(Inside Science) -- On Sunday when the teams for the NCAA's men's basketball tournament are announced, they'll be grouped into four regions, and receive seeds from 1 to 16 that are intended to reflect their rank from best to worst. If the selection committee does a perfect job -- and the teams involved played to their ability -- then the higher seeded team would win each game. The tournament would be predictable and quite possibly boring.

But that's not how it usually goes. Part of what makes the tournament so compelling to fans is watching lower seeded underdogs defeat top-seeded powerhouses. One reason for that pattern of upsets, new research suggests, is because the seeding doesn't sufficiently account for how well teams play in the games that immediately precede the tournament.

That's what economists Daniel Stone of Bowdoin College in Brunswick, Maine, and Jeremy Arkes of the Naval Postgraduate School in Monterey, California, found when they studied performances from the 2001-2016 men's tournaments. They published their findings in the journal Economic Inquiry.

The pair found that the effect grew stronger in 2010, "when the NCAA unofficially adopted a policy of evaluating teams based on their 'body of work' with games from throughout the season treated equally," Arkes wrote in an email to Inside Science.

More on the NCAA basketball tournament from Inside Science:

How Do You Win A March Madness Bracket?

NCAA Tournament Math: More than Adding Up Ones, Twos and Threes

Suggesting that seeds should reflect how well teams are expected to fare in the NCAA tournament, the researchers argue that performances in the two weeks before Selection Sunday should be more heavily weighted, as well as wins and margins of victory in the teams' conference tournaments.

They found that conference tournament runners-up could be good candidates for dark horse teams when filling out your bracket, because they don't usually receive the same reward in seeding as the winners (Purdue, anyone?).

Trying to divine the selection committee's methods for particular decisions can be difficult. One reason why, the researchers suggested, could be the absence of an incentive to seed the teams completely accurately. Upsets are part of why the tournament is popular, they wrote in the paper, "making the persistence of bias less surprising."

That's another way of saying underseeding good teams can add to the tournament's "madness."