Bill Radjewski | CollegeFootballData.com

🏀 The March Madness Pack is live - $25. Clean historical tournament data + guided Jupyter notebooks for: • bracket models • matchup analysis • upset research Perfect timing now that the bracket is out. buff.ly/7stmFQL

Glad to support one of the best in the business and truly a great dude. This is sure to be an awesome way to continue to relive this past season time and again. If you're a fellow Michigan fan, this is a no brainer.

@solo_trader Sounds great. I look forward to seeing what you create.

These are schools guaranteed to *attract* NFL talent via high school and portal recruiting This graph has nothing to do with developing NFL talent. That would require a whole different set of variables, and should naturally include some sort of weighted metric for draft picks that transfer via the portal but were developed at other schools

There’s a clear exponential curve between roster talent and NFL Draft picks. Good teams produce draft picks. Elite teams flood the draft.

It’s hard to find a non-paywalled list of players in the transfer portal. Thankfully @CFB_Data has a great API at collegebasketballdata.com. Here are the top players in the portal (some pending eligibility) by my time-decayed RAPM. Jasper Floyd could be an underrated addition.

@bucknut2734 🤔

@CFB_Data But they didn’t win their Conference Championship. According to the dUMb fans, that makes it a Mickey Mouse Title. 🤷‍♂️ Amiright? 😝

Michigan didn’t just win the title… they ran through everyone. • Beat all 3 Final Four teams • 6 of 7 Elite Eight teams • 9 of 15 Sweet Sixteen teams This is an absurd tournament résumé.

@offlistcomment1 en.wikipedia.org/wiki/2015_Univ…

@CFB_Data What happened in 2013?

Want more historical deep dives like this? I’ve got full datasets for exploring runs like this 👇 CollegeBasketballData.com Relatively newer site if you're familiar with my CFB work. Planning to keep building and expanding. Let me know what tools you would like to see!

Beating 9 Sweet Sixteen teams is even rarer. Only other team to do it: • 1996 Kentucky Now it’s Kentucky… and Michigan.

Updated historical Elo chart for the newly crowned men's basketball National Champion, Michigan. Greatest Michigan team of all time? They have a strong argument.

@RajLakra en.wikipedia.org/wiki/Elo_ratin…

@CFB_Data For us neophytes, what is the ELO?

Season Elo Trends - National Title Game Edition Michigan has hovered near the top of Elo most of the year. UConn has consistently graded out as an elite team. Strong vs steady with it all on the line tonight.

@LogHanRatings Yes it is. Elo ratings for team sports typically consider MOV and these are no different. Michigan's MOV in this stretch was below the Elo expected MOV, hence the downturn.

@CFB_Data Methodological question: Is this actually Elo? Because from Jan 6-30, Michigan's Elo in your chart decreased persistently, while these were their results. One of Elo's defining traits is "wins always increase ratings". Shouldn't there only be 3 downward slopes for them?

Offense gets headlines. Defense wins in March. The Final Four teams? They do both and that’s why they’re still standing. More data: CollegeBasketballData.com

Shot distribution contrast in the Final Four: • Illinois: extreme 3PT reliance (50%) • Arizona: highest mid-range rate (32%) • Michigan: near-even rim/3 split • UConn: efficient modern profile Philosophy clash this weekend.

Well, this aged poorly for you.

@CycloneGrav Last I checked, they weren't a 1 seed and as the chart clearly states, only 1 seeds are circled.

@CFB_Data So... ISU has "Championship DNA" but you don't circle them as well?

The Sweet 16 breaks into tiers: 🏆 Championship DNA: Duke, Michigan, Arizona ⚠️ Dangerous but flawed: Alabama, Purdue, Illinois 🧊 Defense travels: Nebraska, Houston Balance still wins in March… usually. Who survives this weekend?

Seed Says Chalk. Elo Says Elite. With two 1-seeds, a 2-seed, and a 3-seed, this was already one of the chalkiest Final Fours of the seeding era. But by pre-tournament Elo, it was even stronger than that seed profile might suggest.