Le Moyne Dolphins Post Game Thread

[The Yakk]

That is a bad interpretation. Regression toward the mean is a tendency not a immutable law. There are many exceptions to regression toward the mean. Sometimes a 6-7 man and a 6-4 inch woman will have a seven foot child. But more often they will produce a child shorter than either. Sometimes a curve buster like Babe Ruth comes along. But more often, a player will hit a number of home runs closer to the league average.

Regression toward the mean is considered a statistical fact, a fundamental concept in statistics. It is NOT the same as "the law of averages" which is simply conventional wisdom.

Therefore one can reasonably predict that a UConn basketball team going 0-10 from the three point line in the first half will perform better from the arc in the second half due to regression toward the mean.

You are saying the same thing Nick Danger said. There is no beef here.

 

[HooperScooper]

How come nobody talks about regression to the mean when a good shooter is in a slump? My favorite non-regression to the mean example is Neils Giffey. No way a career 34% shooter is going to shoot almost 50% the entire season. Yeah, he is. Including the title game.

If you remember that season he was at like 60% from 3 for much of the early part of the season and then went into a slump from 3 for a while.

 

[Hunt for 7]

His comments were interesting. It sounds to me like we'll be seeing a move with Ross or Stewart entering the starting lineup for Mahaney if Mahaney can't show he can do the things Hurley wants him to do as a starter.

On a team with this much depth, all minutes should be earned in practice and in game play. I start to get concerned when he can’t see the play “because of his height” against a low mid major whose tallest player I think was 6’8”. What happens when the opponent players get bigger. So I think he is sending some of that type message.

We can’t expect to hit home runs on all the portal moves. He and the staff have been the best in the country and it seems they hit on Reed Jr. How many Cam’s can we reasonably expect?

Not giving up yet on Mahaney but if he does not pick it up when we start playing tougher comp it will be tough to get the 15 minutes he is getting.

 

[JustbrewitMan]

How come nobody talks about regression to the mean when a good shooter is in a slump? My favorite non-regression to the mean example is Neils Giffey. No way a career 34% shooter is going to shoot almost 50% the entire season. Yeah, he is. Including the title game.

In many games of skill plus a chance element, most people do not realize what an adequate sample size truly is for what they are looking at. It's nearly always larger than what they may think.

For example poker is a game of skill that also has a significant element of chance. Skilled players will be winners in the long term. Yet horrible players can win money for longer periods than what you might think,..and great players can & will lose money over ridiculous sample sizes due to variance/luck.

There are some really nice elegant examples you can find online that show that you literally need to play hundreds of thousands of hands of poker (some argue millions) to prove with high statistical rigor/confidence that you are a winning player.

You get dealt about 30 hands per hour in live poker. It can take literally years of playing for you to state with confidence you are a winning player.

For 3pt shooting, let's say someone shoots 37.5% over a season of 250 attempts. The range of "true" 3pt shooting ability for that player using a 95% confidence interval is between 33-43% for any given 250-shot sample size season.

 

[nomar]

How come nobody talks about regression to the mean when a good shooter is in a slump? My favorite non-regression to the mean example is Neils Giffey. No way a career 34% shooter is going to shoot almost 50% the entire season. Yeah, he is. Including the title game.

I posted these stats in an old thread:

Giffey shot 48.3% for the season from 3 (58/120). He shot 11/13 from 3 in the first 3 games of the season. Then between 2/23 and 3/14, he went 19/31 from 3 (61.3%). Then he cooled off. But after going 0/7 from 3 against Michigan State and Florida, he made 2 of 4 against Kentucky.

 

[Waquoit]

If you remember that season he was at like 60% from 3 for much of the early part of the season and then went into a slump from 3 for a while.

He reverted towards but not to the mean. Not even close.

 

[HooperScooper]

He reverted towards but not to the mean. Not even close.

What was his mean that he was reverting towards?

 

[NickDanger]

That is a bad interpretation. Regression toward the mean is a tendency not a immutable law. There are many exceptions to regression toward the mean. Sometimes a 6-7 man and a 6-4 inch woman will have a seven foot child. But more often they will produce a child shorter than either. Sometimes a curve buster like Babe Ruth comes along. But more often, a player will hit a number of home runs closer to the league average.

Regression toward the mean is considered a statistical fact, a fundamental concept in statistics. It is NOT the same as "the law of averages" which is simply conventional wisdom.

Therefore one can reasonably predict that a UConn basketball team going 0-10 from the three point line in the first half will perform better from the arc in the second half due to regression toward the mean.

I have taught statistics at the University level. You have no idea what you are talking about.
Regression to the mean is merely a tautology; it is not a “tendency.” a series of numbers must regress to the mean, or it simply would not be the mean.
And going 0 for 10 in the first half of a basketball game does not magically cause you to shoot better in the second half!

 

[NickDanger]

How come nobody talks about regression to the mean when a good shooter is in a slump? My favorite non-regression to the mean example is Neils Giffey. No way a career 34% shooter is going to shoot almost 50% the entire season. Yeah, he is. Including the title game.

Yes, he’s shooting improved, so that generated a new mean

 

[NickDanger]

That is a bad interpretation. Regression toward the mean is a tendency not a immutable law. There are many exceptions to regression toward the mean. Sometimes a 6-7 man and a 6-4 inch woman will have a seven foot child. But more often they will produce a child shorter than either. Sometimes a curve buster like Babe Ruth comes along. But more often, a player will hit a number of home runs closer to the league average.

Regression toward the mean is considered a statistical fact, a fundamental concept in statistics. It is NOT the same as "the law of averages" which is simply conventional wisdom.

Therefore one can reasonably predict that a UConn basketball team going 0-10 from the three point line in the first half will perform better from the arc in the second half due to regression toward the mean.

If you would like, I can explain to you the relationship between the law of large numbers and “regression to the mean”.