NET and KenPom Tracker 23-24 (SU = 84 3/9/24)

[OCESIP]

Just wondering. What is the "luck" factor in the Ken Pom formula.
Oregon State is highest, Navy is lowest. A curious element to
factor into a computerized tool.

 

[Fly Rodder]

Just wondering. What is the "luck" factor in the Ken Pom formula.
Oregon State is highest, Navy is lowest. A curious element to
factor into a computerized tool.

https://kenpom.com/blog/ratings-explanation/

What do all the columns mean?
The new ones are Cons (Consistency) and Luck. The easiest one to understand is Luck, which is the deviation in winning percentage between a team’s actual record and their expected record using the correlated gaussian method. The luck factor has nothing to do with the rating calculation, but a team that is very lucky (positive numbers) will tend to be rated lower by my system than their record would suggest.

 

[blust2i3d4]

We definitely need some sort of wins against Duke, UNC, Miami, Clemson (already lost opportunity on Virginia) and a Pitt win isn't helping nearly as much, but winning or losing the Pitt game next week will be a good indicator for the fans as to whether this team will find it in themselves to climb to the top of the middle of the pack in the ACC and compete for a tournament bid. Bottom of the ACC is very bad. I think the middle is better than the last two years, but not what it should be.

Top four ACC teams are UNC, Duke, Virginia and Clemson. I don't think Miami is good.

 

[CardiacCuse]

Just wondering. What is the "luck" factor in the Ken Pom formula.
Oregon State is highest, Navy is lowest. A curious element to
factor into a computerized tool.

"what should a team's record be" vs. "what a team's record is"

IF they "should" have a lower record than they do...that team gets dinged by his system

I guess the question becomes: how does he know what a team's record should be?

 

[jordoo]

"what should a team's record be" vs. "what a team's record is"

IF they "should" have a lower record than they do...that team gets dinged by his system

I guess the question becomes: how does he know what a team's record should be?

It seems like an attempt to explain a formula/algorithm that doesn't accurately predict what it was designed to. The same person who came up with the original formula/algorithm is also responsible for the LUCK . . . . . . . . . . . LOL

 

[jdubs30]

Would love Oregon to be a quality neutral court win, but it seems like that might not be the case by seasons end

 

[Knicks411]

The luck thing, as far as I can tell, is just looking at teams who have won/lost more games than expected based on their point differential It's pretty standard across sports analysis, you see Pythagorean record in most sports.

 

[Fly Rodder]

It seems like an attempt to explain a formula/algorithm that doesn't accurately predict what it was designed to. The same person who came up with the original formula/algorithm is also responsible for the LUCK . . . . . . . . . . . LOL

Seems like people could make an attempt at understanding how this actually works instead of just being curmudgeons who don't trust data analysis because they don't understand how any of it works. And if they don't understand it, then it doesn't make sense. Nice little closed loop system you got there.

 

[CardiacCuse]

The luck thing, as far as I can tell, is just looking at teams who have won/lost more games than expected based on their point differential It's pretty standard across sports analysis, you see Pythagorean record in most sports.

the problem with this, for me, is playing style changes based on margin for many teams

i.e. relax with huge leads, play conservative with small leads etc

so assuming margin of victory is some unknown constant that we can measure results against ignores the way basketball teams play games.

there might be certain teams that have erratic margins of results but not due to luck rather do to their personalities or strategies...

luck would be a total misnomer for such teams because increased "luck" would thus be a result of inherent qualitites of those teams/coaches

 

[Cheriehoop]

Did we gain 16 points by beating Oregon while Oregon decreased 22 points losing to us?

 

[Knicks411]

the problem with this, for me, is playing style changes based on margin for many teams

i.e. relax with huge leads, play conservative with small leads etc

so assuming margin of victory is some unknown constant that we can measure results against ignores the way basketball teams play games.

there might be certain teams that have erratic margins of results but not due to luck rather do to their personalities or strategies...

luck would be a total misnomer for such teams because increased "luck" would thus be a result of inherent qualitites of those teams/coaches

Sure I get that is the case for some teams in some cases.

On the whole, scoring margin is generally a good predictor moving forward as to quality of teams, so it seems to make sense consider that when trying to evaluate the quality of the team. It's not a one size fits all type deal, but when you're trying to rank 350 teams or whatever, I think it does a decent enough job. Do some teams consistently outperform their "luck" number? Or is it relatively random from year to year? (Hopefully someone has done the research).

For whatever this is worth (probably nothing) I took a look at the 5 "luckiest" teams per this number in 2023. Here is how they ranked the season prior

1) UNC Asheville (luckiest in 23, 260th in 22)
2) Morehead State- 126th in 22
3) UNC Wilmington- 3rd
4) Alcorn State- 4th
5) Kennesaw State- 353rd


So we saw 2 teams that were "lucky" in back to back years (Then i had to go back and check 21, Alcorn state dropped to 241st in 21 and Wilmington 350th) but the other 3 teams were all over the place the year prior. Which says to me, based of course on this impossibly small sample, is that there isn't much correlation from year to year with this kind of thing so it probably isn't very predictive. (That said UNC Wilmington is top 5 in luck this year again, so maybe thats actually a thing? Alcon state is 248th)

 

[CardiacCuse]

Sure I get that is the case for some teams in some cases.

On the whole, scoring margin is generally a good predictor moving forward as to quality of teams, so it seems to make sense consider that when trying to evaluate the quality of the team. It's not a one size fits all type deal, but when you're trying to rank 350 teams or whatever, I think it does a decent enough job. Do some teams consistently outperform their "luck" number? Or is it relatively random from year to year? (Hopefully someone has done the research).

For whatever this is worth (probably nothing) I took a look at the 5 "luckiest" teams per this number in 2023. Here is how they ranked the season prior

1) UNC Asheville (luckiest in 23, 260th in 22)
2) Morehead State- 126th in 22
3) UNC Wilmington- 3rd
4) Alcorn State- 4th
5) Kennesaw State- 353rd


So we saw 2 teams that were "lucky" in back to back years (Then i had to go back and check 21, Alcorn state dropped to 241st in 21 and Wilmington 350th) but the other 3 teams were all over the place the year prior. Which says to me, based of course on this impossibly small sample, is that there isn't much correlation from year to year with this kind of thing so it probably isn't very predictive. (That said UNC Wilmington is top 5 in luck this year again, so maybe thats actually a thing? Alcon state is 248th)

i mean, thats a lot of overlap there...luck should be random...there shouldnt be teams in the top 5 back to back seasons

margin of victory is indicative but not precisely...rather they just use that

but then teams play to WIN they dont usually play to win by as much as they can

some teams dnt want to run up the score even if they could, for example

in Syr recent game vs Niagara, they start imitating the globetrotters late in the game bc they felt comfortable, for example...

basically, the only way luck would be a good indicator is if game-state didnt impact performance but it clearly does ...moreso for some teams than others

 

[Cheriehoop]

i mean, thats a lot of overlap there...luck should be random...there shouldnt be teams in the top 5 back to back seasons

margin of victory is indicative but not precisely...rather they just use that

but then teams play to WIN they dont usually play to win by as much as they can

some teams dnt want to run up the score even if they could, for example

in Syr recent game vs Niagara, they start imitating the globetrotters late in the game bc they felt comfortable, for example...

basically, the only way luck would be a good indicator is if game-state didnt impact performance but it clearly does ...moreso for some teams than others

JB was notorious for not running up the score. He counted possessions and would slow down the offense towards the end of games to use minimizing the opponent’s possessions to ensure wins vs wanting to possibly running up the score more. In blowouts he’d take off the pedal in the last minute or so too.

 

[Knicks411]

i mean, thats a lot of overlap there...luck should be random...there shouldnt be teams in the top 5 back to back seasons

margin of victory is indicative but not precisely...rather they just use that

but then teams play to WIN they dont usually play to win by as much as they can

some teams dnt want to run up the score even if they could, for example

in Syr recent game vs Niagara, they start imitating the globetrotters late in the game bc they felt comfortable, for example...

basically, the only way luck would be a good indicator is if game-state didnt impact performance but it clearly does ...moreso for some teams than others

Sure I dont think anyone is arguing MOV is precisely correct or whatever. It's just another data point that in the aggregate adds value.

 

[Sherman20]

https://x.com/madeformarch/status/1738742322571973015?s=46&t=RtEJIchZ5r8ArjkidyJFyQ

This team (Iowa State, I believe) is now 5th in the NET. Check out their stellar resume. They’ve beaten up on a bunch of tomato cans and went 1-2 in their games against P5 teams, losing to Va Tech and Texas A&M and beating a weak Iowa team. Unbelievable.

Also, 7-5 Alabama is somehow 9th in the NET.

 

[CardiacCuse]

https://x.com/madeformarch/status/1738742322571973015?s=46&t=RtEJIchZ5r8ArjkidyJFyQ

This team (Iowa State, I believe) is now 5th in the NET. Check out their stellar resume. They’ve beaten up on a bunch of tomato cans and went 1-2 in their games against P5 teams, losing to Va Tech and Texas A&M and beating a weak Iowa team. Unbelievable.

Also, 7-5 Alabama is somehow 9th in the NET.

hopefully it will balance out by the end of the season, but yeah, thus far, none of these ranking systems seem very accurate

for example, the disparity between NET, Kenpom and RPI is generally extreme...(and in my opinion, points to the arbitrariness of the ranking systems)

Syr NET (82), KP (86) RPI (16)
Duke NET (18) KP (7) RPI (52)
UVA NET (36) KP (37) RPI (22)
UNC NET (26) KP (15) RPI (20)

unfortunately for SU...the one the NCAA is using at the moment, is also the one that ranks them poorly...

 

[Capt. Tuttle]

https://kenpom.com/blog/ratings-explanation/

Seems to me he ought to re-evaluate his initial formula for expected records if he needs to add a luck component to counter the actual results.
For instance, without looking it up, I would bet the 2009-2010 SU team was one of the luckiest ever.

I will let the math/stats folks let me know why I am not correct.

 

[atlanticuse]

Seems to me he ought to re-evaluate his initial formula for expected records if he needs to add a luck component to counter the actual results.
For instance, without looking it up, I would bet the 2009-2010 SU team was one of the luckiest ever.

I will let the math/stats folks let me know why I am not correct.

I think 2013-14 was lucky - Pitt game one shining example.

 

[blust2i3d4]

hopefully it will balance out by the end of the season, but yeah, thus far, none of these ranking systems seem very accurate

for example, the disparity between NET, Kenpom and RPI is generally extreme...(and in my opinion, points to the arbitrariness of the ranking systems)

Syr NET (82), KP (86) RPI (16)
Duke NET (18) KP (7) RPI (52)
UVA NET (36) KP (37) RPI (22)
UNC NET (26) KP (15) RPI (20)

unfortunately for SU...the one the NCAA is using at the moment, is also the one that ranks them poorly...

I think the problem is that these computer aided ratings programs also use the data in the past two years especially how teams in NCAA performed in NCAA tournament. Not in NCAA tournament makes us outside 64 until one year we make it.

 

[orange79]

I think the problem is that these computer aided ratings programs also use the data in the past two years especially how teams in NCAA performed in NCAA tournament. Not in NCAA tournament makes us outside 64 until one year we make it.

Unless I'm misunderstanding you, there is no way data from previous seasons should be used in any other season. Each season is a stand-alone event.

 

[bnoro]

Data is most accurate with a larger sample size.

So looking at everything now is premature, and premature anything isn't good.

 

[CardiacCuse]

I think the problem is that these computer aided ratings programs also use the data in the past two years especially how teams in NCAA performed in NCAA tournament. Not in NCAA tournament makes us outside 64 until one year we make it.

they use data from other seasons?!?!

that would be really wierd...especially in the portal era

 

[Cusefan95]

the problem with this, for me, is playing style changes based on margin for many teams

i.e. relax with huge leads, play conservative with small leads etc

so assuming margin of victory is some unknown constant that we can measure results against ignores the way basketball teams play games.

there might be certain teams that have erratic margins of results but not due to luck rather do to their personalities or strategies...

luck would be a total misnomer for such teams because increased "luck" would thus be a result of inherent qualitites of those teams/coaches

I don’t think how you view this is accurate - although I don’t know for sure since I don’t know the algorithm.

Margin of victory shouldn’t be a linear correlation, it’s probably some form of power graph. The “value difference” between winning by 1 vs 2 pts is much greater than winning by 8 vs 9 points. A significant amount of the issue with a team taking their foot off the gas doesn’t really matter as a result. But there’s always going to be potential for some weird potential outliers.

Plus I think he used margin of victory to simplify the explanation - which actually isn’t helpful. The whole point of this analysis is on offensive and defensive efficiency, so that’s what would most likely be used to calculate “luck”. Those should both be impacted less than pure margin of victory by a strategy of running the clock out at the end of a game.

As an aside, when hockey analytics were initially taking off, a lot of people mocked the measure of “puck luck”. Then the Kings (#8 seed) and Devils (#6 seed) made the 2012 Stanley Cup Finals as the two worst “puck luck” teams - then everyone thought it was some magic bullet measure (which it isn’t). It might be better if luck was called something else like potential variance, although nobody would know what that meant and it’s not really accurate either. The whole issue here is that most people don’t understand statistical analysis, and efforts to simply enough to explain it generally just confuse the situation for the average person. I know enough to know I don’t know enough to know if how luck is calculated here has any predictive value, and I also know that investigating it to figure that out isn’t a valuable way for me to spend my time.

 

[Ragman2000]

Blind Resume, guess the Net ranking

Q1 0-2

Q2 0-0

Q3 4-1

Q4 5-0

Vs

Q1 0-3

Q2 1-0

Q3 6-0

Q4 1-0

 

[Cuse181]

Blind Resume:

Q1 0-2

Q2 0-0

Q3 4-1

Q4 5-0

Vs

Q1 0-3

Q2 1-0

Q3 6-0

Q4 1-0

Eh, just win the games in front of them. The 10+ point Blowout losses don’t help