My current top 16…

[Aike]

Your model is intended to predict the National Champion based on the entire work of play, correct? In other words, with more game played, the algorithm would be more accurate using historical data as training data?

If this is true, do you have the ability to compare with historical data on the same date? On Dec 17, 2022, I'd assume the projections would look different than the projections on March 1, 2023?

Yes on every question you asked.

For example, Kentucky was 12th on December 19th last year, but had fallen to 24th by the time the tournament started.

 

[Aike]

Yeah, I get that. And it’s not something I care about personally. I’m just curious how an objective methodological model would put the team that seems to have accomplished more than anyone else so far (ie. Purdue) so low.

What objective factors are weighing against them in this model?

Already answered. Has to do with the way they play defense. Even still, they would clearly be one of the favorites if the tournament started today. And they are modeling much better than they did last year.

 

[Padsfs07]

Alabama has a tough schedule. They are playing Arizona this week now. Impressive scheduling, poor results thus far.

 

[Padsfs07]

They lock down defensively, and also rebound and shoot the ball fairly well. Also take pretty good care of the ball.

Speaking of Iowa State, former Lexington Christian kid, Tre King, who went to EKU is doing really well there.

 

[storm1507]

Thanks. I'll know to keep a better watch on them now.

Iowa St. I was keeping up with them last night. Thanks for the heads up.

 

[DerekMcPwn]

These are statistical ratings, not polls. I don’t switch them around every week like Pat Forde.

I continue to be amazed by the percentage of the population that does not understand how statistical models work. These threads always have a half dozen posts like “Hahaha [insert team] lost twice this week and KenPom/Torvik/Aike still has them ranked #2, he’s crazy!!”

PSA for the four hundred and seventeenth time: these models rank teams based on a formula and game data (total points, total possessions, etc.). The person who wrote the model doesn’t do anything to rank the teams except refresh the model with new game data.

 

[runt#69]

As usual, thank you greatly for the updated look.

I love watching AIke's models over time. And it's not just that UNC*** and Duke aren't anywhere near sniffing this. lol ... I've mentioned before it's not how the blue blood or other notable team show up, but the others that make this so interesting to me.

But now that I actually even bothered to mention them, I think I can see how teams like Duke and UNC*** may not be stacking up well in a cold hard, analytical comparison. Neither team seem to me to do anything exceptionally well or great, yeah, they're winning, but they just don't have that "eye test" look to me. UNC*** ain't bad at all though. And by season's end, either one may be rising up thru the system.

That's why I love watching this. How the Iowa States and colorado states do, FAU, etc. then compare how the blue bloods look.

Great take and much food for thought as we move forward.

Well done, Aike !

ps - Iowa state making your model and other computer models look real good ! Helped me to put them on the radar this year! Playing in Hilton Coliseum is a brutal, rough place to go in and play for Big12 teams. A good analogy to me is Miserable Gym in Nashville.

 

[ukdesi]

I'm so using this model for my work March madness pools. It was good last year.

Purdue continually proves they are susceptible to upsets in the NCAA tourney, so despite all of their wins, this ranking system is impressive in that it actually ranks Purdue lower than the top 3.

 

[Ajax10]

Is there a link to his current ranking model?