Aike’s Model - 2/21/25 Update

[Aike]

Florida and Missouri both looking scary. Gonzaga is going to be painfully underseeded. Cats hanging tough through adversity. Pitino primed and ready. Auburn is still Auburn.

1. Auburn 4.29
2. Duke 3.41
3. Tennessee 3.24
4. Houston 2.94
5. Florida 2.83
6. Gonzaga 2.82
7. Arizona 2.73
8. Iowa St 2.63
9. Texas Tech 2.48
10. Missouri 2.31
11. Purdue 2.19
12. Alabama 2.17
13. Mississippi St 2.16
14. Maryland 2.14
15. Connecticut 2.13
16. Michigan 2.12
17. St John’s 2.04
18. Kentucky 2.00
19. Clemson 1.97
20. Kansas 1.96
21. Mississippi 1.91
22. Michigan St 1.86
23. Marquette 1.85
24. BYU 1.84
25. UCLA 1.80
27. Arkansas 1.66
28. Louisville 1.59
30. Vanderbilt 1.52
32. Georgia 1.48
34. Texas A&M 1.40
37. Ohio St 1.38
43. Texas 1.26
47. Oklahoma 1.15
48. N Carolina 1.14
60. Indiana 0.83
71. LSU 0.67
89. S Carolina 0.44

 

[Doc_53]

Wisconsin?

 

[Aike]

Wisconsin?

Closed it down but I think they were 26.

 

[Montana81]

I am not a believer in Gonzaga. They did not look great in the non con and their metrics have once again began to sky rocket as their competition weakened.

In years when they’ve been a real threat they looked great in the non conference too.

That’s the hill i’m dying on. We’ll just have to wait and see.

 

[Montana81]

Also any way you slice it this is the best the sec has ever been. Brutal year for a new coach with a complete roster reset to try and right the ship.

 

[Aike]

I am not a believer in Gonzaga. They did not look great in the non con and their metrics have once again began to sky rocket as their competition weakened.

In years when they’ve been a real threat they looked great in the non conference too.

That’s the hill i’m dying on. We’ll just have to wait and see.

I’ll say what I’ve said about Gonzaga before. They have 4 losses by a combined 17 points, and 3 more losses in OT.

If playing them, I would expect a win in a game that went down to the wire, or a loss. I sure wouldn’t want to have to face them in the second round as a 1 seed.

 

[Digger-Cat]

5 wins over the top 6, gotta like that, especially since some of those wins came with a key player out.

 

[Aike]

5 wins over the top 6, gotta like that, especially since some of those wins came with a key player out.

We do tend to rise to the occasion.

 

[Digger-Cat]

We do tend to rise to the occasion.

This game tomorrow night is another one of those where weird things happen when CMP is coaching. I'm not expecting it, but if he pulls off another win, that could move us into 2 seed consideration.

 

[entropy13]

I’ll say what I’ve said about Gonzaga before. They have 4 losses by a combined 17 points, and 3 more losses in OT.

If playing them, I would expect a win in a game that went down to the wire, or a loss. I sure wouldn’t want to have to face them in the second round as a 1 seed.

Gonzaga won't be an 8 or 9 seed, and thus wouldn't play a 1 seed in the second round. Assume you mean they might be a 4-5 seed and thus possibly face the 1 seed in the Sweet 16?

Also, your model predicts estimated NCAA tournament wins, correct? If so, I think that forecast looks accurate for Kentucky, that is, we make it to the Sweet 16. Get our main dudes back from injury and playing well, and it can only go up!

 

[The_Answer1313]

I am not a believer in Gonzaga. They did not look great in the non con and their metrics have once again began to sky rocket as their competition weakened.

In years when they’ve been a real threat they looked great in the non conference too.

That’s the hill i’m dying on. We’ll just have to wait and see.

I predict Gonzaga is going to be a 8/9 seed and give the 1 seed in that region absolute fits. Probably not win the game as they will still be underdog in that one but a lot closer than what most 8/9 vs 1 games would be.

I think an argument can be made maybe they shouldn't be say a 6th in Aike's model or say 10th like they are in Kenpom and whatnot.

But what a massive underseed they are going to get if Bracket Matrix is correct.

I would really be interested to take two groups of teams. One group where the metrics say they are better than their seed but their resume has brought them down and another group where the resumes are really good but the computer metrics are so so.

I'd venture to guess group 1 overperforms their seed line moreso than group 2 overperforms theirs.

 

[KY Long Rifle]

Aike,

Thanks for tracking this and continually posting updates. I enjoy keeping up with them.

 

[Aike]

Gonzaga won't be an 8 or 9 seed, and thus wouldn't play a 1 seed in the second round. Assume you mean they might be a 4-5 seed and thus possibly face the 1 seed in the Sweet 16?

Also, your model predicts estimated NCAA tournament wins, correct? If so, I think that forecast looks accurate for Kentucky, that is, we make it to the Sweet 16. Get our main dudes back from injury and playing well, and it can only go up!

Gonzaga is projected on that 8/9 line currently. Maybe they’ll move up a little if they win their conference.

And yes, the model predicts “unadjusted” number of tourney wins. Number will change a little based on the bracket, since there are only 63 wins available.

I agree that we have the look and feel of a Sweet 16 level team if we are reasonably healthy.

 

[runt#69]

Auburn has definitely the feel of a great team for all time this year, relative to the competition. And it's been as hard a path for them as anyone. Simply astounding how well they have performed. It would be ashame to see them fall flat and not ball out in March Madness.

But my money ?

Florida

Great job and thanks, Aike. Looks solid.

 

[The_Answer1313]

Gonzaga won't be an 8 or 9 seed, and thus wouldn't play a 1 seed in the second round. Assume you mean they might be a 4-5 seed and thus possibly face the 1 seed in the Sweet 16?

Currently that's exactly where they are

The Bracket Project's Bracket Matrix - 2025

And given their schedule very little room to move up I figure.

 

[yabbadabbadoo]

So Maryland has (2) top 25 wins
UK has (7) top 15 wins
And they are
1,2,3,4 spots above us
Seems like some serious flaws

 

[The_Answer1313]

So Maryland has (2) top 25 wins
UK has (7) top 15 wins
And they are
1,2,3,4 spots above us
Seems like some serious flaws

How many opportunities does Maryland have for top 25 wins?

How many opportunities do we have for those type of wins?

 

[Aike]

Auburn has definitely the feel of a great team for all time this year, relative to the competition. And it's been as hard a path for them as anyone. Simply astounding how well they have performed. It would be ashame to see them fall flat and not ball out in March Madness.

But my money ?

Florida

Great job and thanks, Aike. Looks solid.

Have a feeling that hothead will hurt Auburn in the tournament again. CBN.

Florida is as good a pick as anyone. Especially getting that big man back.

 

[UKBB4Ever]

Take those numbers to Vegas and bet.

Just be sure to save bus money home. 😈

 

[Aike]

Take those numbers to Vegas and bet.

Just be sure to save bus money home. 😈

Not a gambler but they hold up incredibly well where point spreads are concerned. Hard for anyone to beat Vegas consistently, since they get to keep the juice. Pays for a lot of shiny buildings.

 

[Blueheart32]

Not a gambler but they hold up incredibly well where point spreads are concerned. Hard for anyone to beat Vegas consistently, since they get to keep the juice. Pays for a lot of shiny buildings.

Amazes me when people think they know more than the books.

 

[yabbadabbadoo]

How many opportunities does Maryland have for top 25 wins?

How many opportunities do we have for those type of wins?

Duke has (1) top 15 win
Houston has (2)
Yet they stand at #2 and #4 respectively in this list
The tournament will bite these teams
Pretty obvious to me

 

[The_Answer1313]

Duke has (1) top 15 win
Houston has (2)
Yet they stand at #2 and #4 respectively in this list
The tournament will bite these teams
Pretty obvious to me

It probably doesn't matter.

I've seen UK not have great resumes in 2011 and 2014 and turn that into final fours.

Year in and year out we've heard how the Big 12 and Big Ten were dominant conferences and how those tough schedules were going to mean they were "battle tested" come March only for those teams to not live up to the seeding they got.

None of this matters. If you are a good team, your a good team. Duke and Houston will be fine IMO.

 

[Montana81]

I predict Gonzaga is going to be a 8/9 seed and give the 1 seed in that region absolute fits. Probably not win the game as they will still be underdog in that one but a lot closer than what most 8/9 vs 1 games would be.

I think an argument can be made maybe they shouldn't be say a 6th in Aike's model or say 10th like they are in Kenpom and whatnot.

But what a massive underseed they are going to get if Bracket Matrix is correct.

I would really be interested to take two groups of teams. One group where the metrics say they are better than their seed but their resume has brought them down and another group where the resumes are really good but the computer metrics are so so.

I'd venture to guess group 1 overperforms their seed line moreso than group 2 overperforms theirs.

I do the same in regards to that last sentence. I mostly trust the metrics. And aike has a good point about gonzaga ooc losses all being close. I might be being a tad hard on them.

But, and we’ve had this conversation before, a lot of games against the bottom feeders of division one…those are games where stronger teams can name their score. And Arkansas versus Maryland Eastern shore for example. They’re getting a bigger bump in performance for that game than we are for beating Tennessee, Duke or Florida.

And I understand the concept and the math and all and why that is. It’s all about net efficiency. Points per possession scored versus given up vs expected performance. But those are scenarios when it just doesn’t work imo for lots of reasons.

 

[yabbadabbadoo]

It probably doesn't matter.

I've seen UK not have great resumes in 2011 and 2014 and turn that into final fours.

Year in and year out we've heard how the Big 12 and Big Ten were dominant conferences and how those tough schedules were going to mean they were "battle tested" come March only for those teams to not live up to the seeding they got.

None of this matters. If you are a good team, your a good team. Duke and Houston will be fine IMO.

Exactly
That’s why all the Kenpom numbers, BPI, and many others mean basically nothing.

 

[bradyjames]

Aike, has UT beaten a single team in the top 15-20 that hasn’t beaten them worse(UF). Just don’t see them anywhere close to what your model shows them at.

 

[Aike]

Aike, has UT beaten a single team in the top 15-20 that hasn’t beaten them worse(UF). Just don’t see them anywhere close to what your model shows them at.

Baylor was at 13 when they played. UT won by 15.

Arkansas was at 23 when UT beat them by 24.

Georgia was at 23 when UT won by 18.

Mississippi St was at 14 when UT won by 12.

Missouri was at 15 when UT won by 4.

Beat number 5 Florida by 20, which as you pointed out was a revenge game, but they were still good enough to beat Florida by 20.

Lost to Auburn by 2.

They have a pretty strong resume. Definitely susceptible to a great offensive performance, but if you don’t come ready to play, they will cut your water off.

Now do I think they’ll make the Final Four? Seriously doubt it. But depending on matchups, a return to the Elite Eight or at least the Sweet Sixteen shouldn’t be surprising at all.

 

[Aike]

Exactly
That’s why all the Kenpom numbers, BPI, and many others mean basically nothing.

Time and again, advanced metrics do a better job of predicting results than the “eye test.”

The simplest explanation for this is that no one can watch every game, and humans tend to be more biased than the best models (although models have their own biases).

Analytics are tools to enhance your understanding of the game. Nate Oats uses math heavily, and rode it all the way to the Final Four. Mark Pope is deeper into analytics than any coach we’ve ever had. If these are all meaningless, someone should probably let those guys know.

 

[Aike]

I do the same in regards to that last sentence. I mostly trust the metrics. And aike has a good point about gonzaga ooc losses all being close. I might be being a tad hard on them.

But, and we’ve had this conversation before, a lot of games against the bottom feeders of division one…those are games where stronger teams can name their score. And Arkansas versus Maryland Eastern shore for example. They’re getting a bigger bump in performance for that game than we are for beating Tennessee, Duke or Florida.

And I understand the concept and the math and all and why that is. It’s all about net efficiency. Points per possession scored versus given up vs expected performance. But those are scenarios when it just doesn’t work imo for lots of reasons.

I don’t disagree that it’s hard to correctly capture adjustments for SOS. I have some ideas about how to make it better, but they require a bit more work and time then I currently have available.

That being said, these numbers still hold up awfully well.

One example I’ll give is Gonzaga in 2019. They were first going into the tournament with around a 3.8 score. Even though they were the favorite, that score still meant that they were most likely to fall just short of the Final Four.

Ultimately, they lost to Texas Tech in the Elite Eight. Did they lose because the model was wrong and they weren’t used to playing a team as tough as TT? Maybe.

But digging into the numbers (as I recall), TT’s foul rate was abnormally low that game (Rooferee was on the whistle).

Did TT foul less because Gonzaga was soft and bad at drawing fouls? Did Gonzaga get a bad whistle from a Big 12 friendly ref?

End of the day, the predicted margin in most of these games after the first round will be 3-4 points or less. So it doesn’t take much to flip one game the other direction.

In aggregate, the metrics hold up very well. But any one game can and will go sideways. Crazy tournament and only one team is left standing. The expected wins for the best teams are typically 2-3, but a few teams will win 4, 5, or 6 by default.

 

[bradyjames]

Baylor was at 13 when they played. UT won by 15.

Arkansas was at 23 when UT beat them by 24.

Georgia was at 23 when UT won by 18.

Mississippi St was at 14 when UT won by 12.

Missouri was at 15 when UT won by 4.

Beat number 5 Florida by 20, which as you pointed out was a revenge game, but they were still good enough to beat Florida by 20.

Lost to Auburn by 2.

They have a pretty strong resume. Definitely susceptible to a great offensive performance, but if you don’t come ready to play, they will cut your water off.

Now do I think they’ll make the Final Four? Seriously doubt it. But depending on matchups, a return to the Elite Eight or at least the Sweet Sixteen shouldn’t be surprising at all.

There is a lot of WAS. They seem to really be overrated in the polls and your model from what I have seen. Just a thought. I really enjoy your stuff.

 

[Aike]

There is a lot of WAS. They seem to really be overrated in the polls and your model from what I have seen. Just a thought. I really enjoy your stuff.

This model and most of the ones you’re talking about take the whole season into account. I do usually run another model for just the last month of the season to see how teams are trending.

 

[bradyjames]

This model and most of the ones you’re talking about take the whole season into account. I do usually run another model for just the last month of the season to see how teams are trending.

I understand that. But when five of your key wins are against one team not currently in the Top 25. And three of the other four have a losing record in the SEC. It doesn’t scream Top 5 in any measurement today, tomorrow or next month.

Again, just a thought. You have much more expertise but I can think of ten teams that are just better than UT.

 

[The_Answer1313]

I do the same in regards to that last sentence. I mostly trust the metrics. And aike has a good point about gonzaga ooc losses all being close. I might be being a tad hard on them.

But, and we’ve had this conversation before, a lot of games against the bottom feeders of division one…those are games where stronger teams can name their score. And Arkansas versus Maryland Eastern shore for example. They’re getting a bigger bump in performance for that game than we are for beating Tennessee, Duke or Florida.

And I understand the concept and the math and all and why that is. It’s all about net efficiency. Points per possession scored versus given up vs expected performance. But those are scenarios when it just doesn’t work imo for lots of reasons.

I'm not really sure the calculation behind NET since it's not really out there anywhere. I do know that with Kenpom there is a rate of diminishing returns. In other words say you are playing a weak team, the difference in rating between say a 20 point win vs a 10 point win is greater than the gain received from a 30 point win vs a 20 point win (making those numbers up i don't know when the returns start to diminish but you get the point)

So it's not just a matter of "well we can blow out X team by 80" and gain a bunch of EM doing so.

I do think tho that maybe there is something the SOS adjustment isn't exactly capturing. I do think that when you play weaker teams there's a much greater chance those teams are inconsistency and thus you can put up bigger numbers just by them having a crappy performance. I dunno if I'm really explaining my thoughts good on that one. It's like well even tho SOS is technically accounting for the difference in average strength it's maybe not accounting for the fact the weaker team can have absolute stinkers. And obviously by playing more of those teams, there's a greater chance of that.

But overall, I do feel like the metrics capture things well and typically the better metrics/poor resume teams probably do better in the tournament that the great resume/poor metrics teams.

 

[Aike]

I understand that. But when five of your key wins are against one team not currently in the Top 25. And three of the other four have a losing record in the SEC. It doesn’t scream Top 5 in any measurement today, tomorrow or next month.

Again, just a thought. You have much more expertise but I can think of ten teams that are just better than UT.

They also won at UL by 22 before UL was ranked.

They are undefeated on neutral courts, and their only home loss was to us.

Personally, I think it’s pretty great that we seem to have their number. And I have no trouble with the idea that there might be 10 teams more capable of winning a title.

I tend to see them as very solid, but not spectacular. One of the things I would love to do with this modeling is compute a range of scores. For example, If a team is a 3, for instance, do they range from 1.5 to 4.5, or 2.5 to 3.5?

I tend to believe Tennessee would fall in the second category. High floor, low ceiling. When you think of a champion, you tend to think of teams that are more like a 4.5 or 5 when everything is clicking. Explosive.

 

[The_Answer1313]

Exactly
That’s why all the Kenpom numbers, BPI, and many others mean basically nothing.

Sse I actually think just the opposite.

I think those numbers will tell you more in terms of future tournament success than the actual resume itself.

Tho I do want to be clear on this........I don't think resume shouldn't count for something. At the end of the day, this game is based on wins and losses, not a computer rank. If they want to give teams a bump up because they have nice resumes, that's fine.

Last year, St Johns had a better Kenpom rating than UK did. They did not even make the tournament. Not a great resume. Were they a tournament team based on skill level? Of course they were. Would they have won games in the tournament? I have no doubt of this. But ultimately you have to also have resume to back it up.

I'm just saying that if I want to predict who I think would advance further than their seedings would suggest, I'd probably look at the teams with good computer metrics first.

 

[Aike]

Sse I actually think just the opposite.

I think those numbers will tell you more in terms of future tournament success than the actual resume itself.

Tho I do want to be clear on this........I don't think resume shouldn't count for something. At the end of the day, this game is based on wins and losses, not a computer rank. If they want to give teams a bump up because they have nice resumes, that's fine.

Last year, St Johns had a better Kenpom rating than UK did. They did not even make the tournament. Not a great resume. Were they a tournament team based on skill level? Of course they were. Would they have won games in the tournament? I have no doubt of this. But ultimately you have to also have resume to back it up.

If I recall correctly, St John’s was something like 4th overall in my model over the last month of the season last year. That was a team that could have made a lot of noise.

 

[The_Answer1313]

If I recall correctly, St John’s was something like 4th overall in my model over the last month of the season last year. That was a team that could have made a lot of noise.

There was zero doubt in my mind that team was not only a tournament team, but one that could have made a run.

 

[The_Answer1313]

Having said all of this.....

For the most part, especially with power conference teams that have had opportunities for resume building wins, most people/computer metrics agree. At the top the resumes usually do in fact line up with the computer metrics.

Look at any human poll, any of these models........mostly everything lines up at the top. We can debate the likes of Gonzaga or Houston but only because they've lacked opportunities than say an Auburn or Alabama. But for the most part, everyone agrees at the end of the day on who most of the contending teams are in March.

 

[The_Answer1313]

To me, at least on Kenpom......the rankings aren't really all that interesting to me. Everyone knows Auburn is good. Everyone knows Alabama is good.

The team pages on Kenpom and Bartovik I find interesting tho. How a team plays, what they are good at, what they aren't good at.

There's a wealth of information that goes way beyond just an offensive eff and defensive eff number.

 

[The_Answer1313]

One thing tho I do think maybe metrics should factor in consistency more.

So the predictions are always based on what a team does "on average". On average we score this many points per poss. We give up this many points per possession.

I kind of wish they were factoring in the variance between games.

For example, say Team A plays 50 games. They score 1.2 ppp on average. Every game they score 1.2 for that average.
Team B plays 50 games. They also average 1.2 ppp on average. However, 25 of those games they scored 1.00 and the other 25 they scored 1.40.

I'd make the argument that Team A is more likely to win in a one and done tournament compared to Team B, despite them both avging the exact same thing. You score 1.00 in any game, you are probably going home. You score 1.20, you are probably moving on.

So I feel like consistency might be a good thing to look at. Obviously tho, you'd have to do it factoring in the schedule as well. If you are playing a lot of scrubs, it stands to reason you are going to have wild swings (put up a ton vs those teams and normal scoring when playing a decent team). Still, I feel it's something that might be getting overlooked a bit.