Kenpom 2019-2020 rankings

[MGC_07]

https://kenpom.com/

New rankings just came out. MSU is 1st, with almost a 2 point difference to number 2, UK.

Other notable placements

UVA at 5

Purdue at 7

Kansas at 10

Florida State at 14


Take these with a grain of salt. Preseason I tend to think human rankings are better. Kenpom becomes a lot more useful in February/March

 

[imajericho]

4 ACC teams top 6. Notably UL at 3 and UNC at 6.

 

[MGC_07]

4 ACC teams top 6. Notably UL at 3 and UNC at 6.

The top end of the ACC is sick, like always.

 

[KUhawks34]

Lol

 

[ExitFlagger]

The entire concept of preseason computer rankings makes little sense to me. A formula can’t predict a high school player’s performance in college, or the improvement of returnees.

 

[MGC_07]

The entire concept of preseason computer rankings makes little sense to me. A formula can’t predict a high school player’s performance in college, or the improvement of returnees.

Agreed

 

[dukedevilz]

Pretty sure this preseason computer model was devised by @fluoxetine. Some teams are given a little too much respect for last year (ie: Virginia, Purdue, Tennessee).

If you want to hate on a human poll, I'll be releasing my top 50 tomorrow. It's bound to be one of the worst of all-time.

 

[IUfanBorden]

https://kenpom.com/



Take these with a grain of salt. Preseason I tend to think human rankings are better. Kenpom becomes a lot more useful in February/March

THis. KenPom is useless in Nov, Dec and even into Jan.

 

[B-Westy]

THis. KenPom is useless in Nov, Dec and even into Jan.

Is it? For example, Purdue was still a top 20 KP team last year when they were 6-5 and it took them around a month and a half to crack back into the AP rankings. I think KP is way more accurate than any human poll that just looks at wins & losses. That Purdue team was still clearly better than its record indicated, and KP showed that as well.

 

[IUfanBorden]

Is it? For example, Purdue was still a top 20 KP team last year when they were 6-5 and it took them around a month and a half to crack back into the AP rankings. I think KP is way more accurate than any human poll that just looks at wins & losses. That Purdue team was still clearly better than its record indicated, and KP showed that as well.

A system such as KenPom is based off numbers. This is October. Exactly what numbers does Ken have?


Last years.

As I said---KenPom in Nov, Dec and Jan, is pretty useless.

 

[hawkit3113]

Kenpom loves them some Purdue

 

[B-Westy]

A system such as KenPom is based off numbers. This is October. Exactly what numbers does Ken have?


Last years.

As I said---KenPom in Nov, Dec and Jan, is pretty useless.

Well the college basketball community thinks otherwise lol.

 

[IUfanBorden]

Well the college basketball community thinks otherwise lol.

What basketball community? Purdue? Maybe you are misunderstanding my post? I think once Jan, i.e. conference play, rolls around, KP is very accurate. But preseason? No. How can it be? I mean if so, I am really high on Indiana. They just lost to #17 Marquette, by 3, w/o their starting backcourt. Shiiiiiiit, man. Maybe this KenPom shit is right. Hoosiers gonna rock, baby....

 

[KisteK]

The entire concept of preseason computer rankings makes little sense to me. A formula can’t predict a high school player’s performance in college, or the improvement of returnees.

I mean, it did.
... can a human predict better?

Kenpom loves them some Purdue

And it bothers, obviously.

 

[IUfanBorden]

I mean, it did.
... can a human predict better?



And it bothers, obviously.

Who?

 

[FaithPlusOne]

Who?

I’m with you on this one. Statistical models require data. The more good data the better. Until a fair amount of games are played they could potentially give some really odd results. Midway through the year they can be very good.

 

[IUfanBorden]

I’m with you on this one. Statistical models require data. The more good data the better. Until a fair amount of games are played they could potentially give some really odd results. Midway through the year they can be very good.

I mean, to me its common sense. A system that requires data, i.e. numbers, doesn't seem to reliable when there is NO data. In essence, its simply a poll as of now.

 

[lurkeraspect84]

 

[boilerzz]

THis. KenPom is useless in Nov, Dec and even into Jan.

It's actually starting to normalize by mid-December each year. Prior year results fall off as games are played and while prior year is still a factor by then, current year overcomes that for the most part.

But yes, pretty useless in October and November.

 

[hailtoyourvictor]

Pretty sure this preseason computer model was devised by @fluoxetine. Some teams are given a little too much respect for last year (ie: Virginia, Purdue, Tennessee).

If you want to hate on a human poll, I'll be releasing my top 50 tomorrow. It's bound to be one of the worst of all-time.

Fluoxetine works for Kenpom?

 

[dukedevilz]

Fluoxetine works for Kenpom?

Haha no. Just thought some of the rankings were a little strange, much like Fluoxetine's rankings in the early season.

 

[IUBtown]

Purdue ranking makes me pause. I think they'll be pretty solid but never thought they'd be top 10 good. With that said I was wrong last year and KP is a smart man. Will monitor.