After a relatively sedate start to the NCAA men’s basketball tournament, the most shocking result in tourney history was saved for one of the last games in the round of 64. Late Friday, Maryland-Baltimore County did the unthinkable, pulling off the victory that college fans have been waiting decades to see: The Retrievers became the first No. 16 seed in the history of the men’s tournament to knock off a No. 1 when they beat Virginia 74-54.4
(Look, we did say a No. 16 seed could beat a No. 1 this year — never mind which No. 16, and which No. 1…)
No, instead of the obvious pick, it was the obscure one. UMBC went into the game ranked 188th in Ken Pomeroy’s power ratings; Virginia ranked first. Our predictive model gave the Cavaliers a 98 percent probability of winning. Nothing about the two teams’ track records suggested that number was too high. (If anything, facing a Retrievers team that lost by 40-plus points to the likes of Albany, UVA seemed like it might be the safest No. 1 seed in the first round.)
Instead, the Cavs were on the wrong side of history.
Besides the novelty of a No. 16 beating a No. 1 seed — let alone the tournament’s No. 1 overall team — the magnitude of the upset would have been shocking in any context. According to ESPN’s Stats & Information Group, it was tied for the worst loss by an Associated Press No. 1 team to an unranked team at any point in the season, let alone the tournament.5 And from a bettor’s standpoint, UMBC joined Santa Clara in 1993 and Norfolk State in 2012 — a pair of No. 15 seeds that won — as the only teams to win a game outright in the men’s tourney despite being an underdog of 20 or more points.
Virginia can place some of the blame on losing forward De’Andre Hunter right before the tournament. Sidelined with a wrist injury, Hunter was UVA’s best player on a rate basis,6 according to Sports-Reference.com’s Win Shares per 40 minutes. The Cavs no doubt missed his efficient play at both ends of the court against the Retrievers.
But this loss cannot be pinned completely on Hunter’s absence. Against UMBC, Virginia played just about the worst game it possibly could. What had been one of the top offensive teams in the country during the regular season shot just 41 percent from the floor — including 18 percent from 3-point range (to go with a 50 percent showing from the foul line). More shockingly, the nation’s best defense looked completely lost, particularly in the game’s second half.
The Retrievers played their best possible game as well. They shot a stunning 54 percent from the field against the Cavaliers, nailing 12 of 24 attempts from beyond the arc. Going into the game, Virginia hadn’t allowed an opponent to shoot better than 50 percent from the field all season, against a schedule that consisted of many difficult opponents. Though no major-conference school could solve the UVA defense, UMBC — led by an audacious 28-point performance from electrifying guard Jairus Lyles — was the team that finally broke through. Virginia was holding opponents to 53.4 points per game, the lowest in the country. Lyles and the Retrievers got 53 in the second half alone.
In terms of blame, there’s plenty of finger-pointing left for Virginia coach Tony Bennett’s deliberately low-possession brand of basketball. The Cavs were easily the nation’s slowest team all season, and they basically never had to accelerate the pace to make a comeback push all year. As TV analyst Kenny Smith pointed out in the postgame coverage, UVA was unable to adjust once UMBC forced the Cavs to play from behind.
This topic deserves more research, but there is some evidence that slow favorites generally disappoint in March. It makes sense: Unlike underdogs, who should want more variance — and therefore fewer possessions in a game — favorites should theoretically be trying to reduce the randomness and give variance a smaller role in a game’s outcome. So it could be that Virginia-style favorites are inadvertently doing their opponents a big favor by slowing the pace down to a crawl, essentially giving away some of their talent advantage by allowing more randomness to seep into the game. (Whether this works in practice, though, is up for debate.)
Whatever the reason behind it, UMBC’s win was an upset for the ages. On the one hand, it was surprising that no 16 seed had ever beaten a No. 1 before Friday — according to our Elo model, we’d calculate the chances as 1-in-27 (or 3.7 percent) that a No. 16 wouldn’t have broken through in 134 tries. But this was also the victory that just seemed like it might never happen. Now that it has, there are no gimme games in the bracket anymore. Just when you thought it wasn’t possible, March Madness somehow got an extra dimension of madness. And it all comes courtesy of the UMBC Retrievers and their long-awaited underdog victory.