HoH Bracketology

Jeff is really good but hate the roster management. Cycling in 30+ transfers isn't sustainable. Bill eventually comes due and you just miss on everyone (ie our OL). Never hear about young guys pushing for snaps. Going to have to take 35 more again bc that's the only option without a huge reset year

if the pacers were in the CFP without haliburton they would be removed from the game at half

Take it from someone who tried to predict the bracket mostly by making a statistical model

OKC fouling up 3 there is bad data analysis. With no context foul up 3 is the right move, but Denver has the ball 90 feet away needing a 3 without a TO or Jokic on the floor. Play it out

It's crazy how much they emphasized result metrics, as NET, KPI, SOR, and WAB were all at a 73%+ correlation, matching or surpassing their highs from the previous three years

Torvik, which Bubba Cunningham said was used heavily, had one of the lowest correlations among all stats

Completely bygone topic now that we're in the sweet sixteen, but the official bracket emphasized WAB *more* than the mid year reveal. WAB had a 90.7% correlation with ranking within the seed list, crushing every metric and quad stat over the past few years by no less than 7%

We have metrics and data that tell us how good a team is over the course of the season instead of solely emphasizing a subset of a subset of games. Let's use them.

Which leads to ridiculous situations like a top 10 team in the NET being an 8 seed (Gonzaga) and the number 49 team being a 5 seed that's an underdog in round 1 (Memphis). The NCAA came up w a metric to rank teams over a season and only uses it to categorize win quads...

With that being said, the quad system is terrible and arbitrary and tends to suck all the air out of the room when debating these teams, and gets even more ridiculous when you realize the committee came up with the NET but only really uses it to decide what quad each win is in

West Virginia still absolutely got hosed relative to all of San Diego St, Xavier, Texas, and UNC getting in over them, and UNC winning by a billion tonight doesn't necessarily change that

We all need to take a lesson from North Carolina tonight: metrics matter, a lot.

I tweeted about this a while back, but UNC's metrics are great relative to the other bubble teams. They were crushing teams down the stretch and it was reflected in the metrics, which got them in

Year 1's entry finished 19th out of about 250 entries. Chalking that one up as a win

96.7% of the bracket within one seed line, 53 exact seeds, 11 within 1 seed line, miss on UNC and Xavier being in over WVU and Indiana (Xavier is costly)

Louisville and Memphis are the two whiffs seeding teams in the field (went against my own advice with Memphis)

Interested to see where this ends up landing. Some real shockers like UM as a 5, UNC in, IU and WVU out, Louisville an 8, Memphis a 5

Formula whiffed on a couple teams it liked like UC SD and VCU, absolutely nailed the one bid teams though

All in all think this was a good first try

Last 10 games was dumb and could be heavily influenced by SOS being backloaded, but we have plenty of ways to account for how a team has played adjusting for strength of schedule. Let's use them

I think we need a metric that adjusts to how a team is playing at the end of the season vs the beginning. Torvik lets you adjust date parameters, and you can see how a team rates from say, since 1/1/25 rather than since 11/1/24. Teams aren't the same as they are in November (see: Oklahoma)

The only explanation I can really give is Louisville's fate was set getting blown out by Tennessee and Ole Miss at the very beginning of the year- we know beyond all doubt they aren't that team now, but if UofL dropped 1-2 Q3 games the committee might've sent them to Dayton having won like 19 of 23

Tbf, no metric had louisville as an 8

Clemson at 5 will get gripes locally but Memphis at a 5 explains what happened to Louisville better

Louisville played their way in by winning 21 of 23 (!) games. Memphis had 4 great NC wins and was dreadful in conference

Crazy emphasis on the NCSOS in Nov/Dec

WVU had 3 more Q1A wins than UNC had Q1 wins, including Gonzaga and Arizona neutral, Iowa state at home, and at Kansas

North Carolina beat ucla and lost donkey digit Q1 games

Think we can all see what happened here

Here is Louisville’s analytical profile via Torvik, including NET, KPI, SOR, WAB, Results Average

Then BPI, KenPom, Torvik, and the predictive average

Lexington de facto home game is nice

However there were zero metrics on the team sheet that suggested Louisville is an 8. Every single one was above 28, and results based metrics were mostly in the 4-5 range

Committee just ignored *everything* but Q1 wins

For fun, predicting local teams

Indiana will be a 12 and play Texas in Dayton before being sent to play BYU on Thursday

Louisville will be a 6 and play UC San Diego round 1

Kentucky will be a 3 and play Troy round 1

One final bracket for the calm before the storm. March Madness is less than 30 minutes away, and had a ton of fun building the model and following college basketball. Enjoy the selection show!

Georgetown paid 10s of millions of dollars to hire Ed Cooley and not Rick Pitino.

The bubble now comes down to what argument the committee arbitrarily decides it likes most Sunday

IU, Xavier, Texas, Boise, UNC, CO St, maybe VCU (if they don’t win the a10) would vie for 3 spots

I think sdsu is probably in but could add them into the mix for a fourth spot

Boise state may get an at large on their own after winning tonight, but a Broncos-Colorado State final is a nightmare for bubble teams

Boise very well may have already bumped someone out and CSU with a win definitively will pop someone’s bubble

ACC officials doing the thing where they call absolutely nothing for a half and then call everything after the break

This feels like The End™️ for North Carolina ngl

Went ahead and threw Dayton and UC Irvine on the board for the hell of it, and they do better than I thought in the formula! Just behind Boise for both of them

All in all this is my best guess for where things are right now. Still a lot to sort out in the next two days with a few comparisons still being total coin flips

With Texas winning, UNC winning, Boise winning, and Colorado St still lurking out there as a potential bubble run team/bid thief, most results have broken against the Hoosiers and Musketeers.