For those of you new here, Week 1 will give you a full breakdown of what we’re doing. Current bracketology HERE. Most recent Bubble Watch HERE.

This week I am taking on the notion that diagonal lines (by Net efficiency) would be a better way to bucket teams in order to try and find their reliability of advancing in the tournament, which is the goal here. I have below my effort to bucket the same data set of teams into 8 archetypes in the same manner that I originally did, with a similar representative sample size in each archetype. The general idea of critics has been that the overall net efficiency would be more predictive than the archetypes I have “arbitrarily generated to denigrate X team”. What I believe is shown here is that my archetypes do a better job at separating out the reliable from the unreliable. Take a look at the drop-off from the 4th to the 5th bucket, which for me was my final reliable archetype to the unreliable. When you look purely at net, the drop-off is minimal. For mine, it’s around 30 percentage points. It’s not until you get to the 7th or 8th tranche that you find the true unreliable teams, maybe the 6th beyond the 1st round. You also have the obvious distinction in mine where F4 reliability is > 20% for all 4 reliable archetypes, while the unreliable are all < 4%. That is much more clear than NET. What my goal was initially was to find the teams most likely to get upset, or fail to go as far as their seed might suggest. I believe my archetypes do a better job at predicting that than straight net efficiency, and while you may find a specific team this year that may have more or less reliability in NET than in mine, I think in the aggregate being able to separate reliable from unreliable is much easier with mine than straight net, based on historical results. Now at least, the proof is spelled out, and you no longer have to just take my word for it.

Week 9 Thoughts:

Not only is the Juggernaut Watch back but we now have 2 teams crossing the historical threshold of 100% reliability to the Final 4. Never have we seen a team with NET++ > 50 miss the Final 4 as we’re 5 for 5 since 2002. We now have Arizona and Michigan who have crossed that critical threshold, and while I just spit on Net++ in general as a way to analyze, 5/5 to the Final 4 is too notable not to pay attention. Other notables would have to start with BYU, as we’ve seen them fall from borderline Elite to Matador, one of the most dramatic shifts from any team on the season. They have crumbled on the defensive end, and while Dybantsa can clearly carry the load offensively, they gotta get back to guarding. Then we have Virginia, who have been all over this chart since we began back in December, as they’ve gone from Matador – Great with their defense getting better, then back to almost Matador and now the offense has fallen off and they’re barely hanging on to Solid on the other end of the spectrum. What a wild ride, but all along the way they’re 20-3, so hard to argue with the results thus far. Also shout out Clemson, still the only grinder, representing a bit of a throwback style winning with defense. Finally, Illinois, who remains our NET outlier who people are calling to be elite but are lacking the defense to be reliable, which as we showed earlier is a better predictor of March success. We are getting closer and closer to our final picture heading into the tournament, but teams continue to rise and fall along the way, and I for one am enjoying the ride.