the Acc...

ecojew

All-Conference
Feb 1, 2006
9,767
2,271
0
...has just one team in this weeks Coaches poll, that being Duke. Only two others received votes - Miami and UNC, with the latter receiving just two votes. That is the worst showing I can ever remember.

Meanwhile, the B1G and B12 have 5 each, the SEC and BE four each, and the Pac12 has 3. The WCC, AAC, and MVC are tied with acc.
 

goru7

All-American
Dec 12, 2005
6,432
7,710
113
...has just one team in this weeks Coaches poll, that being Duke. Only two others received votes - Miami and UNC, with the latter receiving just two votes. That is the worst showing I can ever remember.

Meanwhile, the B1G and B12 have 5 each, the SEC and BE four each, and the Pac12 has 3. The WCC, AAC, and MVC are tied with acc.
Yeah , Acc is at its lowest level in like 15 years. However, Big 12 and SEC are so clearly ahead of the rest of the conferences from computers and eye test. Kansas State the # 10 team in the Big 12 beat Texas Tech who just beat Baylor and Kansas. Our Nebraska is not beating Purdue . Similarly top 10-11 teams in SEC can beat top teams in auburn and Kentucky and LSU.
Clearly , Big 10 is down from the last 2-3 years where we were clearly the best and the deepest. Not so much this year. We should be able to take advantage by winning our home games and stealing a couple of more on the road.
 
  • Like
Reactions: Antnee79 and ecojew

fluoxetine

Heisman
Nov 11, 2012
23,529
16,898
0
However, Big 12 and SEC are so clearly ahead of the rest of the conferences from computers and eye test. Kansas State the # 10 team in the Big 12 beat Texas Tech who just beat Baylor and Kansas. Our Nebraska is not beating Purdue . Similarly top 10-11 teams in SEC can beat top teams in auburn and Kentucky and LSU.
Clearly , Big 10 is down from the last 2-3 years where we were clearly the best and the deepest. Not so much this year. We should be able to take advantage by winning our home games and stealing a couple of more on the road.
What?

Big Ten is second best conference in every major computer ranking with SEC third.
 

goru7

All-American
Dec 12, 2005
6,432
7,710
113
What?

Big Ten is second best conference in every major computer ranking with SEC third.
Really? Michigan is still ranked high because of last year . Shocked Michigan at 32 in Kenpom at 7-7 and Iowa at 19 and Indiana 27 are ridiculous rankings with some of last year factored in. SEC has Auburn, LSU , Kentucky , Alabama , Tennessee , Miss. State, Arkansas . Florida , Vanderbilt , Ole Miss and South Carolina as their top 11 which are much stronger than Purdue, Illinois , Mich State , Wisconsin , Ohio State , Rutgers ,Indiana, Iowa , Minnesota , Penn State and Northwestern / Michigan our top 11/12. . Eye test for sure says SEC way up and Big 10 down and maybe out of conference strength of schedule for SEC schools must be pathetic for computers to have SEC behind BIG 10.
 

fluoxetine

Heisman
Nov 11, 2012
23,529
16,898
0
Really? Michigan is still ranked high because of last year . Shocked Michigan at 32 in Kenpom at 7-7 and Iowa at 19 and Indiana 27 are ridiculous rankings with some of last year factored in. SEC has Auburn, LSU , Kentucky , Alabama , Tennessee , Miss. State, Arkansas . Florida , Vanderbilt , Ole Miss and South Carolina as their top 11 which are much stronger than Purdue, Illinois , Mich State , Wisconsin , Ohio State , Rutgers ,Indiana, Iowa , Minnesota , Penn State and Northwestern / Michigan our top 11/12. . Eye test for sure says SEC way up and Big 10 down and maybe out of conference strength of schedule for SEC schools must be pathetic for computers to have SEC behind BIG 10.
I don't think there is much if any influence from last year still in those rankings.
 

fluoxetine

Heisman
Nov 11, 2012
23,529
16,898
0
The top of the SEC is better than the top of the Big 10, but the middle and bottom are worse.

South Carolina and Ole Miss are worse than every team in the Big Ten except Nebraska. Vanderbilt isn't any good either. Texas A&M is a complete mirage, 15-2 against the 274th ranked schedule.
 
  • Like
Reactions: shields

fluoxetine

Heisman
Nov 11, 2012
23,529
16,898
0
The have the 7th ranked schedule with an adj EM of 9.31 and have outscored those opponents by 4.5 pts/gm which is roughly 6.8 pts/100 possessions.

So just naive math there I would estimate their kenpom rating at +16.11. Their actual rating is +16.84. I don't know if the difference is do to preseason influence or just the fact that I am doing a very rough estimate but it's not way off.

16.84 puts them at #32, if they were 16.11 it would be #37.

Also Bart has them #39 and if I put on filters to remove any preseason influence they are at #42. So I think you are right that there is a little influence in there but not much, and it should all be gone soon.
 

goru7

All-American
Dec 12, 2005
6,432
7,710
113
The have the 7th ranked schedule with an adj EM of 9.31 and have outscored those opponents by 4.5 pts/gm which is roughly 6.8 pts/100 possessions.

So just naive math there I would estimate their kenpom rating at +16.11. Their actual rating is +16.84. I don't know if the difference is do to preseason influence or just the fact that I am doing a very rough estimate but it's not way off.

16.84 puts them at #32, if they were 16.11 it would be #37.

Also Bart has them #39 and if I put on filters to remove any preseason influence they are at #42. So I think you are right that there is a little influence in there but not much, and it should all be gone soon.
7th ranked schedule is ********. They have literally beaten 1 team with a pulse, San Diego State and Beat Tarleton by 11, UNLV by 12 and Praire View by almost 30 at home. They got pummeled by Arizona , UNC , lost by 10 at home to Minnesota , lost by 14 to UCF and by 8 to us on the road. Their only decent game was a 2 point loss at Seton Hall . How is that the 7th ranked schedule ? Someone is wrong with the computer model and their weighing of inputs
 
  • Like
Reactions: Local Shill

fluoxetine

Heisman
Nov 11, 2012
23,529
16,898
0
7th ranked schedule is ********. They have literally beaten 1 team with a pulse, San Diego State and Beat Tarleton by 11, UNLV by 12 and Praire View by almost 30 at home. They got pummeled by Arizona , UNC , lost by 10 at home to Minnesota , lost by 14 to UCF and by 8 to us on the road. Their only decent game was a 2 point loss at Seton Hall . How is that the 7th ranked schedule ? Someone is wrong with the computer model and their weighing of inputs
No. Michigan SOS, according to:

Massey: 4th
Sagarin: 7th
Kenpom: 6th
Bart: 8th

Have you considered that computers may be better suited to analyzing large datasets of basketball results than your brain is?
 

goru7

All-American
Dec 12, 2005
6,432
7,710
113
No. Michigan SOS, according to:

Massey: 4th
Sagarin: 7th
Kenpom: 6th
Bart: 8th

Have you considered that computers may be better suited to analyzing large datasets of basketball results than your brain is?
Genius, I am saying that the input into the computer is flawed. Can your brain understand that ? No way in hell Michigan has the 7th hardest schedule in the country. Get it.
 
  • Like
Reactions: Local Shill

fluoxetine

Heisman
Nov 11, 2012
23,529
16,898
0
Genius, I am saying that the input into the computer is flawed. Can your brain understand that ? No way in hell Michigan has the 7th hardest schedule in the country. Get it.
What do you think the input is? In what way is it flawed? You're just some guy looking at Michigan's schedule and going "hurr durr that's not the 7th hardest schedule OBVIOUSLY". Meanwhile every reasonable computer model comes up with basically the same answer and yet you think you're smarter than all of them.
 

fluoxetine

Heisman
Nov 11, 2012
23,529
16,898
0
You inspired me to get my own computer ratings to spit out basic SOS rankings since they did not until now.

Michigan's SOS was... drumroll... 7th

And I also hacked together conference rankings which I did not have in the past either.

Drumroll #2

Big 12
1935.223​
Big East
1830.395​
Big 10
1826.285​
Southeastern
1814.415​
Pac 12
1730.719​
Atlantic Coast
1686.318​
Mountain West
1678.611​
American Athletic
1650.455​
West Coast
1633.842​
Atlantic 10
1562.237​
Missouri Val
1495.481​
Conference USA
1477.154​
Southern
1467.251​
Sun Belt
1433.914​
Colonial
1423.607​
Western Athletic
1415.051​
Metro Atlantic
1386.82​
Mid-American
1384.587​
Big West
1380.419​
Ivy League
1343.499​
Summit Lg
1320.414​
Atlantic Sun
1316.633​
Big Sky
1296.48​
OH Valley
1292.7​
Big South
1249.175​
America East
1220.192​
Horizon
1216.358​
Patriot League
1191.514​
Northeast
1159.949​
Southland
1148.964​
Southwestern AC
1118.252​
Mid-Eastern AC
1111.858​


P.S. I know for certain this does not contain pre-season influence because it is my own code.
 
  • Like
Reactions: S.W.A.I.N

goru7

All-American
Dec 12, 2005
6,432
7,710
113
You inspired me to get my own computer ratings to spit out basic SOS rankings since they did not until now.

Michigan's SOS was... drumroll... 7th

And I also hacked together conference rankings which I did not have in the past either.

Drumroll #2

Big 12
1935.223​
Big East
1830.395​
Big 10
1826.285​
Southeastern
1814.415​
Pac 12
1730.719​
Atlantic Coast
1686.318​
Mountain West
1678.611​
American Athletic
1650.455​
West Coast
1633.842​
Atlantic 10
1562.237​
Missouri Val
1495.481​
Conference USA
1477.154​
Southern
1467.251​
Sun Belt
1433.914​
Colonial
1423.607​
Western Athletic
1415.051​
Metro Atlantic
1386.82​
Mid-American
1384.587​
Big West
1380.419​
Ivy League
1343.499​
Summit Lg
1320.414​
Atlantic Sun
1316.633​
Big Sky
1296.48​
OH Valley
1292.7​
Big South
1249.175​
America East
1220.192​
Horizon
1216.358​
Patriot League
1191.514​
Northeast
1159.949​
Southland
1148.964​
Southwestern AC
1118.252​
Mid-Eastern AC
1111.858​


P.S. I know for certain this does not contain pre-season influence because it is my own code.
Ok. Smart ***. Lay out Michigan’s Schedule so far.
Buffalo(H);
Prairie View(H)
Seton Hall( A)
UnLV (N)
Arizona (N)
Tarleton(H)
UNC (A)
San Diego State(H)
Nebraska (A )
Minnesota ( H)
Southern Utah ( H)
UCF (A)
Rutgers( A)
Illinois (A).
They are 7-7 with their wins , Buffalo , Prairie View, UNLV , San Diego State , Tarlton, Southern Utah and Nebraska.
They had one game against a top 10 team , on a neutral court, Arizona , that they got obliterated. They have lost against anyone else with a pulse and by more than double digits to all but Seton Hall. They have no good wins. Please explain how they are 32 in Kenpom . Based on what. Offensive efficiency , defensive efficiency , strength of schedule . Do they not get punished for being blown out by Arizona and UNC and non competitive losses to UCF and Rutgers who are 76 and 102 in Kenpom and a 10 point home loss to # 91 Minnesota . Why are they not 75-100 in Kenpom? Any computer model that says their SOS is number 7 and their rank is 32 is flawed and should be discounted.
 

fluoxetine

Heisman
Nov 11, 2012
23,529
16,898
0
Ok. Smart ***. Lay out Michigan’s Schedule so far.
Buffalo(H);
Prairie View(H)
Seton Hall( A)
UnLV (N)
Arizona (N)
Tarleton(H)
UNC (A)
San Diego State(H)
Nebraska (A )
Minnesota ( H)
Southern Utah ( H)
UCF (A)
Rutgers( A)
Illinois (A).
I don't know what to tell you. I will admit that I looked at that schedule and was also surprised it was ranked so high. But you can't RANK SCHEDULES RELATIVE TO EACH OTHER by looking at only one team's schedule. I've literally calculated it myself, using my own rankings. I'm sure you would disagree with my rankings in some places but they are going to be at least in the right ballpark and changing them slightly would not significantly impact the results.

OpponentHome/AwayOpponent RankingOpponent RatingHome/Away AdjustmentAdjusted Opponent Rating
BuffaloHome
145​
0.39​
-2.8​
-2.41​
Prairie ViewNeutral
311​
-14.29​
0​
-14.29​
Seton HallHome
31​
15.14​
-2.8​
12.34​
UNLVAway
110​
4.26​
2.8​
7.06​
ArizonaNeutral
3​
24.84​
0​
24.84​
Tarleton StHome
187​
-2.89​
-2.8​
-5.69​
North CarolinaAway
43​
13.51​
2.8​
16.31​
San Diego StHome
38​
14.16​
-2.8​
11.36​
NebraskaAway
180​
-2.36​
2.8​
0.44​
MinnesotaHome
56​
10.72​
-2.8​
7.92​
Southern UtahHome
151​
-0.09​
-2.8​
-2.89​
UCFAway
84​
7.36​
2.8​
10.16​
RutgersAway
104​
4.68​
2.8​
7.48​
IllinoisAway
20​
17.30​
2.8​
20.10​
Average
6.62

If you do this for every team you will find that Michigan has the 7th highest average adjusted opponent rating. And I am confident this answer is reasonable because a bunch of other computer ratings that all use different methodologies than mine (and each other) all came up with similar answers.

They are 7-7 with their wins , Buffalo , Prairie View, UNLV , San Diego State , Tarlton, Southern Utah and Nebraska.
They had one game against a top 10 team , on a neutral court, Arizona , that they got obliterated. They have lost against anyone else with a pulse and by more than double digits to all but Seton Hall. They have no good wins. Please explain how they are 32 in Kenpom .
Because they have the 32nd highest adjusted efficiency margin. It's literally a mathematical formula, not voodoo magic or Mr. Pomeroy staring at the teams trying to figure out which one is #31 and which one is #33. It's just math.

Based on what. Offensive efficiency , defensive efficiency , strength of schedule .
Yes.

Do they not get punished for being blown out by Arizona and UNC and non competitive losses to UCF and Rutgers who are 76 and 102 in Kenpom and a 10 point home loss to # 91 Minnesota . Why are they not 75-100 in Kenpom?
For the same reason that 2+2=4. You just do the math. You are trying to out-logic a mathematical formula and it is why you fail.

Any computer model that says their SOS is number 7 and their rank is 32 is flawed and should be discounted.
No. You're just a guy saying things. Trying to analyze a massive dataset with your brain is the domain of idiots.

FWIW my rankings have them #81 because there is a bonus/penalty for winning and losing. Kenpom does not measure wins and losses. it does not care about wins and losses. It's not a resume rank, it's an adjusted efficiency margin. You don't even understand the things you are criticizing, so why should anyone take your criticism seriously?
 

goru7

All-American
Dec 12, 2005
6,432
7,710
113
I don't know what to tell you. I will admit that I looked at that schedule and was also surprised it was ranked so high. But you can't RANK SCHEDULES RELATIVE TO EACH OTHER by looking at only one team's schedule. I've literally calculated it myself, using my own rankings. I'm sure you would disagree with my rankings in some places but they are going to be at least in the right ballpark and changing them slightly would not significantly impact the results.

OpponentHome/AwayOpponent RankingOpponent RatingHome/Away AdjustmentAdjusted Opponent Rating
BuffaloHome
145​
0.39​
-2.8​
-2.41​
Prairie ViewNeutral
311​
-14.29​
0​
-14.29​
Seton HallHome
31​
15.14​
-2.8​
12.34​
UNLVAway
110​
4.26​
2.8​
7.06​
ArizonaNeutral
3​
24.84​
0​
24.84​
Tarleton StHome
187​
-2.89​
-2.8​
-5.69​
North CarolinaAway
43​
13.51​
2.8​
16.31​
San Diego StHome
38​
14.16​
-2.8​
11.36​
NebraskaAway
180​
-2.36​
2.8​
0.44​
MinnesotaHome
56​
10.72​
-2.8​
7.92​
Southern UtahHome
151​
-0.09​
-2.8​
-2.89​
UCFAway
84​
7.36​
2.8​
10.16​
RutgersAway
104​
4.68​
2.8​
7.48​
IllinoisAway
20​
17.30​
2.8​
20.10​
Average
6.62

If you do this for every team you will find that Michigan has the 7th highest average adjusted opponent rating. And I am confident this answer is reasonable because a bunch of other computer ratings that all use different methodologies than mine (and each other) all came up with similar answers.


Because they have the 32nd highest adjusted efficiency margin. It's literally a mathematical formula, not voodoo magic or Mr. Pomeroy staring at the teams trying to figure out which one is #31 and which one is #33. It's just math.


Yes.


For the same reason that 2+2=4. You just do the math. You are trying to out-logic a mathematical formula and it is why you fail.


No. You're just a guy saying things. Trying to analyze a massive dataset with your brain is the domain of idiots.

FWIW my rankings have them #81 because there is a bonus/penalty for winning and losing. Kenpom does not measure wins and losses. it does not care about wins and losses. It's not a resume rank, it's an adjusted efficiency margin. You don't even understand the things you are criticizing, so why should anyone take your criticism seriously?
You said all computer models. Well you have done your own and arrive at 81 not 32 as Kenpom you say doesn’t measure wins and losses or by how much you win or lose. What computer model does? If none take that into account , then you have validated my point the computer model is flawed. The thickness of your brain cannot see how the models are flawed is even more troubling. You have been trying to justify why a ****** 7-7 Michigan team with not 1 quality win and tons of bad blowout or double digit losses is justified being ranked so high and I am telling you the computer models are flawed and you have confirmed it. Why the f—- have you tried so hard to be so snarky when all you had to say was the computers have them rated high but those models do not take everything into account that someone with 2 eyes could see ? You are one big Richard. !!!
 

fluoxetine

Heisman
Nov 11, 2012
23,529
16,898
0
You said all computer models. Well you have done your own and arrive at 81 not 32 as Kenpom you say doesn’t measure wins and losses or by how much you win or lose. What computer model does? If none take that into account , then you have validated my point the computer model is flawed. The thickness of your brain cannot see how the models are flawed is even more troubling. You have been trying to justify why a ****** 7-7 Michigan team with not 1 quality win and tons of bad blowout or double digit losses is justified being ranked so high and I am telling you the computer models are flawed and you have confirmed it. Why the f—- have you tried so hard to be so snarky when all you had to say was the computers have them rated high but those models do not take everything into account that someone with 2 eyes could see ? You are one big Richard. !!!
What the **** is wrong with you? I'm snarky? Every post you make is snarky and you don't even understand what you are replying to.

The "every computer model" agreement I referred to is regarding strength of schedule. Not Michigan's rating. Their strength of schedule. Learn to read and learn to use the enter key before you try to tell people how models you clearly don't understand in the slightest are flawed.

Michigan was a 7.5-point favorite on the road against Maryland today. This is basically exactly what Torvik (who had Michigan #39 before the game) predicted. Presumably you think you are smarter than the market as well? And yet you mysteriously have not cashed in on your genius...
 
  • Like
Reactions: S.W.A.I.N

ecojew

All-Conference
Feb 1, 2006
9,767
2,271
0
What the **** is wrong with you? I'm snarky? Every post you make is snarky and you don't even understand what you are replying to.

The "every computer model" agreement I referred to is regarding strength of schedule. Not Michigan's rating. Their strength of schedule. Learn to read and learn to use the enter key before you try to tell people how models you clearly don't understand in the slightest are flawed.

Michigan was a 7.5-point favorite on the road against Maryland today. This is basically exactly what Torvik (who had Michigan #39 before the game) predicted. Presumably you think you are smarter than the market as well? And yet you mysteriously have not cashed in on your genius...

Michigan's games against SHU and last night against UMd were home games, not road games.
 
  • Like
Reactions: knights1212

fluoxetine

Heisman
Nov 11, 2012
23,529
16,898
0
Michigan's games against SHU and last night against UMd were home games, not road games.
I put SHU as a home game in my chart so not sure why you mentioned that one. You are correct, they were not on the road last night. However, the point still stands as the Torvik prediction was Michigan -7.8.
 

Loyal_2RU

Heisman
Aug 6, 2001
15,234
11,049
113
What the **** is wrong with you? I'm snarky? Every post you make is snarky and you don't even understand what you are replying to.

The "every computer model" agreement I referred to is regarding strength of schedule. Not Michigan's rating. Their strength of schedule. Learn to read and learn to use the enter key before you try to tell people how models you clearly don't understand in the slightest are flawed.

Michigan was a 7.5-point favorite on the road against Maryland today. This is basically exactly what Torvik (who had Michigan #39 before the game) predicted. Presumably you think you are smarter than the market as well? And yet you mysteriously have not cashed in on your genius...
Fluoxetine,
I think that you have your own ratings is really cool and brings something to the boards.

That said the first rule of programming is still GIGO (not directed to your rankings). So in evaluating any of the programs the issue of the quality of the algorithm is a legitimate topic of conversation.

It may be math, but whose math.
I an not interested in rating teams relative to their efficiency ratings. I am interested in wins and losses. And there are teams that lose even when efficient and others that win despite being inefficient. That interests me.

There are teams that play dude to the level of competition, some always win, some not. The latter should be down graded not the former. Etc.

So lots to disagree with regarding rankings, algorithms, and the computers well beyond bias per se. It depends on what you -- and the program -- value as important.

Similarly which of those factors are used to assess SOS (record, home/N/away, efficiency, margin) may matter. The broader the level of consensual validation, the more likely it is accurate, but only within the range of the algorithms contributing.

Wonder if @SkilletHead2 has thoughts

Loyal
 

fluoxetine

Heisman
Nov 11, 2012
23,529
16,898
0
Fluoxetine,
I think that you have your own ratings is really cool and brings something to the boards.

That said the first rule of programming is still GIGO (not directed to your rankings). So in evaluating any of the programs the issue of the quality of the algorithm is a legitimate topic of conversation.
I agree with you. I really should try to be more polite to people but unfortunately I get frustrated as I've had similar arguments before. The issue, and the thing that frustrates me, is that it's very clear that he does not understand how these things work and he has just looked at Michigan's schedule and decided that it could not possibly be the 7th toughest schedule which is really obviously not a valid way to go about it.

With respect to garbage in garbage out, it's* literally just using the score of basketball games. It's tough for that to be garbage.

*Bart is doing some more complicated stuff that discounts garbage time and stuff. But the others (Kenpom, Massey, Sagarin, mine) are literally just using scores.

The influence of preseason ratings is also a reasonable concern, but those do not have significant influence at this point in the season.

It may be math, but whose math.
I an not interested in rating teams relative to their efficiency ratings. I am interested in wins and losses. And there are teams that lose even when efficient and others that win despite being inefficient. That interests me.
That's fine, but it doesn't make efficiency ratings wrong or invalid, it just means that you are interested in measuring a different thing.
There are teams that play dude to the level of competition, some always win, some not. The latter should be down graded not the former. Etc.
I'm not sure I agree. Or at least, I'm not sure I agree from a predictive sense. I think something like Kenpom is superior from a predictive perspective and that the results of close games etc are mostly (not 100%, but mostly) noise.

If I were designing something to select a tournament field I would want wins/losses to be considered more because I'm not just trying to predict the best teams going forward, I'm trying to reward those who won in the past. The reason mine work the way they do is not because I think it generates better predictions (Kenpom is definitely >>>> me from a predictive standpoint) but because I wanted something that lined up with the way humans tend to rank teams.
So lots to disagree with regarding rankings, algorithms, and the computers well beyond bias per se. It depends on what you -- and the program -- value as important.

Similarly which of those factors are used to assess SOS (record, home/N/away, efficiency, margin) may matter. The broader the level of consensual validation, the more likely it is accurate, but only within the range of the algorithms contributing.
Agreed, but I think that's different than them being wrong or invalid.

With regards to SOS I was actually surprised to find so much agreement because everyone has different ways of measuring it and there is not one right answer there.
Wonder if @SkilletHead2 has thoughts

Loyal
 
  • Like
Reactions: Loyal_2RU

G- RUnit

All-American
Sep 13, 2004
14,373
7,976
113
Cuse, Louisville and UVA all very down. It's almost weird. And Boston College used to have a good program.
 
  • Like
Reactions: ecojew

SkilletHead2

All-American
Sep 30, 2005
24,451
9,276
113
Fluoxetine,
I think that you have your own ratings is really cool and brings something to the boards.

That said the first rule of programming is still GIGO (not directed to your rankings). So in evaluating any of the programs the issue of the quality of the algorithm is a legitimate topic of conversation.

It may be math, but whose math.
I an not interested in rating teams relative to their efficiency ratings. I am interested in wins and losses. And there are teams that lose even when efficient and others that win despite being inefficient. That interests me.

There are teams that play dude to the level of competition, some always win, some not. The latter should be down graded not the former. Etc.

So lots to disagree with regarding rankings, algorithms, and the computers well beyond bias per se. It depends on what you -- and the program -- value as important.

Similarly which of those factors are used to assess SOS (record, home/N/away, efficiency, margin) may matter. The broader the level of consensual validation, the more likely it is accurate, but only within the range of the algorithms contributing.

Wonder if @SkilletHead2 has thoughts

Loyal
Thanks for the invite to comment, Loyal!

Way too long an answer below. Short version. It depends strongly on what you are trying to predict (W/Ls or magins), whether you want to rely heavily on last year's results (which are highly reliable but not super valid), and how you want to treat playing weak teams and injuires/illnesses. I have to say that I sometimes look at Kenpom and Sagarin and have a real WTF moment!

Longer version (and happy to answer questions):

I love analytics and probability, and do a fair amount of it in my day job. Whether a team is going to win a game or not is not actually dissimilar to whether a test-taker is going to get the next item on a test right or wrong in a computer-based system. And I work a lot in that arena.

The folks who make the computer models are faced with a number of choices, assumptions, and prior information in setting up their models. They also have to decide just what it is they are trying to predict (win or less, margin of victory -- you can bet on either). They also decide if they are going to start with estimates of ability of the teams, and how much they are going to let this season's wins and losses (and margins) influence changes in those prior estimates (called "priors" in Bayesian stats). An important point here is that there isn't so much rights and wrongs as there are choices, goals, and whether you want to pick the winning team as much as possible or be within the spread as much as possible.

Another factor is recency versus primacy. That means are you more interested in what they've done in the past five games or where they started the season at? And then, and this is what kind of kills everything, what about injuries, a player getting COVID, etc.? Do you factor that in or just go with the data from the scoreboards?

So, here are some of my thoughts. If you play 12 teams who are middling tough and go 6-6 and another team plays 6 patsies and 6 teams in the top 10, and they go 6-6, how do you compare those two? In the testing model I use, you look at the current estimate of your ability and the difficulty of your opponent (or the test item), combine them in a reasonably straightforward mathematical model, and then come up with a probability of getting the next item right. The estimate of ability that I use is dependent on just two pieces of info: how many you got right (total wins), and the ability of each team you played (or how tough the items were). Whenever we see an anomaly, we look at what happened, because highly unusual events should not occur often (a player was sick, an examinee made a lucky guess). We see too many of these, and we get really suspicious (that is to say, cheating).

For me, I think the fairest way to do this is to start out with everybody equal or maybe with just three or so levels of prior estimates of ability. Then, each week, you look at the data, adjust teams' ability estimates, and then continue to re-run the data until it "settles." This is doable, but I'm not sure how many programs do it. I have a buddy who is very good at this and did it for the NFL for a couple of years and had a really nice set of estimates going. I'm not a good enough computer programmer to do that, nor do I want to invest the time in it. Skillethead Jr. is a much better programmer than I am, but he's got a job and a family. But hey, I've got a nephew who just invented a VR game called Gorilla Tag that has over 2 million downloads. Maybe I could talk him into it!
 
  • Like
Reactions: Loyal_2RU

fluoxetine

Heisman
Nov 11, 2012
23,529
16,898
0
What is your day job? There is a ton I agree with in your post so only really going to reply to parts I disagree with or have a comment on.
Thanks for the invite to comment, Loyal!

Way too long an answer below. Short version. It depends strongly on what you are trying to predict (W/Ls or magins), whether you want to rely heavily on last year's results (which are highly reliable but not super valid), and how you want to treat playing weak teams and injuires/illnesses. I have to say that I sometimes look at Kenpom and Sagarin and have a real WTF moment!

Longer version (and happy to answer questions):

I love analytics and probability, and do a fair amount of it in my day job. Whether a team is going to win a game or not is not actually dissimilar to whether a test-taker is going to get the next item on a test right or wrong in a computer-based system. And I work a lot in that arena.

The folks who make the computer models are faced with a number of choices, assumptions, and prior information in setting up their models. They also have to decide just what it is they are trying to predict (win or less, margin of victory -- you can bet on either). They also decide if they are going to start with estimates of ability of the teams, and how much they are going to let this season's wins and losses (and margins) influence changes in those prior estimates (called "priors" in Bayesian stats). An important point here is that there isn't so much rights and wrongs as there are choices, goals, and whether you want to pick the winning team as much as possible or be within the spread as much as possible.

I agree with pretty much all of that. The one thing I disagree with is that modeling margins or win probability are fundamentally different problems. I think you would find that most sports modelers (be they linesmakers, professional sports bettors, or simply hobbyists) are going to be using a model that predicts margins and then a fairly simple logistic regression to transform those margins to winning percentages. Teams that win a bunch of close games aren't going to get a boost in their moneyline relative to the point spread because those results are mostly non-repeatable.

If you play 12 teams who are middling tough and go 6-6 and another team plays 6 patsies and 6 teams in the top 10, and they go 6-6, how do you compare those two?
For the most part your estimates of their ability would be equal, but you would have a lot more uncertainty about that estimate for the first team than you would for the 2nd.

For me, I think the fairest way to do this is to start out with everybody equal or maybe with just three or so levels of prior estimates of ability. Then, each week, you look at the data, adjust teams' ability estimates, and then continue to re-run the data until it "settles." This is doable, but I'm not sure how many programs do it. I have a buddy who is very good at this and did it for the NFL for a couple of years and had a really nice set of estimates going. I'm not a good enough computer programmer to do that, nor do I want to invest the time in it. Skillethead Jr. is a much better programmer than I am, but he's got a job and a family. But hey, I've got a nephew who just invented a VR game called Gorilla Tag that has over 2 million downloads. Maybe I could talk him into it!
If you can articulate your model clearly here (and it relies on scores only, as I do not have a good source of other data) I might (emphasis on might) code it up.
 

SkilletHead2

All-American
Sep 30, 2005
24,451
9,276
113
What is your day job? There is a ton I agree with in your post so only really going to reply to parts I disagree with or have a comment on.


I agree with pretty much all of that. The one thing I disagree with is that modeling margins or win probability are fundamentally different problems. I think you would find that most sports modelers (be they linesmakers, professional sports bettors, or simply hobbyists) are going to be using a model that predicts margins and then a fairly simple logistic regression to transform those margins to winning percentages. Teams that win a bunch of close games aren't going to get a boost in their moneyline relative to the point spread because those results are mostly non-repeatable.


For the most part your estimates of their ability would be equal, but you would have a lot more uncertainty about that estimate for the first team than you would for the 2nd.


If you can articulate your model clearly here (and it relies on scores only, as I do not have a good source of other data) I might (emphasis on might) code it up.
Actually, my day job, until Jan 31, is being Dean of the College of Education at the University of Otago in the South Island of New Zealand (I'm guessing that isn't what you were expecting!). I'm a Professor here and have for years specialized in research methods and quantitative measurement models. Bizarrely enough, I also do a lot of work on the psychology of aesthetics (how people look at art). You can get my latest book on Amazon (shameless plug): Scoundrels, Cads, and Other Great Artists. It's a lot of fun to read -- about the bad boys in art and the great art they made. My PhD is in Measurement, Evaluation, and Statistical Analysis (U of Chicago '77).

Margins and wins models only really differ in the dependent variable, and then the approach taken to estimate those models (as dichotomous variables are inherently more difficult to model). I don't think of those as very different, but that would be a quibble.

Your second point is spot on! The difference from most models is in the standard error here. But that irks me nonetheless. Feeding on the weak teams seems objectionable to me. And we know next to nothing about such teams.

Let me get you a good reference on the model I use. For dichotomous outcomes, it's called "the Rasch model" and it's a special case of something called item response theory. I'll see if I can get you a good explanation of it.
 

Loyal_2RU

Heisman
Aug 6, 2001
15,234
11,049
113
Actually, my day job, until Jan 31, is being Dean of the College of Education at the University of Otago in the South Island of New Zealand (I'm guessing that isn't what you were expecting!). I'm a Professor here and have for years specialized in research methods and quantitative measurement models. Bizarrely enough, I also do a lot of work on the psychology of aesthetics (how people look at art). You can get my latest book on Amazon (shameless plug): Scoundrels, Cads, and Other Great Artists. It's a lot of fun to read -- about the bad boys in art and the great art they made. My PhD is in Measurement, Evaluation, and Statistical Analysis (U of Chicago '77).

Margins and wins models only really differ in the dependent variable, and then the approach taken to estimate those models (as dichotomous variables are inherently more difficult to model). I don't think of those as very different, but that would be a quibble.

Your second point is spot on! The difference from most models is in the standard error here. But that irks me nonetheless. Feeding on the weak teams seems objectionable to me. And we know next to nothing about such teams.

Let me get you a good reference on the model I use. For dichotomous outcomes, it's called "the Rasch model" and it's a special case of something called item response theory. I'll see if I can get you a good explanation of it.
@fluoxetine @SkilletHead2

I called in a ringer, lol. I have one of Skillet's books on my desk as a reference whenever I do statistical programming.

Congratulations on your retiring the Deanship. Staying in NZ for the future or coming back to Piscataway. How did you end up there?

Loyal
 
  • Like
Reactions: fluoxetine

SkilletHead2

All-American
Sep 30, 2005
24,451
9,276
113
@fluoxetine @SkilletHead2

I called in a ringer, lol. I have one of Skillet's books on my desk as a reference whenever I do statistical programming.

Congratulations on your retiring the Deanship. Staying in NZ for the future or coming back to Piscataway. How did you end up there?

Loyal
You have the SAS and Statistics book with Ron Cody? We had to go to 17 publishers to get that published the first time, and then it went through five editions. By far the most profitable book I've ever written!

I'm not retiring yet, just stepping down from Deanship. I'm taking a five month sabbatical to work on a book on how to look at art: Personal Connoisseurship: Making Art Meaningful, and then I'll cut back to half time for about two more years. I really enjoy teaching and researching, and the Uni really wants me to stay on to mentor the young folks. So I'll come in two days a week. Should be fun.

We always tell people we came down here as part of the witness protection program, but really, we just wanted to have a big adventure, and RU and Kean (where Moms Skillethead was a prof) each gave us a two-year leave of absence to try it out. That was 16 years ago. Moms eventually became Dean of the College of Ed before me, and is now retired even though she is five years younger than I am. My plan is to work until I'm 74 and then retire. It's indoor work and no heavy lifting. I love writing books for the general public. It's my new hobby!
 

S.W.A.I.N

All-Conference
Nov 23, 2004
4,579
4,863
81
You have the SAS and Statistics book with Ron Cody? We had to go to 17 publishers to get that published the first time, and then it went through five editions. By far the most profitable book I've ever written!

I'm not retiring yet, just stepping down from Deanship. I'm taking a five month sabbatical to work on a book on how to look at art: Personal Connoisseurship: Making Art Meaningful, and then I'll cut back to half time for about two more years. I really enjoy teaching and researching, and the Uni really wants me to stay on to mentor the young folks. So I'll come in two days a week. Should be fun.

We always tell people we came down here as part of the witness protection program, but really, we just wanted to have a big adventure, and RU and Kean (where Moms Skillethead was a prof) each gave us a two-year leave of absence to try it out. That was 16 years ago. Moms eventually became Dean of the College of Ed before me, and is now retired even though she is five years younger than I am. My plan is to work until I'm 74 and then retire. It's indoor work and no heavy lifting. I love writing books for the general public. It's my new hobby!
You may find my brother’s book of interest - https://www.psupress.org/books/titles/978-0-271-06501-4.html
 
  • Like
Reactions: SkilletHead2

Loyal_2RU

Heisman
Aug 6, 2001
15,234
11,049
113
You have the SAS and Statistics book with Ron Cody? We had to go to 17 publishers to get that published the first time, and then it went through five editions. By far the most profitable book I've ever written!

I'm not retiring yet, just stepping down from Deanship. I'm taking a five month sabbatical to work on a book on how to look at art: Personal Connoisseurship: Making Art Meaningful, and then I'll cut back to half time for about two more years. I really enjoy teaching and researching, and the Uni really wants me to stay on to mentor the young folks. So I'll come in two days a week. Should be fun.

We always tell people we came down here as part of the witness protection program, but really, we just wanted to have a big adventure, and RU and Kean (where Moms Skillethead was a prof) each gave us a two-year leave of absence to try it out. That was 16 years ago. Moms eventually became Dean of the College of Ed before me, and is now retired even though she is five years younger than I am. My plan is to work until I'm 74 and then retire. It's indoor work and no heavy lifting. I love writing books for the general public. It's my new hobby!

Amazing. I'm now a Prof and Vice Chair at RWJMS. Yes I have bought at least 20 coed of the hat 2v editing for fellows and personal use. I've developed a bit of statistical methods but am only intermittently working with data these days. And life has made me a covid researcher. Although for years on the QT I said my field was injuring health by measuring difficult constructs...
 
  • Like
Reactions: SkilletHead2

SkilletHead2

All-American
Sep 30, 2005
24,451
9,276
113
Amazing. I'm now a Prof and Vice Chair at RWJMS. Yes I have bought at least 20 coed of the hat 2v editing for fellows and personal use. I've developed a bit of statistical methods but am only intermittently working with data these days. And life has made me a covid researcher. Although for years on the QT I said my field was injuring health by measuring difficult constructs...
So, do you know Ron? He's a close friend.
 

dconifer0

All-Conference
Oct 4, 2004
4,342
3,355
113
i have often found that computerized strength of schedule rankings do not meet my eye test. the source of the disconnect, in a nutshell is that in my view there is little difference in playing the #150 team and the #300 team; they both suck. but per computer, there is a huge difference (or so it seems, I have not seen the algorithms)...
 
Last edited: