Monte Carlo simulation

[Too Long in the Wasteland]

For those of you not versed in statistics, this is a tool where you have a computer run a model of something over and over again to determine the most likely result. Like all computer programs, it can be GIGO -Garbage In, Garbage Out. After reading (& posting) in the 2-10 thread, I got curious as to what the most likely outcome statistically for the 2010 season is. There are online Monte Carlo simulators, and I plugged the data into one. I tried to be objective, and the result is somewhere around 5 wins.

Here's a link with the results:

http://www.myonlineforecast.com/Results.aspx?id=72da1f6d648a4444a27fbb73085ba82b

Look for the interval where the cumulative line crosses 50%.

Here's what I put in for the median %chances of winning each game:

MTSU 70
SDSU 90
USC 20
N IL 80
NW 50
WISC 25
PURDUE 40
PSU 33
OSU 30
MSU 33
IL 50
IOWA 40

The website also wants lo/hi estimates and confidence %'s. I'm too lazy to type all that data in here. Also, the free version only runs the simulation 100 times. In the real world you typically run this model 1000-10,000 times. So I have no confidence in what the model predicts for the 10% and 90% intervals. I'm also too cheap to buy a license for this website. But it would be an interesting thing to model. Then again, its more fun to see the games for real. Here's the website in case anyone else wants to play around with it or plug in different #'s.

http://www.myonlineforecast.com/HomePage.aspx

 

[frozengopher]

Thank you for reawakening the nightmare of quantitative statistics to me. I'm going to go in the bathroom and ANOVA myself.

Expect whiskey hangover -> receive whiskey hangover! Expect win vs USC -> receive whiskey hangover! (I'm blatantly leaving out the other two options for comedic effect)

 

[KoolAid]

USC Win = Hangover

MIN Win = Hangover

?