The 2 line is getting awfully congested. I think we're big Purdue fans tonight—if Nebraska loses they still wouldn't have a loss as bad as Cincinnati, but Iowa State would have a win over a common opponent. I've been beating this drum for weeks, but after last night you have to beat Kansas. can't drop TCU, Utah, or Arizona State. And as others have implied, you have to at least go 2-2 against Houston, BYU, Tech, and Arizona
2025-2026 MBB computer projections thread
- Thread starter Sigmapolis
- Start date
Assuming we beat TCU, Saturday against KU is going to be ridiculous. getting that W is huge and could afford a loss to UH 2 days later.
Last edited:
Have there ever been a better two games in three days in Hilton than KU and UH coming up?
Last edited:
Went down the rabbit hole on this one. Was curious on game by game performance of the last 5 NC's.
Conclusion: Yes if you take it literally about every single game you can't perform you best that is true. But results show that you need to play "perfect" or near perfect for 6 games in a row to win the NC. Even then it may depend on a little luck. See the close game between Fla and Hou in the NC, Fla got a 98 game score and won by 2.
Also, note that the KU team had the widest deviation of game score with the poorest performance in the NC and still won. Not really that surprising.
Forida '25
Uconn '24
Uconn '23
Kansas '22
Baylor '21
Attachments
Cyclonepride
Thought Police
Interesting, but I think taking tournament data from the tournament champions doesn't provide great support for a regular season argument since they obviously performed well in the tournament because they won it. If you look at their regular seasons, you will almost certainly find a few games where they just barely squeaked by when they should have won easily.
A 96 game score for Florida against Texas Tech shocks me. They were down 9 points with 3 minutes to go. I would say that's not perfect or anywhere close to near perfect. It took an incredible finish for them to win that game. Wondering how Torvik got 96 out of that.
Sig couple probably explain it better but I think the Torvik game score is not based on what we would rate a game through observation. It weighs your metrics for that game against other teams then that game score is the percentage of teams you would have beaten. So 96 would mean that performance would have beaten 96% of the teams.
I searched it and can't find it. Does anyone know if the NCAA is having an early bracket reveal (top16 seeds) in the next week or two?
Metrics aside, would you trade a loss @TCU for a win Saturday against KU?
A loss at TCU for wins against both KU and UH?
I would the latter, just for the eye ball resume.
I saw that it will be February 21st this year which seems later than usual.
I would say there are 12 teams in the running for a 2 seed or better.
Arizona
Michigan
Duke
UConn
Houston
Iowa St.
Illinois
Kansas
Nebraska
Purdue
Michigan St.
Florida
Its hard to see anyone outside that making it up to a 2 seed, but a handful that could replace someone on this list for a 3 seed.
Arizona
Michigan
Duke
UConn
Houston
Iowa St.
Illinois
Kansas
Nebraska
Purdue
Michigan St.
Florida
Its hard to see anyone outside that making it up to a 2 seed, but a handful that could replace someone on this list for a 3 seed.
Bingo.
365 teams in Division I
Game Score = Given how you played (as measured by your margin per possession and the analytical quality of your opponent with a small and consistent adjustment for home/away)... how many of the 365 would you beat on that night if they played their average game against you on a neutral court?
Game Score = 93 * 365 = you played about as well as the #25 or #26 team in the country
It has NOTHING to do with the quality of any traditional counting or advanced metrics.
Shooting percentage?
Assist share?
Assist-to-turnover ratio?
Rebounding metrics?
Fouls?
None of any of those measurements matter.
So a 96 against Tech for Florida last year only cares about the final score. These models don't consider the points scored in a mad and miracle comeback at the end of the game as somehow "worse" than the points scored in a "regular" possession with 14:00 left in the first half. Points are points (or rather more accurately margin of victory is margin of victory) no matter when or how you managed to deliver them.
So, the model isn't judging Florida for pulling a rabbit out of its hat (good or bad).
It's saying "beat a good Tech team 84-79 over 71 possessions that's a top ~25ish performance. Good but not great, the #4 team should have beaten Tech by more but a win is a win."
The game score is a simple calculation based on objective criteria.
Not some sort of eye test.
Is being a 2 instead of a 3 much of a huge statistical difference in the challenge level and/or tourney success?
Also, note how 7 of those 12 are midwest teams. Not bad for maybe 25% of the population.
Hell the corn belt has 5 teams!
I was curious about this as well, so I asked Co-pilot (full disclosure). 2 seeds have twice the chance of making a final four compared to 3 seeds. Another way to look at it: 2 seeds are 32-19 in head to head matchups versus 3 seeds.
Using more recent years only, that advantage is 2 seeds are 1.7X more likely to make the elite 8.
I didn't fact check any of this, all per co-pilot.
Last edited:
From 1985 through last year 32 two seeds have made the final four and 17 three seeds have made the final four (66 one seeds for comparison). Interestingly 12 of those two seeds and 12 of those three seeds made the championship game with 5 two seeds winning it all and 4 three seeds winning it all.
I’m sure there is a big drop from 3 seed to 4 seed as far as tournament success goes.
I found this after asking my question. https://adamcwisports.blogspot.com/search/label/G-Score
I crossed up Torvik's G-Score metric with his Game Script metric. I was thinking Game Script when I read G-Score. Florida's game script in that game against Tech came in at -2. They were -0.1 against UConn, -1.6 against Auburn, and -3.6 against Houston. It really is impressive how many 2nd half comebacks they had in their tourney run, which is why I was relating it to a "teams don't have to be perfect in every NCAAT game to win the whole thing" idea.
I looked it up last season and I think ~40% of 1 seeds have made the Final Four. ~20% of 2 seeds have. I think the percentage drops by half again, with ~10% of 3 seeds making it. From there, while the percentage-change with each seed isn't as large, you're of course talking less than 10%. I'm sure we'll have this discussion again in two weeks, but you're basically hoping against all odds if you're not a 1 or 2, and even a 2 vs. a 1 makes a difference
Records for every seed in March Madness from 1985 to 2025
These are the all-time, round-by-round records for every seed in the March Madness men's tournament from 1985 (when the field expanded) to 2025.
www.ncaa.com
