After a fun (to me, at least… I hope you enjoyed it as much as I did) historical jaunt through past tournaments, it’s now time to do some analysis of the Madness Factor and see how it can aid our quest for a perfect bracket. We’ve already hit March, after all, and conference tournaments are either in full swing or about to begin.
First, a reminder:
Madness Factor: the total seed differential of all the upsets in a given data set (team, bracket, round, tournament)
Here’s a look at the Madness Factor of all the tournaments in the 64+ team era, along with a line marking the average:
The average Madness Factor for a men’s tournament comes in at just higher than 81. We’ve already learned that there are just under 18 upsets annually, on average, meaning that the average seed differential for an NCAA Tournament upset is 4.56. Perhaps this helps explain why 11-seeds are so successful, the 6-11 first round matchup sits right on that average.
Nineteen tournaments have had Madness Factors above the average; twenty below. But there is a clear recency trend. We’ve already pointed to 2010 as a demarcation year, after which the tournament has become even more volatile. The average Madness Factor from 2010-2024 is 93.8, a full 12 points higher than the cumulative average. That’s an extra bracket buster, or 2 additional major upsets—not insignificant. The three maddest tournaments of all time have been within this timespan too.
At a round-by-round level, the first round, unsurprisingly, provides about half the madness of the average tournament (a first round average Madness Factor of 42). The madness reduces at a fairly predictable rate as each round progresses:
· Second round average: 23.5
· Sweet 16 average: 8.5
· Elite 8 average: 4.8
· Final 4 average: 1.8
However, when we split the bracket into its regional parts, we don’t find the same evenness:
What this chart illustrates (somewhat confusingly, sorry) is that madness tends to accumulate within individual brackets. In 20 of the 39 tournaments contested, two of the four brackets had Madness Factors at least 7 points above the other two; so it often plays out that two brackets have significantly more chaos than the other two. Sometimes it works out that one, or three, get disproportionate amounts of madness. But the closest we’ve gotten to an even spread is 2019, when the 4 brackets had Madness Factors of 21, 21, 14, and 12. Madness begets more madness, I guess, is the takeaway.
It should also be noted that only once, in 1985, has a regional bracket scored a Madness Factor of zero. That means that, out of 156 regional brackets, only 1 has gone without any sort of upset. One. That seems impossible, but true. And it gives a sneak peek at an upcoming parameter: Never pick an upset-free bracket.
What about a correlation between Madness Factor and the number of total upsets? It would seem natural: the more upsets, the more madness. This chart shows us:
On the whole there is a clear correlation. But there are some outliers, particularly in recent years. The number of upsets over the past six tournaments has been remarkably steady. Three times we’ve seen 19 upsets, twice we’ve had 20, and one year with 21. Madness Factors have had a far broader range, from a low of 62 in 2019 to a high of more than twice that number, 128, in 2021. And that wild 2021 event, the maddest ever seen, had one fewer upset.
The only other stretch on the graph with a noticeable irregularity is between 1991 and 1998. The Madness Factors for those years stayed in a tight range of 53 to 74, while the total number of upsets jumped around a bit, from a low of 12 to a high of 18.
Why? Who knows. But once again we can see that the level of unpredictability is on the rise.
We’ll look at one last chart before bringing things in for a landing:
This is somewhat tangential from the Madness Factor statistic, but I’ve got to fit it in somewhere. This chart visualizes when certain seeding milestones were set and how long they’ve lasted; the highest seed number to reach the second round, sweet 16, and so on.
The first two tournaments of the modern era set the bar high. Villanova’s 1985 Cinderella run set the 8-seed as the standard to beat for highest seed to reach the Elite 8, National Semifinal, Championship Game, and National Champion. The latter two—finalist and champion—have still not been eclipsed. The ’85 tournament also saw a 13 reach the second round (Navy) and a 12 make the sweet 16 (Kentucky). Those two marks fell a year later during the chaos of 1986, when 14-seed Cleveland State broke through to the sweet 16. That same year, LSU made the 11 seed line the standard for the elite 8 and final 4.
Each of those marks that were established by the end of the ’86 tournament have proven remarkably durable. Dominoes fell in 1991 (a 15 seed, Richmond, to the second round) and 2002 (a 12-seed, Missouri, reaching the Elite 8), but it hasn’t been until the last decade-ish that new milestones were set. The Dunk City Florida Gulf Coast team, UMBC, and Saint Peter’s broke the glass ceilings for the Elite 8, second round, and Elite 8, again, respectively.
I point this out because one variable my parameters won’t account for is when another milestone domino will fall. The Perfect Bracket Parameters will get us in the right ballpark, where a perfect bracket is merely stratospherically unlikely, as opposed to cosmically unlikely. But as we pursue this impossible dream together, some of us will need to go out on a limb and predict something that hasn’t yet happened. Because the increasing level of madness characteristic of the last few years has shown no signs of slowing down, and more of these milestone dominoes will certainly fall. Here’s my educated analysis of which:
· Sweet 16: 16-seed advancing… unlikely; even though two 16’s have won first round games in recent years, there is still little sense in predicting even that to happen again, and they’ve been quite overmatched in round 2.
· Elite 8: 16-seed advancing… extremely unlikely; for the same reason times two.
· Final 4: 12-seed advancing… moderate; if four 11-seeds have made it this far, can a 12 be that far behind? Then again, the 12 line is typically the transition point from the final at-large bids (often from power conference teams with power conference rosters) and the best conference winners from one-bid leagues, often with less roster depth and in-season tempering than power conference teams.
· Finalist: 9-seed advancing… likely; this would be the next domino I’d expect to go down. Nine teams seeded 9 or higher have made the final 4; don’t be surprised if the 10th sticks around for more than a game.
· Champion: 9-seed or higher… unlikely; 30 of the 39 champions in the modern era were seeded 1 or 2, then the bottom falls out. Villanova’s ’85 win was truly remarkable, and it would be shocking if it were even duplicated, let alone eclipsed.
Now, about those Parameters. Because the Madness Factor is directly tied to how many and what kind of upsets there are, I only have two parameters directly tied to it. But one other emerged from our exploration that we’ll reveal here:
PARAMETER 4: Aim for a Madness Factor between 70 and 125
I should note that 14 tournaments have had Madness Factors below 70, but only 3 of these have happened after 2010. And we have one outlier above our range, the bonkers 2021 tournament, with 128. This 70-to-125 range ought to cover us, but feel free to scribble a bit outside the lines.
PARAMETER 5: Don’t spread the Madness evenly across all 4 regional brackets
You’ll almost always see one bracket that is a madness magnet and one that is a madness repellent; the other two may lean mild or wild, depending on the year. Recent results suggest more wild than mild.
PARAMETER 6: Never pick an upset free regional bracket
No surprise here; I gave this one away a few paragraphs ago. Just remember that an all-chalk bracket happened only once, and that was 40 years ago.
Between now and Selection Sunday, look for one last analytic look at tournaments of the past and my final Perfect Bracket Parameters. Seedigami is coming…
Comments