dook only drops...

[gary-7]

...to #4. Typical.

Nova #1 --- expected because they're undefeated, although I'll say without any reservation that MSU is better.

We're #7 in both polls, which again, is fine with me.

Laughing about ASU getting 1st place votes. Oh, they've played well and deserve a top-10 appearance, but we'd absolutely boat-race those schmoes.

http://www.espn.com/mens-college-basketball/rankings

 

[MWHeels]

I knew Duke wouldn't drop out of the top 5.

 

[pln2013]

Duke has early wins over Michigan State, Florida, and Texas.

Dropping 3 spots isn't unreasonable.

 

[MWHeels]

Duke has early wins over Michigan State, Florida, and Texas.

Dropping 3 spots isn't unreasonable.

Florida lost at home to Loyola. It remains to be seen how good they really are.

 

[2DDIMOND]

I knew Duke wouldn't drop out of the top 5.

I didn't expect them to, they actually dropped 1 spot further than I thought they would.

 

[Hark_The_Sound_2010]

Conference road games can be tough, especially when you don't proactively book true road games in the OOC schedule.

 

[al would]

I didn't think they would drop that far either .. I mean really .. it was a tough ACC road game with a hostile crowd and they are taking in to consideration that
it's early in the year and dook is a young team with very little experience .. especially off the bench . Which surprisingly enough is how they will end the season somewhere around mid-March 2018. jmho

 

[pln2013]

Florida lost at home to Loyola. It remains to be seen how good they really are.

Florida doesn't look like the elite team they were thought to be, but they're still very good.

Florida #21 on Kenpom. Texas is #26. MSU is #2. Those are 3 very good wins.

 

[Archer2]

I'm surprised they dropped at all. They are just that good. Print the 39-1 shirts now.

 

[Steat]

I'm surprised they dropped at all. They are just that good. Print the 39-1 shirts now.

Heck, print the 79-1 shirts now.

 

[viking131]

I'm surprised they dropped at all. They are just that good. Print the 39-1 shirts now.

Good one Archer!

 

[TPFKAPFS]

Forget that! After the Rat Lord makes his demands, the loss will go on Gaudet's record and they will print the 39-0 tee-shirts!

 

[TPFKAPFS]

BTW: Kenpom has so little to do with the real world that it is laughable! Texas and Florida are no where near as good as advertised. MSU is indeed a good win though. puke should be outside the top 10, but who really expected them to be?

I'm not surprised by puke's weak drop, but I am very surprised that we did not rise very far!

 

[FlaTarHeel]

I'm surprised they dropped at all. They are just that good. Print the 39-1 shirts now.

Best laugh I had all day, thanks man.

 

[RoseHeel]

Agree with pln.

Dook dropping out of the top 5 was highly unlikely. Outside of MSU and Nova, not many teams you can put ahead of them right now, IMO.

I'm more surprised that ASU and Miami leapfrogged us, despite our only loss being to a team that hasn't dropped out of the top 5 since Week 1.

 

[gary-7]

BTW: Kenpom has so little to do with the real world that it is laughable!

I hafta say I cringe a bit every time I see it referenced. Oh, some of the obvious team stuff they track is interesting --- although some of it devolves into statistical wanking.--- but any rankings they offer at this point of the season on a short sample space with little contextul comparison availability is less than meaningless.

 

[pln2013]

Computer models capture far, far more information than any human can. Basketball is extremely complex, and the more information you use, the better your understanding can be. As of 12/10, there had been 1678 games this year. 47 days of basketball - not days since the season started, but total minutes of game clock. That's a lot of information. Models don't interpret the information perfectly, but neither do humans. Well-tuned models are significantly more predictive than human predictions. That's a well-established fact. If you disagree, go make your money beating Vegas.

 

[gauchoheel]

I hafta say I cringe a bit every time I see it referenced. Oh, some of the obvious team stuff they track is interesting --- although some of it devolves into statistical wanking.--- but any rankings they offer at this point of the season on a short sample space with little contextul comparison availability is less than meaningless.

Kenpom's rankings are simply point differentials weighted for opponent strength. So unless you think us beating Arkansas by 19 points, beating Michigan by 15, or winning all of our games by double digits is less than meaningless info, it really doesn't follow that the Kenpom rankings would be meaningless. There are certainly flaws, but all rankings/systems have those. Heck, this topic is about a much more flawed system than Kenpom, the AP Poll.

 

[pln2013]

If you prefer to keep modeling out of it (the inclusion of a pre-season prior to stabilize rankings, adjustments to avoid over-weighting blowouts, etc --- things that increase predictive power), you can use sports-reference.com's simple rating system (SRS). This is the average difference between your margin and your opponent's average margin. So if you win by 18 over a team that on average wins by 12, that'd be +30.

By that metric, MSU is #3, Texas is #23, and Florida is #27. Barely different from Kenpom's.

'Nova is #1. Kansas is #2. We're #6. ASU is #7. Duke is #11. Tennessee is #16.

 

[Archer2]

Pretty big difference between 4 and 11, but that is easily explained by d00k being coached by the GCOAT and all the preseason buildup. They're invincible, I tell you. Oh, wait. Well, they won't lose another game, that's for sure.

 

[jah420]

The dookie always drops

 

[TPFKAPFS]

The second model seems to fit what I see much better than Kenpom so I would take it as the lesser of two evils! Since I never gamble, I'm unlikely to bet either for or against Vegas, but I do enjoy the cheap food/booze/hotels, the scantily clad servers, and the awesome shows!

I share one thing in common with the forces of nature, college sports, humanity, and computer indices: We all must bow before the Rat Lord!

puke will never lose another game!

 

[pln2013]

The second model seems to fit what I see much better than Kenpom so I would take it as the lesser of two evils!

But Kenpom fits what actually happens better. That's the point of predictive analytics - to create the most accurate forecast, independent of subjective biases.

 

[CrazyHeel]

Dookies shouldn't panic, they'll move up next time they lose.

 

[TPFKAPFS]

You presume that I am interested in the most accurate prediction possible. I'm not. What I find interesting is comparing what I see when watching teams play each other to what other experts see when they watch. I like the human polls because they can take emotion into account and other intangibles like revenge, chemistry, and momentum. I like that the best team doesn't always win and I like to watch hot/motivated teams out play their predictions. I have no interest what so ever in certainty or in the accuracy of a predictive model. (I do understand how they work and what data is part of their algorithms though) I want to see Lehighs, Mercers, and GCUs make it to the money rounds and show that emotion!

I want to watch a beautiful game with an uncertain outcome and see the tears in the eyes of the victors!

I also want that victor to always be the Tar Heels!

 

[gunslingerdick]

Screw Kenpom, BPI, RPI and any other number crunching analytics. I've watched basketball long enough to know I don't need analytics to tell me who's good and who's not. I trust my eye test far more than some computer model that doesn't take real life circumstances into account. Such as, does Kenpom reflect that Roy called off the dogs in pretty much every game we've played by playing wacky line ups and zone defense. If we wanted to have beaten Tulane and Western by 60, we could have. Does Kenpom take into account Berry's hand injury? Cam's absence and what he means to our team? Does Kenpom take into account we're the f*cking defending national champions and we're just biding our time until real and meaningful games start? Our guys have done this and we know it's a long season and what it takes to win a championship. And going "balls to the wall" in a game against Davidson in November just so we look better to Kenpom or in some other formula doesn't factor into that. I guess it's something for people to talk about in December but it's completely meaningless to me.

 

[pln2013]

You presume that I am interested in the most accurate prediction possible. I'm not.

You're ostensibly interested in things that relate to "the real world". Data tells you a lot about the real world. If you're trying to evaluate how good a team is and how good its wins are, the computer rankings are a valuable tool.

What I find interesting is comparing what I see when watching teams play each other to what other experts see when they watch. I like the human polls because they can take emotion into account and other intangibles like revenge, chemistry, and momentum. I like that the best team doesn't always win and I like to watch hot/motivated teams out play their predictions.

I agree - if everything played out as it was supposed to on paper, basketball would be super boring. The variance is what makes it fun. I'm not saying you should trust the spreadsheets and then never watch games.

I have no interest what so ever in certainty or in the accuracy of a predictive model. (I do understand how they work and what data is part of their algorithms though)

And that's fine - I'm not saying you have to care about the models. But to say they have "so little to do with the real world that it is laughable!" is to ignore how much information they can provide. If we're evaluating how good Duke is, they provide a lot of information that helps inform that evaluation.

I want to see Lehighs, Mercers, and GCUs make it to the money rounds and show that emotion!

Same!

I want to watch a beautiful game with an uncertain outcome and see the tears in the eyes of the victors!

I also want that victor to always be the Tar Heels!

Ditto!

 

[pln2013]

Screw Kenpom, BPI, RPI and any other number crunching analytics. I've watched basketball long enough to know I don't need analytics to tell me who's good and who's not. I trust my eye test far more than some computer model that doesn't take real life circumstances into account.

All the same, the computer models will outpredict you.

Such as, does Kenpom reflect that Roy called off the dogs in pretty much every game we've played by playing wacky line ups and zone defense.If we wanted to have beaten Tulane and Western by 60, we could have.

To an extent, yes. For good models, a 60 point win isn't treated much differently from a 30 point win - a blowout is a blowout.

Does Kenpom take into account Berry's hand injury? Cam's absence and what he means to our team?

Not sure if Kenpom specifically does, but a lot of rankings do adjust for injuries (particularly longer-term ones).

Does Kenpom take into account we're the f*cking defending national champions

It has a pre-season prior that incorporates overall program success. The "defending national champions" part doesn't mean much, but the baseline performance of Roy Williams teams does. Plus it accounts for returning and incoming players.

and we're just biding our time until real and meaningful games start?

Not much evidence to support this claim. Our guys are busting their asses out there. This is hardly the '08-'09 team.

I guess it's something for people to talk about in December but it's completely meaningless to me.

You don't have to care about it. But it can tell you a lot all the same.

 

[TPFKAPFS]

The only information I need about puke is how much they suck! I ignore anything that might even hint that they might suck less this year! I don't need every bit of information possible to enjoy the game and I don't want anything more than my eyes/knowledge of the college landscape to predict games!

If others want this info than have at it, but it means nothing to me! The AP poll does mean something even though I disagree with the current rankings.

The only time I ever even see these models is when peeps bring them up on here!
(and the occasional reference by a talking head with nothing else to say!)

 

[gary-7]

Computer models capture far, far more information than any human can. Basketball is extremely complex, and the more information you use, the better your understanding can be. As of 12/10, there had been 1678 games this year. 47 days of basketball - not days since the season started, but total minutes of game clock. That's a lot of information. Models don't interpret the information perfectly, but neither do humans. Well-tuned models are significantly more predictive than human predictions. That's a well-established fact. If you disagree, go make your money beating Vegas.

Pfffttt. Basketball ain't a video game, bubba.

 

[pln2013]

Pfffttt. Basketball ain't a video game, bubba.

Never said it was.

 

[gary-7]

Never said it was.

Maybe not in so many words, but that's how you treat it on here, and not just in this thread.

I have nothing against stats. Hell, one of my degrees is in mathematics, and I utilized stats when I coached --- more than my counterparts, probably --- but I also know their limitations. Point is, stats can be helpful in analyzing/improving your team's performance, but you're throwing around kenpom like it's an omniscient authority on all things basketball.

When you play against an actual team, a basketball game is about systems, personnel, match-ups, coaching, execution and officiating. You don't scout an opponent by running to kenpom, you dissect what you see on film or in person.

 

[gunslingerdick]

To an extent, yes. For good models, a 60 point win isn't treated much differently from a 30 point win - a blowout is a blowout.
.

What about Davidson? My eye test tells me we could have won that game by 30 and we might have if we were in late season mode where Roy has his line ups set. I would guess a 10 point win is viewed differently than a 30 point win.

Not sure if Kenpom specifically does, but a lot of rankings do adjust for injuries (particularly longer-term ones).
.

Do they adjust for Roy calling the dogs off like he has in many, if not all our wins? Does Kenpom take into account that Roy is a hell of a guy and chooses to go zone when we're absolutely throttling a team so that it doesn't crush that opponent's spirit and gives us a chance to work on a defense that we're unlikely to employ throughout the year?

It has a pre-season prior that incorporates overall program success. The "defending national champions" part doesn't mean much, but the baseline performance of Roy Williams teams does. Plus it accounts for returning and incoming players.
.

Well, then I've already found at least one fault with the formula.

Not much evidence to support this claim. Our guys are busting their asses out there. This is hardly the '08-'09 team.
.

Not much evidence to prove it false either. It's what I believe.

You don't have to care about it. But it can tell you a lot all the same.

It can tell you a lot. It's not telling me anything because I choose to disregard it.

 

[pln2013]

What about Davidson? My eye test tells me we could have won that game by 30 and we might have if we were in late season mode where Roy has his line ups set. I would guess a 10 point win is viewed differently than a 30 point win.

Every single game has quirks and caveats like that.

Have you watched every game 'Nova has played? MSU? UVA? Wichita State? Duke? Texas A&M? Kansas? Purdue? TCU? St. John's? New Mexico State?

Computer models account for those quirks and caveats imperfectly. Your eye test doesn't account for 99% of them at all because your eyes don't see them.

Rankings and predictions are ultimately comparative - if you make those "eye test" adjustments for UNC but not other teams, your comparison will have a systemic bias.

Do they adjust for Roy calling the dogs off like he has in many, if not all our wins? Does Kenpom take into account that Roy is a hell of a guy and chooses to go zone when we're absolutely throttling a team so that it doesn't crush that opponent's spirit and gives us a chance to work on a defense that we're unlikely to employ throughout the year?

Do you know when other teams do that?

And some models factor in concepts like game control - if you spend the game up 30 and win by 15 after a late push against your bench players, that's treated differently from a game in which you pull away late. Not sure about the specifics of Kenpom with regards to that.

Well, then I've already found at least one fault with the formula.

Winning the championship is too "random" to have much predictive power. Winning isn't destiny. We could've easily lost to Arkansas in R32.

Not much evidence to prove it false either. It's what I believe.

If players aren't giving their all, Roy needs to address that.

It can tell you a lot. It's not telling me anything because I choose to disregard it.

There's a wealth of information out there. Use it or don't. Up to you.

 

[gunslingerdick]

Ever single game has quirks and caveats like that.

Have you watched every game 'Nova has played? MSU? UVA? Wichita State? Duke? Texas A&M? Kansas? Purdue? TCU? St. John's? New Mexico State?

Computer models account for those quirks and caveats imperfectly. Your eye test doesn't account for 99% of them at all because your eyes don't see them.

Rankings and predictions are ultimately comparative - if you make those "eye test" adjustments for UNC but not other teams, your comparison will have a systemic bias.



Do you know when other teams do that?

And some models factor in concepts like game control - if you spend the game up 30 and win by 15 after a late push against your bench players, that's treated differently from a game in which you pull away late. Not sure about the specifics of Kenpom with regards to that.



Winning the championship is too "random" to have much predictive power. Winning isn't destiny. We could've easily lost to Arkansas in R32.



If players aren't giving their all, Roy needs to address that.



There's a wealth of information out there. Use it or don't. Up to you.

Right...I don't think you're understanding what I'm saying. You see, there's this thing called the NCAA tournament that ends each season. The team that wins all their games is the champion. And whether Kenpom said that team would win or not is irrelevant. They're not handing out trophies according to kenpom. So what's the point?

 

[pln2013]

Maybe not in so many words, but that's how you treat it on here, and not just in this thread...but you're throwing around kenpom like it's an omniscient authority on all things basketball.

If I come off that way, it's not my intent. I don't view Kenpom as "an omniscient authority". It's a helpful tool.

My opinion of UNC is mostly informed by my own eye test, expert analysis (including yours!), and more granular data (especially from Adrian Atkinson).

But what do I know about Texas? Basically nothing. I only watched them against Duke. Kenpom and other models help contextualize what I saw. Texas is a good-not-great team. Texas was a good win (and your resume is ultimately about wins and losses, not margin), but Duke barely beating them indicates Duke isn't as good as I might've thought.

What about Purdue? I haven't seen them once this year. Kenpom tells me they're pretty damn good. That's helpful, because otherwise I'm blind.

When you play against an actual team, a basketball game is about systems, personnel, match-ups, coaching, execution and officiating. You don't scout an opponent by running to kenpom, you dissect what you see on film or in person.

Agreed!

Predictive analytics don't have much prescriptive power. They don't tell you, as a coach or player, what you should do. Coaching and player execution win games, and trying to figure out how to play using only numbers won't go well.

But they do have predictive power. They tell me, as an observer, what is likely to happen. They do very well on that front - better than the vast majority of "experts".

 

[pln2013]

Right...I don't think you're understanding what I'm saying. You see, there's this thing called the NCAA tournament that ends each season. The team that wins all their games is the champion. And whether Kenpom said that team would win or not is irrelevant. They're not handing out trophies according to kenpom. So what's the point?

The team that wins all their games is the champion. And whether your eye test said that team would win or not is irrelevant. They're not handing out trophies according to your eye test. So what's the point?

 

[gunslingerdick]

The team that wins all their games is the champion. And whether your eye test said that team would win or not is irrelevant. They're not handing out trophies according to your eye test. So what's the point?

touche'

Sort of. But I'm not making poasts here attempting to justify team X or team Y's ranking based on my eye test like you are with some statistical formula. In fact, I couldn't care less about rankings - especially in December. If you go back to my original poast in this thread, my last statement was "I guess it's something for some people to talk about but it's completely meaningless to me." Why do you give a shit what I choose to subscribe to and don't. Look, I'll leave you and the other number crunchers to pray at the alter of Kenpom and you leave me to fan however I like.

 

[Archer2]

I take all polls and basketball analysis sites for what they are, a tool for helping me to form my own opinions about how good teams. As Slinger said, they fail to take many subjective human points into account. Of course, I take anything I read on the internet with a grain of salt this big:

 

[gary-7]

If I come off that way, it's not my intent. I don't view Kenpom as "an omniscient authority". It's a helpful tool.

My opinion of UNC is mostly informed by my own eye test, expert analysis (including yours!), and more granular data (especially from Adrian Atkinson).

But what do I know about Texas? Basically nothing. I only watched them against Duke. Kenpom and other models help contextualize what I saw. Texas is a good-not-great team. Texas was a good win (and your resume is ultimately about wins and losses, not margin), but Duke barely beating them indicates Duke isn't as good as I might've thought.

What about Purdue? I haven't seen them once this year. Kenpom tells me they're pretty damn good. That's helpful, because otherwise I'm blind.



Agreed!

Predictive analytics don't have much prescriptive power. They don't tell you, as a coach or player, what you should do. Coaching and player execution win games, and trying to figure out how to play using only numbers won't go well.

But they do have predictive power. They tell me, as an observer, what is likely to happen. They do very well on that front - better than the vast majority of "experts".

Fair enough, with the exception of predictive power. What kind of tempo Team-A likes to run?... or how much Team-B relies on 3-pointers? Sure.
Outcomes of games? Not so much.

Admittedly, I probably watch too much college basketball for my own good --- "Feast Week" was a hoops orgy --- so my perspective is different from most fans, but I understand why stat sites get so much play. They are a fun short-cut to the landscape in a given point of a season and folks like to look at ESPN's indices, RPIs, BPIs, what-have-you.

When I look at an impending game it's much like boxing in that styles and match-ups matter. Ali was the GOAT, but a relentless pit-bull like Joe Frazier was always gonna be a challenge for a classy boxer. In contrast Frazier had almost zero chance going in against George Foreman because Joe's style was a setup for a big mega-puncher (and when the actual fight happened, yikes, it was ugly). However, a big ol' lumbering bear like George (or Sonny Liston before him) was a setup for Ali.

Now fortunately in basketball there are 10 guys on the court instead of 2 in the ring, but the point is that the nuances of human competition, coaching and officiating are more important in a given game than trends.