Monday, September 08, 2008

Waist high and knee deep

I currently am [title] in NFL data. I figure that I need a data set that I know the quirks of so I can do some better testing, so I have decided to type up the NFL data for this year. Since week one is almost done I have enough data to start playing around so I have been seeing how things look. Here is an example of a graph I have made for factor loadings to predict wins:
Basically the way to road this graph is that the further right you go the more that stat helps you win NFL games, so Rushing Attempts are good, Kick Return Yards are bad. (Think about why getting Kick Return Yards are bad, it makes sense.)

In other news: Lollipop by Mika is a song that is one huge hook and nothing else, I'm not saying that the song is good, but it is brilliant. And to top it off the music video is a visual version of the song. This is as close to candy as you can get while listening to music (diabetics beware):


B.E. Earl said...

I used to do similar modeling for predicting NFL wins for use in an office pool. Then I created another model using several factors tied against random number generators to predict the wins.

The random number generator version worked hella better than the more "scientific" version. In fact, I won the pool (over 150 participants) 2 different weeks in the first half of that season and was leading the cumulative until about week 10.

The winner of cumulative pool? She picked the winner based on which city she liked better and if she had never visited either, she picked the one that she would first like to visit. Won around $6,000 for her troubles.

It's a crapshoot, is what I'm sayin'!

Ookami Snow said...

Did you makes the picks against the spread or straight up?

I wouldn't try to pick against the spread, Vegas has alot of smart guys setting the line.

B.E. Earl said...

I don't remember all the variables exactly. It was a while ago.

Weird thing is that I know so many people who make a lot of money gambling and they always say the same thing. That they can spot a bogus spread from a mile away. If they all say that, how does Vegas make any money on it? Are they throwing them a bone every once in a while? Maybe. Gamblers tend to talk about their big wins, not their big losses.

Ookami Snow said...

One thing to remember about Vegas odds is that they are made so that half the people will pick one way and half will pick the other. A 50% split of bets is the best way to maximize profits. So that doesn't necessarily mean that the spread will actually be the best guess at the actual score difference.

So if I was setting odds I may have to change the New York teams a couple of points just because there will be more people who will be betting on the New York teams no matter the spread. So to counteract that the odds makers need to entice others to bet against New York.

Most companies have no idea what to do when it comes to making decisions based on the science of statistics. Vegas was built on probability and statistics, they know what they are doing. :)