical to the new version.

Thanks for the follow-up, Sean.

I did Fenway on Fenway, ground-outs, LHH vs. all pitchers and there is definite evidence of shifts against LHH. It looks like most of them were at David Ortiz’s expense.

If it is feasible, the LHH/RHH breakdown might be interesting if it could be done for switch-hitters. I don’t want to see the interface get too complicated but if would make it simple to see how they take advantage (if they do) of friendly fences.