NHL Mathematical Prediction Model

i'm writing this program that will download a bunch of statistical data and make predictions on the winner of nhl games ...

of course the most challenging part is determining what data to use and how to use it ...

i've come up with the following rough outline ...

teams are divided into elite, average, and bad ... elite are the top 8, bad are the bottom 8, and average are the rest .. i'll use powerrankings to determine the team categories ...

1. weighted wins

elite = 3, average = 2, bad = 1

the last 5 games that a team has played is used to determine a "team point" .. if they beat an elite team they get 3 points and so on ..

2. goal differentials

goals for MINUS goals against for the last 5 games is determined .. this is divided by 2 ... and this result is added or subtracted from part 1 depending on if it's positive or negative ..

3. head to head matchup

3 points are awarded to the winning team in previous head to head games ...



what do you guys think of this? post some other criteria that can be used in this model if you think of anything that might be a good predictor ...

Some other major factors to look for in NHL

Shots/game
ShotsAgainst/game
faceoff%
PP%
PK%
5-5 FA

makaveli

 

Agree!!!