3PA and 2PA correlations with wins

[Millhouse]

Correlation between 3 pt attempts and wins = -22%
Correlation between 2 pt attempts and wins = 18%

Don't settle! Wide open threes and desperation, that's it

Boeheim knows this, of course, just confirming what everyone knows

 

[NineOneSeven]

Wonder how that works in the NBA.

Also expect more teams to shoot threes when they are losing.

 

[dollarbill44]

Is this for SU or NCAA?
Also, I took enough stats classes to know that those aren't very strong correlations (0.18 and 0.22).

 

[Millhouse]

Wonder how that works in the NBA.

Also expect more teams to shoot threes when they are losing.

this was just for our games this year fwiw

also, it could be totally wrong, i did it in 2 minutes and i'm listening to a conference call

 

[CusefanATL]

Correlation between 3 pt attempts and wins = -22%
Correlation between 2 pt attempts and wins = 18%

Don't settle! Wide open threes and desperation, that's it

Boeheim knows this, of course, just confirming what everyone knows

i feel like we are missing some variables here. how many 3 attempts? or are you saying for every 3 we take there is a 22% chance we dont win?

 

[Millhouse]

i feel like we are missing some variables here. how many 3 attempts? or are you saying for every 3 we take there is a 22% chance we dont win?

no, it's just a correlation between the number of 3s or 2s and wins (win=1, loss=0)

it's not super strong, it only explains a little bit of wins and losses. but all else equal, we're better off not shooting lots of threes (which we all already knew because we suck and making 3s - don't do things you suck at)

 

[moqui]

i feel like we are missing some variables here. how many 3 attempts? or are you saying for every 3 we take there is a 22% chance we dont win?

assuming his calculations are correct, r= -.22 means 3 pt attempts explains just under 5% of the variance in wins & losses

 

[Millhouse]

assuming his calculations are correct, r= -.22 means 3 pt attempts explains just under 5% of the variance in wins & losses

0.22 = 22% not 5%

 

[moqui]

0.22 = 22% not 5%

-0.22 is your r. r- square is your effect size, .0484

 

[Millhouse]

-0.22 is your r. r- square is your effect size, .0484

yes - sorry

 

[Pearl309]

This math is making my head spin

 

[Millhouse]

This math is making my head spin

correlation tells you which direction a relationship between two numbers goes (go up or down together vs one goes up other goes down)

r squared tells you how strong the relationship is

there is a lot more to the game than just how many 3s you take so it doesn't explain that much but we do know that all else equal, shooting more threes is bad but doesn't strongly predict anything

 

[Brooky03]

SHUT UP NERDS!