to ganchrow about a basketball stats question

[pico]

since you're a resident whiz kid and i am too lazy to look up the stats, i am wondering what is the correlation (R^2) between the winner of the all star game and the nba champion...that is if the west wins the all star game, how likely they're going to win the championship?

i am curious now, since the west won the all star game and the spurs are most likely going to win the championship.

[Ganchrow]

since you're a resident whiz kid and i am too lazy to look up the stats, i am wondering what is the correlation (R^2) between the winner of the all star game and the nba champion...that is if the west wins the all star game, how likely they're going to win the championship? </script>

i am curious now, since the west won the all star game and the spurs are most likely going to win the championship.

I don't have the data, but I have little doubt a correlation exists.

I'd just have a very hard time believing that knowing the winner of the All-Star game would provide you any information not already implicitly valued in the market price.

[BuddyBear]

even if there were somehow to be an association between All-star region winner (East/West) and Finals Winner (East/West) it would not be the basis for any sort of cause-and-effect reltionship. All it would be is an association...you couldn't possibly rule out all the other variables that would confound that relationshp.

Personally, I have a difficult time believing such an association exists. I can look it up if I get a chance.

[Ganchrow]

even if there were somehow to be an association between All-star region winner (East/West) and Finals Winner (East/West) it would not be the basis for any sort of cause-and-effect reltionship. All it would be is an association...you couldn't possibly rule out all the other variables that would confound that relationshp.

Personally, I have a difficult time believing such an association exists. I can look it up if I get a chance.

You're right -- correlation certainly does not imply causation. But that doesn't necessarily render knowledge of a potential relationship worthless.

Two canonical examples of what's known as spurious correlation:

1. The relationship between the number firetrucks at a fire scene and the amount of fire damage -- a greater number of fire trucks implies there's likely to be more fire damage, and more fire damage implies there's likely to be more fire trucks. However, one does not cause the other. Both are the result of a separate underlying factor, namely that a greater intensity fire causes both the need for more fire trucks as well as more fire damage. Nevertheless, if you were some sort of fire damage claims adjuster, all else being equal you'd certainly be correct to target those fires attended by more firetrucks.
2. The relationship within a population between ice cream sales and mosquito bites -- more mosquito bites implies higher ice cream sales and vice-versa. Again, one doesn't cause the other, instead the underlying cause is that on hot and humid days people tend to consume more ice cream, and on hot and humid days there also tend to be more mosquitoes present.

The situation with the NBA Finals/All-Star game is analogous. A conference winning the All-Star obviously doesn't cause the team from that conference to win the finals. Instead, the conference that wins the All-Star game will tend to have superior players, a disproportionate number of whom will tend to be on the team that won that conference's championship.

Without looking at the data I feel confident in hypothesizing that a relationship between the two exists, and suspect that if all you knew was the winner of the All-Star game and were able to bet at even odds (minus some modest vig) you'd have a profitable long-term strategy. That said, unless you're betting with your 9-year old nephew chances are you won't generally be able to wager at even odds (and the times you could would be the times you wouldn't want to). Therein lies the problem.

Knowing the winner of the All-Star doesn't reveal any new information not already well-known to the market. Sure if you're betting a game in a naïve market at pick'em it has positive value, but in the real the world with real markets that's just not going to be the case.

[Arilou]

It would make sense that the side with better players is a favorite to win each, creating a (small) correlation. But that should be all.