Rankings & Ratings after Bloody New Year

[visitingcock]

What a glorious weekend for chaos!

We have to wait til Noonish for the AP poll. However, the first computer rating to reflect the upsets is Massey which updated at 5:00 am this morning.

5. Oklahoma (undefeated up 3 places)
6. LSU (even after consecutive losses)
7. Vandy (undefeated up 9 places)
8. Iowa (up 5 places)
9. U$C (even after a beat down)
10. Maryland
11. Kentucky (up 7 places)

TCU fell 7 places to 13 and Iowa State fell 9 spots to 16 following their first losses. Oh yeah, Notre Dame fell 10 places and Duke rose 7.

Cant wait to see what the AP eyeball tests do at noon. The present battle for two seed spots is fascination while Texas and SCar were underwhelming yesterday. Pretty certain LSU won’t be 5th among humans.

 

[BRS24]

Very interesting re LSU. They played a week early schedule and have now lost to KY and VU back to back and they didn't move? Right now they are #5 in AP poll, and I hope it gives some of the voters a chance to really rank them where they should be. It also seems as if the preseason voting is sketchy, as they've not moved from the 5 spot. And this is why I'd love the AP poll to not start until January, as that's when things start to get sorted out as to what votes thought a team might be, vs the reality.

Conversely, Duke's preseason rank was 10, then they went 8, 16, UR, and the last 5 weeks didn't even get a vote, but that may change within the next week or so. Meanwhile (and I snicker about this), there are folks out there that want to jettison UConn out of the #1 spot, when they've played Louisville, OSU, MI, Utah, USC, Iowa, and have ND and Tenn to go.

 

[cancontent]

Very interesting re LSU. They played a week early schedule and have now lost to KY and VU back to back and they didn't move? Right now they are #5 in AP poll, and I hope it gives some of the voters a chance to really rank them where they should be. It also seems as if the preseason voting is sketchy, as they've not moved from the 5 spot. And this is why I'd love the AP poll to not start until January, as that's when things start to get sorted out as to what votes thought a team might be, vs the reality.

Conversely, Duke's preseason rank was 10, then they went 8, 16, UR, and the last 5 weeks didn't even get a vote, but that may change within the next week or so. Meanwhile (and I snicker about this), there are folks out there that want to jettison UConn out of the #1 spot, when they've played Louisville, OSU, MI, Utah, USC, Iowa, and have ND and Tenn to go.

Thought there was a holiday break and the updated AP poll comes out later this morning/afternoon. There should be a fair amount of movement for several teams, not only LSU.

 

[Viet Vet]

"there are folks out there that want to jettison UConn out of the #1 spot"..........

UConn hoops are like the Yankees, Patriots and being in the Army. You love them or hate them; there is very little middle ground.

 

[BRS24]

Thought there was a holiday break and the updated AP poll comes out later this morning/afternoon. There should be a fair amount of movement for several teams, not only LSU.

I should have made that clear, as yes, we've been on the same ranking for two weeks. I wouldn't be surprised to see some teams not move down as much as they should, but what do I know.

 

[Centerstream]

Gotta remember that the AP voters are humans (or AI?) and probably use Massey, NET and other algorithm based rankings to help them select. Then you throw in their partiality and the fun begins. If LSU can lose their first two conference games, to ranked teams, and not hardly move in Massey and NET, why would the voters disagree? Same goes for other teams that lost over the last two weeks.

 

[FanFromMaine]

Very interesting re LSU. They played a week early schedule and have now lost to KY and VU back to back and they didn't move? Right now they are #5 in AP poll, and I hope it gives some of the voters a chance to really rank them where they should be. It also seems as if the preseason voting is sketchy, as they've not moved from the 5 spot. And this is why I'd love the AP poll to not start until January, as that's when things start to get sorted out as to what votes thought a team might be, vs the reality.

Conversely, Duke's preseason rank was 10, then they went 8, 16, UR, and the last 5 weeks didn't even get a vote, but that may change within the next week or so. Meanwhile (and I snicker about this), there are folks out there that want to jettison UConn out of the #1 spot, when they've played Louisville, OSU, MI, Utah, USC, Iowa, and have ND and Tenn to go.

Massey also might as well have started after New Year. The data are just now coming into focus with so many teams living near Dick Vitale (Cupcake City).

For example, looking at the PWR column in Massey, UConn finally outranks South Carolina by one one-hundredth of a point. This is the column that most obviously carries over data from previous seasons. But, despite two major victories by the Huskies over the Gamecocks, the algorithm required 15 games this year to reflect what we all saw.

Maybe this is the new reality in the world of mega conferences with a large number of league games.

 

[diggerfoot]

In general I’ve always trusted algorithms over humans for mitigating bias. The bias of an algorithm is static, based solely on the decisions for the initial inputs. Over time those inputs may change slowly, generally to reflect better informed decisions of what should be measured. The biases of human opinions in the moment are far more arbitrary and dynamic.

I’ve always respected Massey in particular, but the lack of movement for LSU after two consecutive losses to finally decent competition seems to warrant a rare adjustment of the metrics.

 

[Brewski]

OKLAHOMA is not undefeated. They lost to UCLA. It makes no difference in the ranking though. I think they will be ranked 5 in the upcoming AP poll.

 

[visitingcock]

OKLAHOMA is not undefeated. They lost to UCLA. It makes no difference in the ranking though. I think they will be ranked 5 in the upcoming AP poll.

oops. Sorry

 

[NycUcWbbFan]

In general I’ve always trusted algorithms over humans for mitigating bias. The bias of an algorithm is static, based solely on the decisions for the initial inputs. Over time those inputs may change slowly, generally to reflect better informed decisions of what should be measured. The biases of human opinions in the moment are far more arbitrary and dynamic.

I’ve always respected Massey in particular, but the lack of movement for LSU after two consecutive losses to finally decent competition seems to warrant a rare adjustment of the metrics.

Massey’s Model, at its core, is a system of equations where each team has one equation and that equation is a linear combination of its games that is set equal to a combined MOV.

So feasting on cupcakes does make Massey sticky.

The same thing holds for Pythagorean Expectation methods such as Torvik and NET. A severe diet of cupcakes can be sticky for a time.

The Committee’s Quad System becomes crucial here. Perhaps it’s humorous poetic justice that LSU lost its first Quad 1 game for being ill-prepared, and the lack of preparation was due to trying to boost its NET with cupcakes, which in turn was due to its attempt to avoid avoidable OOC Quad 1 losses.

 

[YoDore]

After beating LSU, Vandy dropped a spot in the NET from 13 to 14. LSU stays the same at 5.

 

[FanInNC]

After beating LSU, Vandy dropped a spot in the NET from 13 to 14. LSU stays the same at 5.

Yeah the NET is a bit of a joke at times. For instance :

Team A:
SOS 1
Quad 1: 2-3
Quad 2: 3-1
Quad 3: 2-0
Quad 4: 4-0

Team B:
SOS 2
Quad 1: 1-5
Quad 2: 3-1
Quad 3: 2-0
Quad 4: 3-0

Team C:
SOS 65
Quad 1: 0-3
Quad 2: 3-1
Quad 3: 2-0
Quad 4: 8-0

To me it's pretty obvious Team C has the weakest resume, by far the weakest schedule, no Q1 wins, same record in Q2 games and just has loaded up against the sister's of the poor Q4 games. Yet according to the NET, Team C is ranked the highest of these 3 teams.

Team C 18th (UNC)
Team B 19th (Duke)
Team A 25th (State)

 

[BigTenGuy]

Yeah the NET is a bit of a joke at times. For instance :

Team A:
SOS 1
Quad 1: 2-3
Quad 2: 3-1
Quad 3: 2-0
Quad 4: 4-0

Team B:
SOS 2
Quad 1: 1-5
Quad 2: 3-1
Quad 3: 2-0
Quad 4: 3-0

Team C:
SOS 65
Quad 1: 0-3
Quad 2: 3-1
Quad 3: 2-0
Quad 4: 8-0

To me it's pretty obvious Team C has the weakest resume, by far the weakest schedule, no Q1 wins, same record in Q2 games and just has loaded up against the sister's of the poor Q4 games. Yet according to the NET, Team C is ranked the highest of these 3 teams.

Team C 18th (UNC)
Team B 19th (Duke)
Team A 25th (State)

Suspect you're at least somewhat a fan of the RPI, which as of this morning is:

State 9
Duke 22
UNC 44

 

[FanInNC]

Suspect you're at least somewhat a fan of the RPI, which as of this morning is:

State 9
Duke 22
UNC 44

Becoming a bigger fan by the minute.

I mean honestly, State isn’t a top 10 team but unranked and behind UNC and Duke in the NET and getting less votes than Stanford after the result this past week all feels like a major slight. Not to mention being behind Notre Dame in votes lol.

 

[triaddukefan]

Becoming a bigger fan by the minute.

I mean honestly, State isn’t a top 10 team but unranked and behind UNC and Duke in the NET and getting less votes than Stanford after the result this past week all feels like a major slight. Not to mention being behind Notre Dame in votes lol.

Duke > State.

 

[Suebird10fan]

Massey’s Model, at its core, is a system of equations where each team has one equation and that equation is a linear combination of its games that is set equal to a combined MOV.

So feasting on cupcakes does make Massey sticky.

The same thing holds for Pythagorean Expectation methods such as Torvik and NET. A severe diet of cupcakes can be sticky for a time.

The Committee’s Quad System becomes crucial here. Perhaps it’s humorous poetic justice that LSU lost its first Quad 1 game for being ill-prepared, and the lack of preparation was due to trying to boost its NET with cupcakes, which in turn was due to its attempt to avoid avoidable OOC Quad 1 losses.

On Torvik I use a filter to look at ratings based only on games vs. top 50 or top 100 teams.