We’ve just added brackets to our college basketball bracketology, and it seemed like a good idea to let everyone know how they’re built, because they have some confusing aspects.
First of all, if you’re unfamiliar with our bracketology model, you should know that it simulates the rest of the college basketball season 1,000 times, spitting out 1,000 different seed lists based on the outcomes of each simulation. This gives us the ability to look not at where things stand now, but where things are likely to stand come Selection Sunday. (editor’s note: For more on this model, read Joe’s explanation of how it works)
Still, people like looking at brackets just as much (if not more) than they like looking at charts. There are good reasons for this. While our chart outlining each team’s chances of receiving different seeds is useful for fans of an individual team, or for comparing the outlooks of bubble teams, it takes a lot of work to get a solid picture of the overall environment. Brackets, despite downplaying the uncertainty of things, give a good picture of the overall environment.
For our brackets, we’ve chosen to use predictive bracketology. Again, we think it’s more useful to look at where things are likely to stand come Selection Sunday than it is to look at where things stand now. But there are still a few different ways to handle that, so in the interest of alleviating confusion, here’s what we do:
- We take each team’s median overall seed from the 1,000 simulations and rank them against each other. For example: While Virginia’s median overall seed is 4th, that’s the best of any team, so it comes in as our overall 1-Seed.
- We take the favorite (not necessarily the highest-seeded team) in each conference tournament and plug it in as its conference’s NCAAT automatic bid.
- We build out NCAAT bracket based on these.
Next, we turn to the NIT:
The NIT does not have a set number of automatic bids. Different NIT Bracketology’s handle this differently, but our goal is to at all times give fans the most accurate look we can at the final NIT Bracket, so we place a lot of importance on giving this number our best guess.
Similarly, the NIT is significantly impacted by NCAA Tournament “bid thieves.” The NIT’s at-large field moves up if there aren’t many bid thieves. It moves down if there are a lot of bid thieves. For NCAA Tournament bracketology, it’s impossible to account for this while creating a complete bracket unless you start putting weird things in the bracket (i.e., putting bid thieves in even though no one specific team isn’t likely to be a bid thief). For the NIT, though, we think it’s easy to account for this, and we think it’s important, because it gives fans the most accurate look at where the cut lines will fall.
Here’s how we handle those two issues:
- We take the median number of NIT automatic bids across our 1,000 simulations and use that number for our model.
- To fill the automatic bids, we take the most likely teams to supply them. This late in the season, a maximum of only one team per conference is a possible NIT automatic bid, but when we’ve produced brackets earlier in the season, we avoid taking multiple “automatic bids” from the same conference. We don’t think it’s useful to put anything impossible in a bracket for either tournament. Contrarily, we do think it’s useful to put teams in both brackets if they’re both the favorite to win their conference tournament and one of the most likely teams to wind up an NIT automatic bid. For example: If a low-major team has a 30% chance of winning their conference tournament, they may be the favorite, but there’s still a 70% chance they’ll end up in the NIT.
- We set our cut lines in a simple way: We simply take the median cut lines from our 1,000 simulations. Again, this means that teams end up in both our brackets (as NIT 1-Seed’s and teams in the NCAAT’s First Four), but this reflects the likelihood of bid thieves.
So, that’s how it all works. As always, feel free to reach out if you have questions. We know it’s confusing, but at the end of the day, if you’re interested in how college basketball’s postseason will look, these brackets (along with our bracketology model) are good resources.