MUScoop

MUScoop => Hangin' at the Al => Topic started by: Carl on February 24, 2019, 09:21:45 PM

Title: Is our statistical ranking valid?
Post by: Carl on February 24, 2019, 09:21:45 PM
I'll keep this specific to Kenpom, which I respect as unbiased, if for no other reason than Vegas holding so much stock in it.

-- With our (23-4) record, there is not a single team ranked behind our 24 ranking with fewer losses. 

-- The ones ranked behind us with 4 losses as well... Wofford(26), Cincinnati(31), Murray St(53), Belmont(56), New Mexico St(60).

-- 16 of the 23 teams ranked above us have 4 or more losses.

--  5 of those 16 are Big Ten teams which average 6.2 losses apiece.

The Big Ten undoubtedly had a great non conference performance.  My question is whether or not that is overvalued in the context evaluating the relative strength of a team at this point in the season?  The non-conference schedule is not only a relatively small sample size (so many cupcakes), but also inherently occurs early in the season.

Middle of the road BE teams are a case study in that all of their wins aren't valued as much due to a "down" BE but any loss for any B10 team is a "good" loss.  I think it will net out with too few BE teams in the tourney and a lot of early exits for middling B10 teams.

Bottom line is that Non-Con performance is overvalued and skews all advanced metric models from that point forward.

*Only using the B10 as an example because I'm familiar with the teams and they've had an outlier positive Non-Con
Title: Re: Is our statistical ranking valid?
Post by: TallTitan34 on February 24, 2019, 09:53:44 PM
The committee had us as the 12th best team at the last reveal and we haven't lost since. I trust we will be seeded appropriately even with the models having us where we are.

The statistical models are just a tools to go long with eye tests.

As others have pointed out, not blowing out weaker opponents hurt us in the models. Also someone else pointed out how closely the teams around us are bunched. There's not a big difference from 14 to 31 in KenPom anyway.
Title: Re: Is our statistical ranking valid?
Post by: TallTitan34 on February 24, 2019, 09:56:01 PM
Quote from: TallTitan34 on February 24, 2019, 09:53:44 PM
I trust we will be seeded appropriately even with the models having us where we are.

I do fear where they send us, however.
Title: Re: Is our statistical ranking valid?
Post by: Cheeks on February 24, 2019, 09:56:29 PM
KenPom is not the end all be all.  There are other highly regarded options.

Sagarin has us at 14th, for example.

He's been rating teams since the 1980's.

Title: Re: Is our statistical ranking valid?
Post by: Carl on February 24, 2019, 10:05:13 PM
Fully agree that Kenpom, nor any model, is perfect.  Just opening up a convo on how Non-Con and early season results in general can skew models.

Interestingly, Marquette was 8-4-1 against the spread in their 13 Non-Con games.

10-4 in their 14 conference games.

Maybe this is more about investment advice than banging on the stats guys!  :)

Title: Re: Is our statistical ranking valid?
Post by: Jay Bee on February 24, 2019, 10:15:16 PM
Lotsa close wins early on for us. Had we lost a few of those, people would have crapped on us (here included) even more than they actually did.

We've had some luck, but this is a great team. Top 15 team with potential to go to the top. Just as expected for the past year+.
Title: Re: Is our statistical ranking valid?
Post by: brewcity77 on February 24, 2019, 10:18:36 PM
Quote from: Cheeks on February 24, 2019, 09:56:29 PMHe's been eating teams since the 1980's.

(https://media.giphy.com/media/Ui0xDRG7kxbuU/giphy.gif)
Title: Re: Is our statistical ranking valid?
Post by: Heisenberg on February 24, 2019, 10:30:04 PM
Quote from: brewcity77 on February 24, 2019, 10:18:36 PM
(https://media.giphy.com/media/Ui0xDRG7kxbuU/giphy.gif)

How do I unsee this?

Title: Re: Is our statistical ranking valid?
Post by: Cheeks on February 24, 2019, 10:34:13 PM
Quote from: brewcity77 on February 24, 2019, 10:18:36 PM
(https://media.giphy.com/media/Ui0xDRG7kxbuU/giphy.gif)

Lol.

I hate Apple autocorrect.  Absolutely hate it.  Will fix

Dahmer was eating people my senior year, one was right across the street from where I lived...what an insane Summer that was.
Title: Re: Is our statistical ranking valid?
Post by: Heisenberg on February 24, 2019, 10:39:36 PM
If we replaced Kansas and IU with two RPI 150 blowouts, and everything else the same, any idea where Kenpom would have us?

scheduling hard teams and losing versus blowing out easy teams ... which is the better approach to maximize statistical rankings?
Title: Re: Is our statistical ranking valid?
Post by: Herman Cain on February 24, 2019, 10:46:23 PM
Quote from: Rick Majerus' Towel on February 24, 2019, 10:39:36 PM
If we replaced Kansas and IU with two RPI 150 blowouts, and everything else the same, any idea where Kenpom would have us?

scheduling hard teams and losing versus blowing out easy teams ... which is the better approach to maximize statistical rankings?
That is what The Johnnies did this year.
Title: Re: Is our statistical ranking valid?
Post by: TAMU, Knower of Ball on February 24, 2019, 11:08:57 PM
Quote from: Rick Majerus' Towel on February 24, 2019, 10:39:36 PM
If we replaced Kansas and IU with two RPI 150 blowouts, and everything else the same, any idea where Kenpom would have us?

scheduling hard teams and losing versus blowing out easy teams ... which is the better approach to maximize statistical rankings?

Kansas doesn't hurt us much. It's a "good loss". Replacing Indiana with a blowout of a 150ish team would likely be worth a good amount. I have no idea exactly where we would be but I would be surprised if we were in the top 15. The 7 point win against UTEP is probably the other one you want to replace.

Quote from: Herman Cain on February 24, 2019, 10:46:23 PM
That is what The Johnnies did this year.

Except they struggled against their mediocre teams, hence the KenPom ranking in the mid 50s.
Title: Re: Is our statistical ranking valid?
Post by: brewcity77 on February 25, 2019, 12:07:11 AM
Quote from: Herman Cain on February 24, 2019, 10:46:23 PM
That is what The Johnnies did this year.

More accurately, that's what they tried to do.
Title: Re: Is our statistical ranking valid?
Post by: bilsu on February 25, 2019, 06:46:20 AM
The rankings are not accurate, because they put equal weight on November games. I am confident that we would beat IU at IU right now and Kansas on a neutral court.
Title: Re: Is our statistical ranking valid?
Post by: MUDPT on February 25, 2019, 06:57:21 AM
Quote from: bilsu on February 25, 2019, 06:46:20 AM
The rankings are not accurate, because they put equal weight on November games. I am confident that we would beat IU at IU right now and Kansas on a neutral court.

K state and Louisville are also probably better now then when MU played them.

Pomeroy on his pod talks about running into a hot shooting night/ team and not a ton you can do about it. Indiana and St. John's (part 1) did that to MU which has skewed the rankings. I checked a few weeks ago, but if MU/ IU shoot their season 3% that night, MU still loses, but by 3 points.

I believe Pomeroy is still best predictor going forward. I think Wins Above Bubble is the best metric in looking at resumes.
Title: Re: Is our statistical ranking valid?
Post by: The Sultan on February 25, 2019, 08:08:07 AM
I think what you are going to see from the Big Ten this year is similar to what we saw from the Big East last year.  They are going to get a lot of teams in, but I'm not sure how many will be playing in the second weekend.  I really only feel confident about MSU and Michigan. 
Title: Re: Is our statistical ranking valid?
Post by: BrewCity83 on February 25, 2019, 09:20:37 AM
Quote from: bilsu on February 25, 2019, 06:46:20 AM
The rankings are not accurate, because they put equal weight on November games. I am confident that we would beat IU at IU right now and Kansas on a neutral court.

Sagarin includes a "Recent" ranking which weights the more recent games more heavily than the early-season ones.  Marquette is #7 in today's "Recent" rankings...
https://www.usatoday.com/sports/ncaab/sagarin/
(see far right column).
Title: Re: Is our statistical ranking valid?
Post by: copious1218 on February 26, 2019, 10:08:47 AM
Quote from: MUDPT on February 25, 2019, 06:57:21 AM
K state and Louisville are also probably better now then when MU played them.

Pomeroy on his pod talks about running into a hot shooting night/ team and not a ton you can do about it. Indiana and St. John's (part 1) did that to MU which has skewed the rankings. I checked a few weeks ago, but if MU/ IU shoot their season 3% that night, MU still loses, but by 3 points.

I believe Pomeroy is still best predictor going forward. I think Wins Above Bubble is the best metric in looking at resumes.

Can someone describe this?
Title: Re: Is our statistical ranking valid?
Post by: Silkk the Shaka on February 26, 2019, 10:12:00 AM
Quote from: copious1218 on February 26, 2019, 10:08:47 AM
Can someone describe this?

https://sports.sites.yale.edu/yusag-bracketology

Wins Above Bubble (WAB): The difference in the number of wins a team has compared to the expected number of wins an average "bubble" team would earn against a given teams' schedule. We compute WAB by computing a team's win expected win difference against each of the at-large teams ranked 32-40 by YUSAG coefficient and averaging these differences
Title: Re: Is our statistical ranking valid?
Post by: copious1218 on February 26, 2019, 10:43:08 AM
Quote from: Ellenson Family Reunion on February 26, 2019, 10:12:00 AM
https://sports.sites.yale.edu/yusag-bracketology

Wins Above Bubble (WAB): The difference in the number of wins a team has compared to the expected number of wins an average "bubble" team would earn against a given teams' schedule. We compute WAB by computing a team's win expected win difference against each of the at-large teams ranked 32-40 by YUSAG coefficient and averaging these differences

Thank you!
EhPortal 1.39.9 © 2025, WebDev