https://kenpom.com/index.php

Kinda surprised it's that good

Thought they didn't use recruiting data.

Always the problem with this cr@p every year. They starts out with arbitrary data (Memphis #50, what if he decided they were #10 or #90?) and then tries to explains that it disappears over time, but the new data came from the made-up old data. This type of ranking should Start January 1 for basketball and the sample size is too small for football until the season is over.

(10-20-2019 12:50 PM)cmt Wrote:  Always the problem with this cr@p every year. They starts out with arbitrary data (Memphis #50, what if he decided they were #10 or #90?) and then tries to explains that it disappears over time, but the new data came from the made-up old data. This type of ranking should Start January 1 for basketball and the sample size is too small for football until the season is over.

People do this every year...

The preseason bias is gone mid season.

If the data can be used on January 1st to rank teams. All the preseason bias can be removed by January 1st. It is just math and the numbers would be exactly the same

How can a computer model already have teams ranked?

Sent from my SM-G965U using Tapatalk

(10-20-2019 11:16 AM)AlonsoWDC Wrote:  Kinda surprised it's that good

Thought they didn't use recruiting data.

According to his blog 'the top 30 or so have an impact on team rating but beyond that it is mostly blind to newcomers'.

Projected record: 21-10, 11-7

Projected losses:
Oregon 77-73
NC State 83-80
Tennessee 81-75
@Wichita State 77-75
@South Florida 76-75
@Cincy 80-74
@UConn 78-77
@Houston 78-72

Iffy wins:
Cincy 77-76
@Tulsa 78-75
@UCF 76-73
Houston 76-75
@SMU 76-75

I don't think people would be happy with that.

lost at USF?

(10-22-2019 08:10 AM)fsquid Wrote:  lost at USF?

Yep but 49% so on the cusp currently. Oddly enough he has us winning by 7 against them at home. Tennessee (27%), @Cincy (29%) and Oregon (37%) are the games we have the least chance of winning...for now.

2019 KENPOM end of season ranking for your Memphis Tigers
#58 19-12 / 11-7 / 22-14 including post season

So, the almighty, non-biased, algorithm driven, "I don't tweak it," best guess is that this year's Tigers, with 3 returning contributors, the number one recruiting class in the country including a 7'1" number one player in the country projected as the number player in the NBA draft, Kentucky's small forward, Kansas' power forward, Duke's point guard, Indiana's shooting guard, plus Thomas, Baugh, Dandridge, Boyce and Stokes will be

#50 21-10 / 11-7

Yet our friends a little to the west
2019 KENPOM end of season ranking for the Arkansas Razorbacks

#54 17-14 / 8-10 / 18-16

Arkansas has a recruiting class with one 4* and three 3*s ranked 157th nationally including the ever popular Conner Vannover, but return almost everyone from a team that went 18-16 except their 6'11" leading scorer. They will have no inside play and a first year coach, but they are int he SEC. Kenpom predicts

#44

He can talk all he wants about how "its the computer, I'm not manipulating anything" I call BS just like I do every year.

(10-22-2019 08:29 AM)cmt Wrote:  2019 KENPOM end of season ranking for your Memphis Tigers
#58 19-12 / 11-7 / 22-14 including post season

So, the almighty, non-biased, algorithm driven, "I don't tweak it," best guess is that this year's Tigers, with 3 returning contributors, the number one recruiting class in the country including a 7'1" number one player in the country projected as the number player in the NBA draft, Kentucky's small forward, Kansas' power forward, Duke's point guard, Indiana's shooting guard, plus Thomas, Baugh, Dandridge, Boyce and Stokes will be

#50 21-10 / 11-7

Yet our friends a little to the west
2019 KENPOM end of season ranking for the Arkansas Razorbacks

#54 17-14 / 8-10 / 18-16

Arkansas has a recruiting class with one 4* and three 3*s ranked 157th nationally including the ever popular Conner Vannover, but return almost everyone from a team that went 18-16 except their 6'11" leading scorer. They will have no inside play and a first year coach, but they are int he SEC. Kenpom predicts

#44

He can talk all he wants about how "its the computer, I'm not manipulating anything" I call BS just like I do every year.

He's just running a macro in excel. I'm sure he's not changing anything.

It's a poor model for forecasting. I wouldn't worry too much about this.

(10-22-2019 10:31 AM)FlyingTiger2016 Wrote:  

(10-22-2019 08:29 AM)cmt Wrote:  2019 KENPOM end of season ranking for your Memphis Tigers
#58 19-12 / 11-7 / 22-14 including post season

So, the almighty, non-biased, algorithm driven, "I don't tweak it," best guess is that this year's Tigers, with 3 returning contributors, the number one recruiting class in the country including a 7'1" number one player in the country projected as the number player in the NBA draft, Kentucky's small forward, Kansas' power forward, Duke's point guard, Indiana's shooting guard, plus Thomas, Baugh, Dandridge, Boyce and Stokes will be

#50 21-10 / 11-7

Yet our friends a little to the west
2019 KENPOM end of season ranking for the Arkansas Razorbacks

#54 17-14 / 8-10 / 18-16

Arkansas has a recruiting class with one 4* and three 3*s ranked 157th nationally including the ever popular Conner Vannover, but return almost everyone from a team that went 18-16 except their 6'11" leading scorer. They will have no inside play and a first year coach, but they are int he SEC. Kenpom predicts

#44

He can talk all he wants about how "its the computer, I'm not manipulating anything" I call BS just like I do every year.

He's just running a macro in excel. I'm sure he's not changing anything.

It's a poor model for forecasting. I wouldn't worry too much about this.

LOL, he's not using Excel.

(10-22-2019 10:31 AM)FlyingTiger2016 Wrote:  

(10-22-2019 08:29 AM)cmt Wrote:  2019 KENPOM end of season ranking for your Memphis Tigers
#58 19-12 / 11-7 / 22-14 including post season

So, the almighty, non-biased, algorithm driven, "I don't tweak it," best guess is that this year's Tigers, with 3 returning contributors, the number one recruiting class in the country including a 7'1" number one player in the country projected as the number player in the NBA draft, Kentucky's small forward, Kansas' power forward, Duke's point guard, Indiana's shooting guard, plus Thomas, Baugh, Dandridge, Boyce and Stokes will be

#50 21-10 / 11-7

Yet our friends a little to the west
2019 KENPOM end of season ranking for the Arkansas Razorbacks

#54 17-14 / 8-10 / 18-16

Arkansas has a recruiting class with one 4* and three 3*s ranked 157th nationally including the ever popular Conner Vannover, but return almost everyone from a team that went 18-16 except their 6'11" leading scorer. They will have no inside play and a first year coach, but they are int he SEC. Kenpom predicts

#44

He can talk all he wants about how "its the computer, I'm not manipulating anything" I call BS just like I do every year.

He's just running a macro in excel. I'm sure he's not changing anything.

It's a poor model for forecasting. I wouldn't worry too much about this.

It is a poor model for forecasting an upcoming season..

But once enough games are played it has proven to be a very good model for measuring a teams strength.

Honestly I don't think he puts much effort into trying to get his pre-season model right because he knows it will correct itself.

(10-22-2019 10:31 AM)FlyingTiger2016 Wrote:  

(10-22-2019 08:29 AM)cmt Wrote:  2019 KENPOM end of season ranking for your Memphis Tigers
#58 19-12 / 11-7 / 22-14 including post season

So, the almighty, non-biased, algorithm driven, "I don't tweak it," best guess is that this year's Tigers, with 3 returning contributors, the number one recruiting class in the country including a 7'1" number one player in the country projected as the number player in the NBA draft, Kentucky's small forward, Kansas' power forward, Duke's point guard, Indiana's shooting guard, plus Thomas, Baugh, Dandridge, Boyce and Stokes will be

#50 21-10 / 11-7

Yet our friends a little to the west
2019 KENPOM end of season ranking for the Arkansas Razorbacks

#54 17-14 / 8-10 / 18-16

Arkansas has a recruiting class with one 4* and three 3*s ranked 157th nationally including the ever popular Conner Vannover, but return almost everyone from a team that went 18-16 except their 6'11" leading scorer. They will have no inside play and a first year coach, but they are int he SEC. Kenpom predicts

#44

He can talk all he wants about how "its the computer, I'm not manipulating anything" I call BS just like I do every year.

He's just running a macro in excel. I'm sure he's not changing anything.

It's a poor model for forecasting. I wouldn't worry too much about this.

Not worried. It’s like a weatherman’s prediction.

(10-22-2019 10:31 AM)FlyingTiger2016 Wrote:  It's a poor model for forecasting.

I mean it's not, in general. But the model doesn't really account for a team that's finished 87, 75, 101, 161, and 56 the past five years signing the number one recruiting class. And because of that...

Quote: I wouldn't worry too much about this.

..is probably the right way to look at it.

(10-22-2019 10:50 AM)mairving Wrote:  ...

LOL, he's not using Excel.

It wouldn't really matter if he was.

Deleted my previous post. Going there is pointless.

It is such a different scenario that our ranking might as well be decided by a 3 year old. But it seems as if it was. Can you fathom a team with top 10 talent being lower than 50?

Luckily basketball championships aren't decided like football championships. The ability to decide it on the court is HUGE!

GTG

As I posted in the preseason rankings thread a couple of weeks ago, KenPom's preseason ranking was shown to be the largest deviator of the same KenPom postseason ranking results in a few recent seasons, when compared to other preseason lists.

The algorithm methods are just not good at this - especially with a program like Memphis that has changed so much in the past couple of years. They rely too much on recent seasons and returning players.

They will get the vast majority of teams right but miss on a couple every year. The human preseason polls do as well, but at least they do apply some common sense - which is usually more helpful in preseason.