Monday, April 07, 2014

Gridiron Solitaire #100: Punts!

Well, I never thought I'd get to 100 of these. Hopefully that isn't dreadful for anyone in terms of dangerous levels of boredom.

Also, this may be the only time the word "Punts" is followed by an exclamation mark.

I've been finding that every time I get something closer to "real", it seems to be better, so I decided to do that with punts. Previously, while ratings did have an influence, punts were relatively simple and only vaguely reflective of the real world. The strategy, yes, but the actual numbers, not so much.

This seemed like a simple thing. Guy kicks a ball, guy returns a ball. What's complicated about that? Well, as it turns out, quite a lot, and I'm going to tell you about it today. This is a good example of how deep a small detail can make you dig, if you want to get it absolutely correct (or almost).

First, I needed real punt data, so I downloaded the individual play database for last season, allowing me to load the details from every punt play in the 2013 NFL season into an Excel spreadsheet. I didn't know how I was going to use it--not yet--but I did know I was going to need it.

Then I sat down and thought about punting. Punters have different strategies, depending on where they are on the field. Sometimes they're trying to throw a dart, essentially, and sometimes they're using a rocket launcher. Because of this, there needed to be multiple routines to account for field position, and after looking at the real data (that spreadsheet already coming in handy), I divided the field into three zones, based on punting strategy used in that part of the field.

Then I sorted the spreadsheet data by those same zones. Now I had actual data corresponding properly to the zones.

Still easy, right? Here are the variables I needed to account for, by tier, in addition to punt distance and return distance:
1. Blocked kick chance
2. Touchback chance
3. Fair catch chance
4. Ball downed chance (where the kick isn't touched by the receiving team, which is different from a fair catch)
5. Ball kicked out of bounds chance
6. If it's a fair catch, check for fumble
7. If there's a return, check for fumble
8. If there's a return, check for touchdown based on return distance

All of these chances were calculated using NFL data from the spreadsheet. Plus, the ratings influence needs to correspond to the best in the NFL (with an A+ Special Teams rating) or the worst (with an F Special Teams rating).

All of the possible outcomes needed appropriate messaging, too, and that messaging needed to be limited to a certain number of words, because the message has to both fit into a certain space and be displayed for as little time as possible.

This simple thing needed about 20 hours of intense work to be accurate. Well, and to work.

Site Meter