The Perfect Bracket Project, Year 2

Last year, I and a small (very small) coterie of friends set our sights on doing the impossible: picking a perfect March Madness bracket. Going 63 for 63 and beating literally astronomical odds. Our little project didn't gain the momentum I'd hoped, but we're back for year 2: The Bracketologist Strikes Back. Or something like that. Maybe "Perfect Bracket 2: Tokyo Drift"?

I'd love to rally more people to the quest this time, so I'll make it my aim to post more frequently and dive a little deeper into bracket dynamics. Last year's parameters weren't bad, but an abnormally bland tournament arrived like a mild Hurricane season--perfectly fine and enjoyable, but absent the chaos many of us irrationally seek.

I'll also be doing this in video form too, so be sure to follow, like, subscribe, smash those notifications and whatnot: https://www.youtube.com/@ellomenobrackets

NEVER TELL ME THE ODDS

We'll start in the same place we did last year: how quixotic this quest truly is.

Statistically, the raw odds of picking a perfect NCAA bracket stand at 1 in 9,223,372,036,854,775,808. That's 2 to the 63rd power, following the simple math flowing from the fact that there are 63 tournament games with 2 possible outcomes in each. Exponential math is nothing if not unforgiving. Spelled out, that's

One in nine quintillion, two-hundred and twenty-three quadrillion, three-hundred and seventy-two trillion, thirty-six billion, eight-hundred and fifty-four million, seven-hundred and seventy-five thousand, eight-hundred and eight.

I did a little sleuthing to get a sense of how massively improbable this actually is, since a number like that is so large that it's literally impossible to imagine. Here it is compared with other perceived statistical improbabilities:

Apart from the reality that we should all be a lot more careful when thunderstorms roll in, this chart shows just how much more unlikely any one person is at bracket perfection than the most improbable things we can think of. It is an astonishing 31 billion times MORE unlikely than winning the Powerball. The number is so large that, in order to ensure that it would happen, every man, woman, and child on the planet would EACH need to fill out 1,152,921,505 brackets, with every one of them different from the brackets filled out by every other person. Over one BILLION brackets per person.

Visually speaking, we can picture picking a perfect bracket as being equivalent to randomly flipping open a book to the one page that has been pre-marked as the correct one. Only a book with over 9.2 quintillion pages would be wider than the solar system.

Imagine a single book wider than everything in this image. Then imagine flipping to the one page marked as 'correct' in said book. That's the silliness of the Perfect Bracket Project.

BUT WAIT, THERE'S MORE

On the surface of things, then, no one will ever accomplish the feat of a perfect March Madness bracket. It's insurmountable.

Yet for some reason I've become fixated on this quest. I believe that with a lot of data analysis, a lot of buy-in from the basketball-loving public, and a metric ton of luck, it can happen. My two reasons for believing are ACTUAL PROBABILITY and VOLUME.

The 1 in 9.2 quintillion figure assumes that each game has a 50/50 chance of going either way. But we know this isn't true. This infographic shows the actual probabilities of a given seed reaching the various stages of the tournament, based on 40 years of results from the modern era of March Madness (1985-present, events with 64 or more teams). Lets's look at just the first round, for example:

Only the 8/9 game is a true coin flip. The 5/12, 6/11, and 7/10 matchups are roughly 60/40 (and, interestingly, somewhat even in terms of odds; something we'll examine in a future post), and the rest increasingly lopsided in favor of the higher seed.

Because the true seeding matchups have more variability in later rounds, it's harder to nail down all the raw probabilities (and some matchups, like an 8 v 15 in the Elite 8, are so infrequent that the sample size isn't large enough to treat it reliably...though we did have that one 8-15 matchup in 2022!). However, 40 years of tournaments have yielded an average of 17.7 total upsets. These tend to be spread fairly evenly throughout the six rounds of the event, so we can work on the assumption that about 30% of games from the second round on will result in an upset.

Applying cold exponential math to these real, data-informed trends, we arrive at the true odds of picking a perfect bracket:

1 in ~3,000,000,000

This is still quite ridiculous. You're still 10 times more likely to win the Powerball, but it's a far cry from 9.2 quintillion-to-one. Going back to our book analogy, a book with 3 billion pages is about as wide as the Florida peninsula.

So, without doubt, it will take a good deal of luck to accomplish our dream. But that's where my second reason for believing comes in: VOLUME.

Most bracket sites allow each username registered to fill out multiple brackets. The one I prefer, ESPN, gives 25 brackets each. So if we can get enough people to fill out the maximum number of brackets, maybe even gaming the system a bit by creating logins with all their email addresses, one of those brackets may just land on perfection. We'll be rallying the masses toward that during the run up to Selection Sunday.

Now, a detractor might object that filling out that many brackets invalidates calling them 'picks.' Fair enough. I've long been a "one-man, one-bracket" advocate, and I will still have one bracket that I consider my true predictions. I encourage everyone to do the same. All the others are essentially lottery tickets, and, really, who cares if a bunch of people fill out a bunch of brackets? If one of those brackets makes it through the first weekend unscathed (48-for-48), imagine the media buzz that would ensue. Just once in all the years people's brackets have been archived and internet-searchable has anyone made it perfect that far. CBS and ESPN would send a camera crew to watch games along with the author of that lucky bracket, and the quest for perfection would captivate the nation. It would be like the Manning-cast, except with some lucky fan from Temecula or something. And it would be even better if you were like, "yeah, this is bracket number 15 of mine, but I did it as part of this crazy group of hoop-heads actually trying to get someone, somewhere to pick a perfect bracket."

Let's do this, then. Over the next 2 months, in video and in print, we'll take a crazy deep dive into recorded March Madness history in order to parse out a set of 'Perfect Bracket Parameters:' ironclad (well, maybe rubber-clad, these will have a bit of flexibility) guidelines to follow when filling out your pile of brackets during the third week of March. And, if we have time (read: my life isn't overwhelmed by day job and kids and whatever other diversions captivate my squirrel-like attention span), we'll take a good look at this year's field to see what we might learn by comparing it to previous ones.

It'll be fun. And, while you're here, check out my semi-weekly bracketology forecast too. Peace!