Franketology: January 29, 2025
There are many bracketologists out there far more knowledgeable and experienced than I am. I started last season. But I’m a quick study, and my ADHD hyperfocus and love of spreadsheets and data has made me a fairly quick study. On the same token, I do not come with years of ingrained practice and I am entirely self-taught. Last year I literally said “I want to do bracketology,” and then figured out how to best go about it on my own. I do not think I have done anything unique or original. I pull all the data (metrics, records, SOS, etc.) into a spread sheet, and start eliminating teams from there. Easy.
My take on evaluating teams is also informed by last season. Infamously, the Big East only got 3 bids last year, to the chagrin of many. St. John’s fans like myself, as well as Providence and Seton Hall fans were universal in our anger at the committee. Many fans of all 3 fan bases pointed out how stupid the metrics were. The gripes certainly seemed legitimate…
St. John’s was 32nd in NET, and 4-10 (.286) in Q1 games, but watched the team ranked 54th in NET, with a 2-7 (.222) Q1 record—Virginia—get the bid. Seton Hall and Providence had 5-8 (.385)and 6-9 (.400) Q1 records, respectively, but also had to watch UVa get in over them. Providence also had 0 losses in Qs 3&4. PC’s 6 wins tied them for 14th among all D1 teams. Seton Hall’s 5 tied them for 24th.
Every. Other. Team. That had 5 or more Q1 wins made the NCAAT except for Seton Hall and ProvidencePlenty of teams got in with 4 or less, and some teams even got in with only 3 Q1 wins! That latter category included Auburn, 3-7 (.300), ranked 5th in NET, and got a 4-seed. Gonzaga, 3-6 (.333), ranked 17 in NET, got a 6-seed. Dayton, 3-4 (.429), ranked 23rd in NET, got a 7-seed. Michigan St., 3-9 (.250) was 24th in NET and got a 9-seed. FAU, 2-2 (.500), ranked 39th in NET, and got a 10-seed.
If you’re an astute observer, you’ll notice that not only did Hall and Providence have more Q1 wins than all the listed teams, but they also had higher winning percentages than: Michigan St., UVa, Auburn and Gonzaga.
So what went wrong? Seton Hall and Providence both had weak NET rankings: 67 & 58, respectively. But NET isn’t supposed to matter. It certainly didn’t for Virginia, who was significantly worse in Q1 than PC or Hall, and made it with a mere 54 NET. Northwestern, 53rd in NET with only 4 Q1 wins (4-7, .364) had fewer wins and a lower win % in Q1 than either Hall or PC, but got a 9-seed.
Hall, PC, and SJU all had fairly weak Q1&2 combined records: SJU 10-12 (.455), Hall 9-11 (.450), and PC 9-13 (.409). Once again, that couldn’t be the factor. Texas was 8-11 (.421) in Q1&2, with 5 Q1 Ws and still got a 7-seed, Miss St. 8-12 (.400), with 4 Q1 wins got an 8-seed, and TCU, also 8-12 (.400) with 5 Q1 wins got a 9 seed.
Indeed, if you line Hall’s resume up against Texas—a 7-seed—and remove the metrics, they’re virtually indistinguishable:
20-12/20-12
Q1: 5-8/5-9
Q2: 4-3/3-2
Q1&2: 9-11/8-11
Q3: 2-1/5-1
Q4: 9-0/7-0
Left those team name-less to prove a point. The first is Seton Hall if you hadn’t figured it out. Thanks to metrics, those incredibly similar resumes are rated 27 spots apart: Seton Hall as 2nd team out was 47th in “true seed,” while Texas, a 7-seed, is 27th. So let’s look at the metrics, aside from NET, since we know NET is a sorting tool, not a ranking tool:
SOS: 33/17
OOCSOS: 67/36
At least we see some separation here, and in a less opaque metric. The slight SOS disparity is supported by their quad records, with Hall having only 1 more Q1&2 game, but 2 more Q4 games. However, the disparity is might be even smaller than it appears, as Texas’ average opponent was 103 NET to 112 for Hall’s average opponent. While there’s a disparity here, I do not think it’s one that could account for a 20-true seed difference.
Let’s move on to the resume metrics:
SOR: 41/31
KPI: 66/47
Avg: 53.5/39
If you’ve followed me in the past, you’ll know I have major problems with KPI. It’s a deeply flawed metric. For these pages, let’s suffice to ask the rhetorical questions why did the NCAA rely so heavily on a metric that varies so greatly from what the NCAA has identified as a quality win? Why is the NCAA leaning on a metric, as it did last year, that can take 2 resumes the NCAA views very similarly—NCAA created the NET to rank and sort wins and losses into the now-familiar quads, presumably they would view similar quad records as equivalent—and ranks them 20 true seed lines apart? KPI was the most determinative individual metric last year, and it’s crystallized here with this Texas comparison. It was also an outlier. Hall’s SOR, the other resume metric on team sheets last year, was a 41, versus a 66 in KPI. Bart Torvik’s highly respected WAB
Predictive metrics aren’t really meant to be used for determining bids, but it’s clear the committee will boost a team with poor resume metrics if it has elite predictive metrics. Texas had a 25 predictive metrics average versus a 65.50 for Hall. We know KPI is the likely outlier because the only other available resume metric—as far as this blogger knows—is Torvik’s WAB. While it wasn’t on the team sheet last year (the NCAA’s own version appears on team sheets this year), WAB had Seton Hall ranked 43. That brings the resume average down to a 50, but obviously if you remove the flawed KPI, you have a resume average of 42. Similarly, Providence had a WAB of 45 per Torvik, versus a KPI of 69 (not nice lol) and SOR of 44. Once again, WAB proving that KPI is an outlier here.
A resume average of 42 would put them squarely in line with teams like TCU (44) and Michigan St. (47), both of whom it had a better Q1 record than, and it has a better Q2 record than Michigan St. So in sum, what really did the Hall in was KPI. I’m not saying that’s right, it’s just the facts. In my campaign against KPI I learned a lot about the metric, but also the shadowy, opaqueness behind it.
For starters, KPI stand for Kevin Pauga Index. Kevin Pauga created the KPI essentially to replace and improve upon the RPI as a results-based metric. Of course you don’t know who Kevin Pauga is, but he’s an assistant AD for a division 1 school…but not just any school…he’s an assistant AD for:
Michigan St!
Do you believe that shit?! The guy controlling the metric that pushed Michigan St. to a 9-seed over otherwise more-deserving Big East squads is controlled by the Michigan St. assistant AD.
Not only that, but he is tied in deep with the NCAA, and his metric is used for countless other sports and divisions. Don’t believe me? Go check out FaktorSports.com and look at the drop down lists for divisions and sports. I actually had a fellow CBB/FCS fan tag me in a post in the fall about how FCS was going to be relying on KPI for selections to its post-season tournament. An absolute travesty.
BUT WAIT, THERE’S MORE! Not only does Kevin Pauga control this incredibly important metric, he then goes out—as an incredibly lucrative side-hustle from his AD responsibilities—and sells those same schools that he controls the rankings for his services as a “scheduling consultant.”
So just think of this racket:
create a black box metric with only the vaguest possible public descriptions
get in good with the NCAA so that said metric is used across many sports
get 5-figure schedule consulting services contracts from various schools, presumably to assist in gaming the metric you created (and it is VERY game-able)
PROFIT!!
Good work if you can get it! So why do I bring this up? Do I think Kevin Pauga is sitting around tweaking the KPI rankings on the back-end? There were certainly plenty of anomalies that I profiled extensively in my Tweets last March. But no, I do not think he’s doing that, but I can’t rule it out in the absence of transparency. Do I think he’s selling consulting services to teach schools how to game his own metric? Yes, yes I do. But this could easily be rebutted out with a bit of transparency.
That is the key here: metrics transparency. While I understand the importance of a secret, proprietary formula to these metrics guys, we need some type of transparency around these metrics. For all I know, the NCAA is already engaging in the type of audits, checks, and balances that would give folks confidence in the team sheet metrics. If they do that, they need to be more open and transparent about it to give the rest of us the comfort they have with these metrics.
Of course this is the NCAA, so there’s an incredibly good chance they don’t have any checks and balances, let alone audits. We simply don’t know. At the end of the day, the only way to restore fan faith in the selection process is greater transparency.
The other reason I raised this is that last year was an excellent year to start bracketology. It threw into stark focus, in my first year as a bracketologist, the importance of resume metrics, and the basic axiom that resume metrics help determine who gets bids, while predictive metrics help determine seeding. While I believe most serious bracketologists know this (in case you can’t tell I am NOT a serious bracketologist), it is a fact that’s lost on the general public. Inevitably some team with solid KenPom/Torvik/NET figures will be left out this year (looking at you BYU), and inevitably those fans will whine about how metrics are irrelevant. Or fans will wonder how did THAT team get in because their NET/KenPom/Torvik figures are too high. They’ll be in good company, as hall of famer Rick Pitino tweeted similar sentiments earlier this year about NET and KenPom.
When you see those complaints, now you know better. You know to focus on the resume metrics to see why that team made or missed the NCAAT.
tl;dr - resume metrics are more important than NET or KenPom for inclusion in the tournament
With that out of the way, let’s get to the current Franketology bracket:
Direct link to spreadsheet here.
True Seed List
Auburn
Duke
Iowa St.
Alabama
Houston
Tennessee
Purdue
Kentucky
Florida
Marquette
Kansas
Texas A&M
Arizona
Illinois
Oregon
Michigan St.
Wisconsin
Mississippi St.
Ole Miss
Texas Tech
Michigan
Maryland
Gonzaga
St. John’s
Louisville
UCLA
Missouri
Utah St.
Memphis
West Virginia
Baylor
Creighton
UConn
Clemson
Texas
Ohio St.
Georgia
Last 4 Byes
Pittsburgh
Oklahoma
Vanderbilt
Arizona St.
Last Four In
New Mexico
UCF
SDSU
Wake Forest
Other Auto-Bids
UC San Diego
Arkansas St.
VCU
McNeese St.
Bradley
Liberty
GCU
NDSU
Samford
Yale
Lipscomb
High Point
Akron
UNCW
Northern Colorado
Milwaukee
Norfolk St.
Bryant
Merrimack
Southern U.
Quinnipiac
Little Rock
Colgate
FIRST FOUR OUT
Drake
Indiana
UNC
Nebraska
NEXT FOUR OUT
St. Mary’s (CA)
Northwestern
Xavier
BYU