Picking favorites in the NCAA tournament is the smart play. In matchups where there is a favorite in terms of seeding, the higher seed has won 71 percent of the time since 2011, the first year under the current tournament format. But it’s the upsets and Cinderella teams that get you bragging rights — and the edge you need to win your bracket pool.
And we’ve developed a new lens to spot them before you make your bracket picks.
The teams best built to topple favorites are the ones that have a high scoring margin while creating extra possessions through turnovers and offensive rebounds. Since 2011, the first year the NCAA used a 68-team format, there have been 50 instances in which a double-digit seed upset a lower seed. More than half (28, 56 percent) had a possessions edge from turnovers and offensive rebounds. To help you find and evaluate those diamonds in the rough, I have developed DAViD — the Data-Assisted Victory Detector — for the NCAA tournament.
DAViD, a reverse acronym named after the plucky shepherd who vanquished the warrior Goliath with a sling, looks at each individual matchup from the ground up, starting with an estimated number of possessions for each team, taking into account any additional possessions to be had via offensive rebounds and turnovers. Then, estimated scores are derived by multiplying those respective possessions by the team’s adjusted offensive rating found at Ken Pomeroy’s site. Once we know the projected scoring margin, we can infer an implied win percentage. For instance, teams that are favored by two points would have an expected win probability of 55 percent. That rises to 72 percent if the predicted scoring margin is seven points.
Finally, these win rates are compared to what we would expect based on looking at seed matchups in a vacuum, with stronger upset potential reflected in a higher DAViD score. Historically, keying in on a 12-seed to upset a 5-seed has paid off, but in larger pools with hundreds, perhaps thousands, of competitors, you need other ways to distinguish your bracket, and that involves taking some chances. DAViD incorporates a value element for each particular upset in relation to what we would expect on seed matchup alone, with higher DAViD scores indicating not only strong upset potential but significant value as well. This helps you separate from the pack in your bracket pools when you pick correctly.
Let’s walk through an example from 2017, when No. 12 Middle Tennessee upset No. 5 Minnesota.
Based on its adjusted scoring margins, Middle Tennessee had a 34 percent win expectancy in the first round. However, the Blue Raiders created almost five extra possessions per game via offensive rebounding and turnovers, while Minnesota broke even (0.6 per game), boosting Middle Tennessee’s projected win rate over Minnesota to 58 percent. Over the past seven years, the No. 12 seeds have a 12-16 record (43 percent) against the No. 5 seeds, giving Middle Tennessee a DAViD score of 135 in 2017.
As a handy cheat sheet, here’s how you should interpret the DAViD scores:
|0-50||Not a great upset pick|
|51-100||Below-average chance to upset/poor value|
|101 to 125||Above-average chance to upset/good value|
|126+||Potential bracket buster|
Check back after the bracket is set Sunday for a rundown of the most likely, and most valuable, upset picks.
Read more on college basketball: