The framework of âhow many points is he anticipated to score in each gameâ multiplied by âhow many games is he anticipated to playâ seems good, so I would just try to improve how you get to each of those factors. T-Rank or KenPom has you covered for win probabilities so I would just use that to calculate expected games played. To refine expected points-per-game, you could add in anticipated opponentsâ defensive efficiency rankings as well as anticipated possession counts in the matchups. The easiest is just doing the first matchup since itâs known; if you want to get fancy, you could calculate an expected opponent defensive efficiency for the tournament run weighted by the probability of facing that opponent. Thatâs probably a little extra though, lol.
I tried to add in anticipated opponent pace averaged with a given playerâs teamâs pace last year, but once you weighted the probabilities the effects of potential opponents ended up being super minimal. The teams expected to play a lot of games have so much uncertainty about who they will play by game 3 or 4 that it doesnât end up mattering much. Also it was a massive pain in the ass (the play-in games were especially difficult to make work because we draft while those games are on)
Will probably stick with Torvik advancement probabilities (used to use 538 but RIP) and PPG. Might do a slight adjustment to bias the PPG towards last 5 or 10 games played, but simple is probably best. Last year my big winner after Zach Edey in the first was Jared McCain who I took late because he had been lighting it up recently and that did end up continuing in the tournament.
@jazznutUVA I just built out a google sheet based off of the RealGM site that in theory should autoimport those tables and then I can just run lookups on them from there. I have it working for 2024 at least, weâll see what technical horrors await when the tournament starts. Thanks.
They update it pretty quickly⌠even the Euro basketball leagues⌠so Iâd assume the NCAA tournament should be even faster.
A brief note as we discuss the portal and now that the Miya rankings are the default in LRA article: When you see a big difference between Miyaâs predicted BPR for next season and what the BPR was for this past season, itâs related in part to how he accounts for team effects. Hereâs the plot of the difference between a playerâs predicted 2026 BPR and his most recent BPR (positive number = predicted 2026 BPR is higher than most recent BPR) against his most recent teamâs adjusted efficiency margin:
Basically, the guys who have bigger predicted improvements in BPR come from worse teams and the guys who have the biggest predicted decreases in BPR come from the better teams.
Evan Miya has a post out on roster construction: How To Build A CBB Roster In 2025 - by Evan Miyakawa
The headlines are:
1. Fill your roster with good basketball players
2. Returning players should account for at least 50% of the playing time
3. Prioritize recruiting players who will play at least 2 seasons for you
He acknowledges that #1 is obvious and not really worth discussing in much detail so he really just goes into detail on the other points.
Has some interesting charts, of note to us is the chart for the âflipped rostersâ teams:
The results are pretty disappointing for this group. Only 6 teams out of the 20 made the tournament, and only three advanced past the first round. Almost half of the group finished outside the Top 100 at EvanMiya.com. Of the six high-majors this past season who had entirely new rosters, only Kentucky and Louisville made the tournament. It takes an excellent coaching staff to get the most out of a roster comprised entirely of new faces.
Good read overall. I have some minor quibbles with some analytic decisions (e.g. he loves making continuous variables into discrete variables), but I like all the charts and tables.
From our perspective, would be nice to see a deeper dive into what did and didnât make the flipped rosters work. But thatâs a lot of workâŚ
On the talent question, it looks like 6 of 8 teams who had top 50 preseason teams made the tourney. The two that didnât are LSU and Memphis, and Iâd argue that their coaches are ⌠meh.
For season 1, if our goal is make the tournament, getting into top 50 preseason with our talent will be important.
For season 2, it seems like we wonât be hitting Miyaâs other two factors no matter how our final portal additions this season shake out this year, too many one year guys. Heading into season 2 is where Iâd want to see Odom try to target some good high schoolers, sophomores and juniors using his first season momentum. That paves the way for year 3 making a splash hitting all three factors.
It probably wonât be as buzzy as the transfer haul, but the live weekends and HS recruiting in general may be more impactful over the longer haul.
One takeaway I had from looking at the flipped rosters chart is that Penny is something elseâŚappearing on this chart two seasons in a row and as a not-new coach is impressive in a way. Runner-up award to Madsen at Cal for taking two bites at the total flip apple and mostly getting worms.
The (way-too-early) T-Rank projections for 2025-26 are out: 2025-26 Projections - - Customizable College Basketball Tempo Free Stats - T-Rank
They will change (a lot) as teams finalize their rosters, but I do find it useful as a place where % of returning minutes is kept visible.
Much better than I wouldâve guessed with no Ugo/GrĂźnloh included yet. (65)
@haney so you can look super smart once we pick up Grßnloh⌠when you spell his name and cannot type the umlaut, you put an E after the U⌠so Gruenloh. German crossword puzzles and website names (among other things) do not use umlauts⌠they use the E after the letter for the same thing.
WIth out the E and no umlaut⌠it changes the pronunciation to Gr-UN-low
I fuexid it (pronounced like the ue in baseball player Bill Mueller)
Do you pronounce it MOO-ler? That is correct German. MULE-er is not.
For some reason, he pronounced it Miller. Maybe an Americanization
That is what Mueller means in German⌠or can be farmer too⌠so cool as well.
And thatâs with walk ons getting meaningful minutes.
If you remove the walk ons and add Onyenso and Malik Thomas, we jump to 43rd.
Itâs all super noisy at this point but generally points to us being top 50 if we add a replacement level P5 front court.
Curious what would happen if we also threw in Gruenloh
Yep, but doubt we will get high quality answers. I recall last season Torvik not knowing how to deal with euros pre season. Too much complexity and not common enough to deal with euro transfers, but maybe that changes this year.