Jump to content

  • Welcome!

    Register and log in easily with Twitter or Google accounts!

    Or simply create a new Huddle account. 

    Members receive fewer ads , access our dark theme, and the ability to join the discussion!

     

So I modeled the season (and you aren't going to like it)

Happy Panther

By Happy Panther,
in Carolina Panthers

Recommended Posts

Happy Panther Grand Master

Happy Panther

Posted
  • Author
Posted

Does the model take each game as independent events?  Any after bye week biases (for or against depending on the team trends)?  Etc...

Yes each game and score is completely independent (sort of). Meaning I haven't attempted to build in correlations or relationships which would be way beyond the capabilities of excel.

 

The independent scores are then adjusted for a predetermined power ranking line (determined by me) where teams are either given points or taken away points.

 

So if Denver and Houston both get a randomly selected score of 24 they would tie under a coin flip scenario assuming they are equally matched teams. However in the real world Denver wins by a landslide because they are spotted 10+ points.

 

Another way to look at it is Houston has to roll a 6 to win on a regular die within the simulation otherwise Denver wins.

Link to comment
Share on other sites
Happy Panther Grand Master

Happy Panther

Posted
  • Author
Posted

Did u run the simulation 1000 times for each game? That is only way to get accurate numbers.

Again sort of. 

 

I ran the score for each individual game 10,000 times (based on predetermined parameters selected by me!) and then set them in stone. So imagine a list of 10,000 numbers posted on your wall which represent the possible scores for any NFL game. There are lots of scores of 24, some scores of 12 and 35 and very very few scores of 3 or 50.

 

I then set the model up so that every game (256 of them) randomly selects two scores from the 10,000 independent scores on your wall.

Link to comment
Share on other sites
BANE Newbie

BANE

Posted
Posted

Does the model take into account weather in the late months? For example GB or NE will more than likely be much tougher at home in Nov, Dec than they would in say September, even if said team is already tough to beat at home.

 

 

I mean for the most part I think you make good points, in that there WILL be a team or teams that surprise, either up or down, but Im not sure an algorithm can predict that better than a human looking at the schedule, when each game is played and the players playing in said game. 

Link to comment
Share on other sites
sharkkiller Rookie

sharkkiller

Posted
Posted

Again sort of. 

 

I ran the score for each individual game 10,000 times (based on predetermined parameters selected by me!) and then set them in stone. So imagine a list of 10,000 numbers posted on your wall which represent the possible scores for any NFL game. There are lots of scores of 24, some scores of 12 and 35 and very very few scores of 3 or 50.

 

I then set the model up so that every game (256 of them) randomly selects two scores from the 10,000 independent scores on your wall.

 

 

The last part is what I have a problem with. Why would you randomly select the scores individually, unless I am misunderstanding. You should randomly select the scores of 1 of the 10,000 games, not randomly select one score, then randomly select the other. Otherwise, with that number of simulations, you could just put up every possible combination of scores, and select 2 numbers at random and get the same results.

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

Go to topic listing
×
×
  • Create New...