2022-23 KenPom Rankings | The Boneyard

2022-23 KenPom Rankings

The analytics here have us around where others have us, as a borderline top-25 team. It seems fair, although I agree with Hey Adrien that Ken Pom is a lot more valuable once games have started. Our best shot at cracking the top-15 or so is improvement from Hawkins, which a projection system like this would miss. This team can win the Big East.
 
.-.
It isn’t worth a new thread, but we are #34 at EvanMiya.com, and fourth in the Big East.

evanmiya.com

EvanMiya CBB Analytics

Advanced college basketball analytics, home of the Bayesian Performance Rating
evanmiya.com evanmiya.com

5CC1A286-6F98-4DB9-9461-D5F3C424C429.jpeg
 
Hey Adrien! said:
I love KenPom more than most here, but honestly, the website has little value in the current season until at least Christmas.
Expand ...
Does that explain creighton then who is a top 10 team in all polls but kenpom doesn’t even have in the top 20?
 
Jbirdp5 said:
Does that explain creighton then who is a top 10 team in all polls but kenpom doesn’t even have in the top 20?
Expand ...
Kinda. It’s just an algorithm. A very good one, though.

There are myriad reasons why, but for example, how do you quantify with math: freshmen; players expected to be in expanded roles; new coaching staffs…all this becomes clearer with actual data from the season.
 
.-.
Hey Adrien! said:
Kinda. It’s just an algorithm. A very good one, though.

There are myriad reasons why, but for example, how do you quantify with math: freshmen; players expected to be in expanded roles; new coaching staffs…all this becomes clearer with actual data from the season.
Expand ...
I don't know why he bothers this early, too much guesswork. Creighton is probably better than Nova.
 
Hey Adrien! said:
I love KenPom more than most here, but honestly, the website has little value in the current season until at least Christmas.
Expand ...

You are correct.

I run a python model via the kenpompy built-ins to capture divergences in the early season from vegas vs the stats. I'd have to have a look over my sports book but I was hitting 4+ leg parlays at a near 60% clip last year, until about the end of Nov / early December when there's enough data to support the betting lines.

Typically don't bet too much. Just enough to cover the cost of my kenpom subscription and streaming for the cbb season.

But there is value in it early on in the year.
 
Hey Adrien! said:
I love KenPom more than most here, but honestly, the website has little value in the current season until at least Christmas.
Expand ...
Yes and truthfully we don’t know what the hell we even have until then.
 
NYC_HUSKY22 said:
You are correct.

I run a python model via the kenpompy built-ins to capture divergences in the early season from vegas vs the stats. I'd have to have a look over my sports book but I was hitting 4+ leg parlays at a near 60% clip last year, until about the end of Nov / early December when there's enough data to support the betting lines.

Typically don't bet too much. Just enough to cover the cost of my kenpom subscription and streaming for the cbb season.

But there is value in it early on in the year.
Expand ...
Were you betting in KenPom's favor? The guy you're agreeing with was saying the opposite.
 
NYC_HUSKY22 said:
You are correct.

I run a python model via the kenpompy built-ins to capture divergences in the early season from vegas vs the stats. I'd have to have a look over my sports book but I was hitting 4+ leg parlays at a near 60% clip last year, until about the end of Nov / early December when there's enough data to support the betting lines.

Typically don't bet too much. Just enough to cover the cost of my kenpom subscription and streaming for the cbb season.

But there is value in it early on in the year.
Expand ...
Can we get in on this action?
 
NYC_HUSKY22 said:
You are correct.

I run a python model via the kenpompy built-ins to capture divergences in the early season from vegas vs the stats. I'd have to have a look over my sports book but I was hitting 4+ leg parlays at a near 60% clip last year, until about the end of Nov / early December when there's enough data to support the betting lines.

Typically don't bet too much. Just enough to cover the cost of my kenpom subscription and streaming for the cbb season.

But there is value in it early on in the year.
Expand ...
I remember when kenpom first came out I was using that along with other models and absolutely killing it on college bball over/unders. There would be some smaller conference games where the line would move 10 points cause a lot of handicappers were doing the same thing. Vegas got killed and adjusted accordingly but for a 3 month period it was a fantastic time
 
.-.
methodology said:
I remember when kenpom first came out I was using that along with other models and absolutely killing it on college bball over/unders. There would be some smaller conference games where the line would move 10 points cause a lot of handicappers were doing the same thing. Vegas got killed and adjusted accordingly but for a 3 month period it was a fantastic time
Expand ...
Post/handle
 
TheChamps990411 said:
Its not where you start its where you finish.
Expand ...
Sort of... KenPom's pre-season rankings factor into where you finish. The pre-season data eventually is pulled out completely from the modeling, but the reality is by the time that data is removed it's already had a massive influence on how every team is perceived for the entire season, who makes the tournament, etc...

RPI and NET are more "pure" rankings, where it isn't calculated pre-season and pre-season perceptions have no influence at all on the outcome.
 
KenPom gets better as the season goes but it fails to capture sophomore jumps, chemistry, some aspects of freshman ect. After 10 games or so it straightens out and starts to be meaningful but right now his model has no way to account for hawkins/jackson in their bigger roles, the freshman or any of the transfers chemistry/roles
 
PNolanFan said:
KenPom gets better as the season goes but it fails to capture sophomore jumps, chemistry, some aspects of freshman ect. After 10 games or so it straightens out and starts to be meaningful but right now his model has no way to account for hawkins/jackson in their bigger roles, the freshman or any of the transfers chemistry/roles
Expand ...
Is that not the arbitrage opportunity for a betting angle?

The stats can’t meaningfuly adjust for players who can make a big leap. Stud freshmen they can certainly put a number on, but a guy like Hawkins, how do you factor in his improvement accurately?

Fascinating discussion. Very volatile early in season as the baseline data turns over to up to date sampling.
 
.-.
JohnFSilver said:
Is that not the arbitrage opportunity for a betting angle?

The stats can’t meaningfuly adjust for players who can make a big leap. Stud freshmen they can certainly put a number on, but a guy like Hawkins, how do you factor in his improvement accurately?

Fascinating discussion. Very volatile early in season as the baseline data turns over to up to date sampling.
Expand ...
You can project certain players forward based on recruiting rankings. Essentially assign them stats based on historical comparison. 4* top 50 SG average soph stats when dealing with players with lower minutes totals from their freshman year (otherwise you just project average improvement by age from previous stats). I know Hoop-Explorer does this, not sure about KenPom. You can also project usage, which I know Torvik does, and make assumptions from there. (Ie if you're given a big role for a team that historically has a good offense but your stats or minutes were lower, you're probably better than expected). SI's former analyst Dan Hanner took this to an extreme with a comprehensive lineup model, but he's out of the industry (or hired by a team maybe). You can also feed into it consensus NBA Big Boards to supplement talent projection.
 
Balker said:
Sort of... KenPom's pre-season rankings factor into where you finish. The pre-season data eventually is pulled out completely from the modeling, but the reality is by the time that data is removed it's already had a massive influence on how every team is perceived for the entire season, who makes the tournament, etc...

RPI and NET are more "pure" rankings, where it isn't calculated pre-season and pre-season perceptions have no influence at all on the outcome.
Expand ...
We determine where we finish.
 
Balker said:
Sort of... KenPom's pre-season rankings factor into where you finish. The pre-season data eventually is pulled out completely from the modeling, but the reality is by the time that data is removed it's already had a massive influence on how every team is perceived for the entire season, who makes the tournament, etc...

RPI and NET are more "pure" rankings, where it isn't calculated pre-season and pre-season perceptions have no influence at all on the outcome.
Expand ...

This is the one bone i would pick with Ken (and have tried). He should publicize his "curve" using prior / non-game data, and possibly also give a "pure this-season" number along side it.
 
You must log in or register to reply here.

Forum statistics

Threads
168,285
Messages
4,561,420
Members
10,455
Latest member
UConnGabby


Top Bottom