Don’t Forget Those Bid Thieves: 2/5 Advanced Stats Check-In

Who’s in? Who’s out?

It’s February, and that means that we can stop thinking of Bracketology as an interesting little diversion and can start thinking of it as The Most Important Thing in the Universe.

There’s only a month left to solidify that resume. For teams on the bubble, or for Big Ten teams hoping to earn their way there (Hi, Maryland, Penn State, and Nebraska!), that cut line is a big deal. Last Four In feels a whole lot better than First Four Out.

But if you’re sitting in that Last Four In category, you might not really be in. And I don’t mean that you could still lose a bad game or that the Selection Committee could screw your team by putting in Syracuse instead. Both those things could happen. But there’s something else that can happen, too, and I don’t see any Bracketologists who account for it in their projections.

Bid thieves.

So what’s a bid thief?

A bid thief is a team that:

1. Wouldn’t have gotten an at-large bid
2. But wins its conference tournament
3. From a league whose best team will get an at-large bid

In other words, one-bid leagues get one bid no matter what. And non-one-bid leagues usually give their autobid to a team already projected to make the field of 68. But sometimes those autobids get stolen when bid thieves win their conference tournament. It happens every year.

So if we really want to know where to draw the NCAA Tournament cut line, we need to know how many bid thieves we should expect. You could look back at past history, but that’s a subjective exercise since in some cases it’s not clear whether a bid was really stolen or not. And recent conference realignment makes bid thieves less likely than they were in the past.

But this is 2018, and we have nerds who can help us model this out.

One of the coolest advanced stats sites is barttorvik.com, where you can find T-Rankings. T-Rankings are essentially a free alternative version of KenPom, but there’s a bunch of other cool stuff that lets you use those rankings in different ways.

(Incidentally, Mr. Torvik is a Big Ten fan, and he writes at bigtengeeks.com, which you Big Ten fans should add to your Big Ten basketball reading lists.)

One of the cool features of the site is the ability to simulate conference tournaments. That feature is essential if we want to identify bid thieves.

So here’s what we’re going to do. We’re to pull the probabilities that each team in a multi-bid league wins its conference tournament. Then we’ll add up those probabilities for teams who are currently projected to not be in the field. To determine which teams those are, we’ll cross-reference bracketmatrix.com.

Below are the multi-bid leagues and the possible bid thieves in each. (Note that if you’re reading this a day or two from now, the projected field at Bracket Matrix may have changed slightly, but that’s OK. There isn’t much difference between the probability that a team in the First Four In wins its tournament vs a team in the Next Four Out. In other words, though you could swap some teams in or out, their numbers are going to be similar enough that it won’t skew the analysis too much.)

• Big East: Georgetown, DePaul, St. Johns
• ACC: Syracuse, Notre Dame, Wake Forest, Boston College, Pittsburgh
• Big 12: Oklahoma State, Baylor, Iowa State
• SEC: Mississippi State, LSU, Georgia, South Carolina, Ole Miss, Vanderbilt
• Big Ten: Everybody but Purdue, Michigan State, Ohio State, and Michigan
• AAC: Everybody but Cincinnati, Wichita State, and Houston
• WCC: Everybody but Gonzaga and St. Mary’s
• Pac-12: Everybody but Arizona, Arizona State, Washington, and USC
• A10: Everybody but Rhode Island
• MWC: Everybody but Nevada

Here are the T-Rank probabilities that a bid thief comes out of each of the ten conferences above:

• Big East: 0.6%
• ACC: 0.8%
• Big 12: 3.2%
• SEC: 4.1%
• Big Ten: 4.9%
• AAC: 8.5%
• WCC: 11.2%
• Pac-12: 30.6%
• A10: 53.3%
• MWC: 60.1%

(If you’re on the bubble, you really, really, hope that Rhode Island and Nevada win their conference tournaments.)

Using these probabilities, we can simulate 10,000 series of tournaments to estimate the number of expected bid thieves. Here’s what that distribution looks like:

• Zero bid thieves: 9%
• One bid thief: 32%
• Two bid thieves: 37%
• Three bid thieves: 18%
• Four bid thieves: 4%
• Five or more bid thieves: 0%

That averages out to an expected 1.76 bid thieves. We’ll round that up to two.

So, basketball fans, as you look at the projections of your favorite Bracketologist, keep in mind that those last two slots—and maybe as many as four—are likely to be filled not with the teams shown, but with unexpected conference tournament champions. If you’re on the bubble, you have to be just a little bit better than it looks right now.