Kenpom has us 12th in the ACC

[Cusefan0307]

And 68th overall.

This would be our worst Kenpom ranking ever going back to 2002 when the formula started.

 

[OttoinGrotto]

What on Earth is this projection based on? Tyus Battle's stats? There's no other data to work with.

 

[rrlbees]

Good. We have them right where we want 'em.

 

[cuseinseattle]

And 68th overall.

This would be our worst Kenpom ranking ever going back to 2002 when the formula started.

He also has us above Yukon. Just sayin.

 

[moqui]

What on Earth is this projection based on? Tyus Battle's stats? There's no other data to work with.

it's not a prediction; he begins with last season's final numbers then subtracts players who are not returning. it is just his baseline measure to begin the season

 

[OttoinGrotto]

it's not a prediction; he begins with last season's final numbers then subtracts players who are not returning. it is just his baseline measure to begin the season

Ok.

 

[cuseattle]

Here's my thing with Kenpom.
I know it's pretty accurate and good, but at some point, no matter how well the poll is set up analytically, a human still has to put in the initial rankings.

 

[jordoo]

Here's my thing with Kenpom.
I know it's pretty accurate and good, but at some point, no matter how well the poll is set up analytically, a human still has to put in the initial rankings.

I get you on human error and such. In the end though this is just a dude making money using math and his interest in basketball. It's a pretty good predictor/tracker overall especially if you are aware of how he calculated stuff.

Funny thing about humans . . . . The brain is tech that far exceeds anything else known.

 

[Knicks411]

I also think there's a rough formula to account for freshmen as well.

2016 preseason ratings | The kenpom.com blog

 

[Fly Rodder]

What on Earth is this projection based on? Tyus Battle's stats? There's no other data to work with.

I don't even pretend to know how his algorithms work or what exactly goes into them, but there's plenty of data to work with. Freshmen play every season. They don't need to how a specific freshman is going to play to make projections, just maybe how an average freshman for the position ranked as a certain recruit level has historically played (e.g., freshman who score more than 15 pts a game are rare. Freshman SFs who are ranked outside the top-20 and score 15 pts a game are even more rare., etc.). So with some finagling through an algorithm, they can generate a potential stat line based on historical averages.

Outliers happen, but these are general projections built around the "most likely scenario" - not every scenario. A baseball site I frequent does projections every preseason and presents 5 intervals of projections for each player (20%, 35%, baseline, 65% 80%). The baselines are aggregated for each team and then those are run through 10,000 "seasons". So, for example, this season Aaron Judge was projected to have a baseline of 25 HRs, and then "that player" was used for 10,000 simulated seasons. He never reached 50 HRs in 10,000 seasons. What he did this season was an absolute outlier based on what data was available, but that doesn't mean it was impossible just really unexpected. It ain't perfect, but these projections aren't predictions and not intended to be so.

Fangraphs has an explanation here.

 

[Cusefan0307]

 

[Digger]

I also think there's a rough formula to account for freshmen as well.

2016 preseason ratings | The kenpom.com blog

From the link:

It doesn’t know about transfers, and all but 5-star recruits are virtually ignored. If you think your favorite team is ranked too low, the reason is probably that there are really good transfers or recruits arriving.

Projected offense is largely determined by the quality of a team’s offense over the previous three seasons and its defense from last season.

I have no clue if we're ranked too low, but if we are I assume it's in large part because of this

 

[Ringostar57]

Kenporn...

 

[moqui]

his baseline defensive efficiency for the orange is 97.2, a 5.1 point improvement over last year. based entirely on the guys who left. #additionbysubtraction

btw, last year's defensive efficiency was 102.3, nearly 10 full points higher than the 92.4 average of the previous 8 seasons. i believe that we have the personnel to get back around that average. and in the minutes when sidibe is on the floor, we'll beat that average

 

[Cusefan0307]

his baseline defensive efficiency for the orange is 97.2, a 5.1 point improvement over last year. based entirely on the guys who left. #additionbysubtraction

btw, last year's defensive efficiency was 102.3, nearly 10 full points higher than the 92.4 average of the previous 8 seasons. i believe that we have the personnel to get back around that average. and in the minutes when sidibe is on the floor, we'll beat that average

No doubt, but the guys who left were all better offensive players than who we added. We need to get back to an elite level defensively if we are going to win 20 games because we won’t be as good at the offensive end.

 

[moqui]

No doubt, but the guys who left were all better offensive players than who we added. We need to get back to an elite level defensively if we are going to win 20 games because we won’t be as good at the offensive end.

that remains to be seen

 

[pearl31]

From the link:

It doesn’t know about transfers, and all but 5-star recruits are virtually ignored. If you think your favorite team is ranked too low, the reason is probably that there are really good transfers or recruits arriving.

Projected offense is largely determined by the quality of a team’s offense over the previous three seasons and its defense from last season.

I have no clue if we're ranked too low, but if we are I assume it's in large part because of this

good find

 

[Brooky03]

Sometimes I feel like I'm the only one who isn't in love with KenPom's rankings. Love his advanced stats, but not his rankings.

I don't think his algorithms produce any more reliable rankings than the average 'expert'. I think his methods would work better in professional leagues with fewer teams and more head-to-head games between everybody.

 

[Cusefan0307]

Sometimes I feel like I'm the only one who isn't in love with KenPom's rankings. Love his advanced stats, but not his rankings.

I don't think his algorithms produce any more reliable rankings than the average 'expert'. I think his methods would work better in professional leagues with fewer teams and more head-to-head games between everybody.

It's not perfect, but I would rather the committee rely on a metric like this than something like the eye test and the AP poll/RPI when it comes to tournament seeding which are heavily weighted with confirmation bias and Duke is given one or two seeds every year even if they have 7-9 losses.

 

[Knicks411]

It's not perfect, but I would rather the committee rely on a metric like this than something like the eye test and the AP poll/RPI when it comes to tournament seeding which are heavily weighted with confirmation bias and Duke is given one or two seeds every year even if they have 7-9 losses.

I make this point a lot, but the projected scores they come up with are almost always within a point of the Vegas line, which makes me think they're pretty good.

 

[upperdeck]

you can play better D and worse offense but create more points.. if we get 2-3 more blocked shots and 2-3 steals/deflects a game that lead to 4-5 break outs you can flip a bad shooting team to decent pretty quickly . even being a better rebounding team would add pts and who knows how we will be there.. shooting lots of 3s and getting no rebounds wasnt a recipe that worked for us.

 

[Brooky03]

It's not perfect, but I would rather the committee rely on a metric like this than something like the eye test and the AP poll/RPI when it comes to tournament seeding which are heavily weighted with confirmation bias and Duke is given one or two seeds every year even if they have 7-9 losses.

I would like the committee to include more advanced stats in the decision making process; I think we agree there. What I wouldn't like is the committee narrowing it down to Team A and Team B, looking at their Kenpom rankings, and just going with the higher/better ranked team. His rankings aren't refined enough for that, imo.

 

[Cusefan0307]

you can play better D and worse offense but create more points.. if we get 2-3 more blocked shots and 2-3 steals/deflects a game that lead to 4-5 break outs you can flip a bad shooting team to decent pretty quickly . even being a better rebounding team would add pts and who knows how we will be there.. shooting lots of 3s and getting no rebounds wasnt a recipe that worked for us.

When’s the last time we rebounded well?

 

[Knicks411]

When’s the last time we rebounded well?

Offensively? Always. Defensively? Never

I would like the committee to include more advanced stats in the decision making process; I think we agree there. What I wouldn't like is the committee narrowing it down to Team A and Team B, looking at their Kenpom rankings, and just going with the higher/better ranked team. His rankings aren't refined enough for that, imo.

For sure, I hope no one is suggesting that. If Team A is #40 and team B is #75, that's one thing, but if the difference is like 10 spots, that's not really much.

 

[HOFCeluck]

Revoke Kenpom's voting privleges!