Champ Check: Eight is Enough (Sorry Gophers, Buckeyes)

[InstiGATOR1]

This is a weekly check on those teams doing the things associated in the past with winning the NCAA tourney. It starts:

Last week, 10 teams had the stats to be champs. But a lot has happened in the last seven days. Many of the nation's top 20 squads suffered tough losses -- Duke got blown out by Miami, Louisville dropped its third straight, and Syracuse lost as well. None of those teams played their way off the champ list, but two Big Ten teams no longer make the grade.

UF does pass the screens. Here is the link:

http://www.cbssports.com/collegebask...tball/21622236

The latter part of the article is also very interesting as the author speculates this years may be a chalk type tourney ala 2007. He posts a graph in the link and then says:

One thing jumps out on this table: Florida's efficiency numbers are head and shoulders above the rest of the top 20. How good are the Gators' KenPom stats? Chew on this: no team in the last nine years has gone into the tournament with a Pythag as high as Florida's. Not Kentucky last year. Not the Gator squads that won back to back championships in 2006 and 2007.

Or course as we all know UF's SEC schedule is back loaded so all the stats people are wowed by may well drop a bit.

[Bradass]

Enjoyed that read. Lots of good info, and nice to have the stats laid out like that comparing the different seasons.

[themistocles]

That was a very interesting read.

As a Numbers runner, I don't honestly think one can say that .965 differs from .958, however, it is quite possible that .987 differs from .96.

Wow - these Gators are incredibly efficient.

Did anyone else notice the average points per game against. The Gators were at 51, while other top defensive teams, like Kansas 59 and Louisville 58, were substantially higher.

The idea that you need to average 73 points per game when you are only giving up 51 is absurd. If this year's Gators end up averaging under 73 and still win the National Championship, he will need to reset his "At least averaging to whatever the Gators average this year."

That is a real problem with setting fractional levels to define something - within 3 points is about the best precision one could possible hope for in such things.

[GatorRade]

Originally Posted by themistocles

The idea that you need to average 73 points per game when you are only giving up 51 is absurd. If this year's Gators end up averaging under 73 and still win the National Championship, he will need to reset his "At least averaging to whatever the Gators average this year."

This is the major failing to this type of method, which you could categorize under the label of "overfitting" the model. Ohio State averages 73.0 ppg and suddenly they aren't fit for being champions. Craziness.

I saw an earlier version of this analysis (I think it was this one), which also said that recent champions have not had win streaks of over 17 games in a row. This is the kind of thing that I am talking about. So a team wins 20 games in a row, so then you believe they can't become champions? Overfitting the model.

Florida's points per possession differential is in the stratosphere. 73 points per game is silly.

[malscott]

That girl in the video is hot! That guy is a dick and should spend his next check on an attitude overhaul and a nose-job. Graph...good.

[corpgator]

Originally Posted by GatorRade

This is the major failing to this type of method, which you could categorize under the label of "overfitting" the model. Ohio State averages 73.0 ppg and suddenly they aren't fit for being champions. Craziness.

I saw an earlier version of this analysis (I think it was this one), which also said that recent champions have not had win streaks of over 17 games in a row. This is the kind of thing that I am talking about. So a team wins 20 games in a row, so then you believe they can't become champions? Overfitting the model.

Florida's points per possession differential is in the stratosphere. 73 points per game is silly.

That's why Pomeroy's numbers are so much better. They account for pace.

Also, they account for the backloaded schedule. Our wins against weaker teams count for less, so closer wins against tougher competition will count the same as long as we keep playing up to our potential.

[InstiGATOR1]

Originally Posted by themistocles

That was a very interesting read.

As a Numbers runner, I don't honestly think one can say that .965 differs from .958, however, it is quite possible that .987 differs from .96.

Wow - these Gators are incredibly efficient.

Did anyone else notice the average points per game against. The Gators were at 51, while other top defensive teams, like Kansas 59 and Louisville 58, were substantially higher.

The idea that you need to average 73 points per game when you are only giving up 51 is absurd. If this year's Gators end up averaging under 73 and still win the National Championship, he will need to reset his "At least averaging to whatever the Gators average this year."

That is a real problem with setting fractional levels to define something - within 3 points is about the best precision one could possible hope for in such things.

It is a screening system. The first time the screen fails, he will change it. So yes it is possible that UF will average less than 73 points a game and still win it all.

[gogators73]

He should do points per possession, not points per game.