Jump to content

Recommended Posts

Posted

The problem with using statistics to analyze sports games is they tend to group games together, rather than looking at each game for what it is, a single data point in time that will never happen again.  All of the variables for a football game are not accounted for because the rosters, the coaches, the weather, and the stadium are never exactly alike.    Additionally, the point spread is a made up number calculated in some form, by a human.   I love numbers, but I will freely say applying statistics to sports has more to do with bored people who like both statistics and sports wanting to mash them together.

 

I am not saying numbers are useless in sports, by they need to be applied to the portions of the sport with less variables.   Win/Loss predictions include way too many variables to be accurate. 

Posted
7 hours ago, PlayoffsPlease said:

Found this chart and article interesting. 

 

 

1*aGXm_kssZ18BD2pWuPxxiA.png

 

https://medium.com/the-intelligent-sports-wagerer/what-point-spreads-can-teach-you-about-implied-win-probabilities-a8bb3623d2c5

 

According to Oddshark, currently 64% of moneyline bets are on the Bills and against the Jets.   :)

 

So you have a correlation, over 32 teams x 16 games.  I can see where it might be handy to a better in identifying possible "value" moneyline picks.

 

But as a football fan, what one would really like to know is: does this have value as a predictive model for a specific team on a specific game?

I'm gonna guess, "maybe not"

Posted
Just now, Hapless Bills Fan said:

 

So you have a correlation, over 32 teams x 16 games.  I can see where it might be handy to a better in identifying possible "value" moneyline picks.

 

But as a football fan, what one would really like to know is: does this have value as a predictive model for a specific team on a specific game?

I'm gonna guess, "maybe not"

Your guess would be wrong. Unless this formula says 100 % win or 100% loss it cannot be wrong.  Underdogs win regularly.  This model accurately predicts how likely an underdog is to win. 

Posted

This is just common sense. I don't mean the exact formula is, but what the formula is basically saying is that there's an excellent correlation between who the favorite is and who wins the game and the higher the point spread, the more likely the team is to win. Of course there's exceptions such as Vikings - Bills 2018, but hence the 98% r squared vs. 100%.

×
×
  • Create New...