Why did UVA lose to UMBC?

It is sad that UVA’s basketball team’s historic season will likely be remembered for all the wrong reasons. UVA was great during the regular season, and for their efforts earned a #1 seed in the NCAA tournament, only to be the first #1 to fall to a #16.

In several attempts to analyze the historic upset, the writers have repeated a familiar trope: UVA isn’t built for the postseason. Though under Tony Bennett UVA has been a reliable force in the regular season, somehow they’ve been unable to find even modest success in the Big Dance.

The question is, “Should UVA change?”

Lots of basketball pundits dislike the slow pace and deliberate tactics that UVA uses, but at least this year they’ve been forced to admit that the strategy works pretty well. Yet now some who seemed to grudgingly accept the effectiveness of the Pack Line will surely want to revisit their concessions.

The thought is that the slow pace and suffocating defense typically comes at the expense of offense, and so when the need is for better scoring, UVA just doesn’t have what it takes. Therefore, the argument goes, UVA’s consistency and discipline undermines its ability for success in the post-season, when it is more likely that they will need to make up deficits quickly. UVA would do better if they would sacrifice a little defensive skill in exchange for some better scorers. They might even do better if they would play defense the “normal” way, thereby increasing the number of possessions and making it easier to catch up.

I think that this argument is understandable, but mistaken. More to the point, I think using Friday’s night’s upset by UMBC as evidence for the claim that UVA should change misunderstands what happened.

The upset

So why did UBMC win? A few factors seem relevant.

First, UMBC has a smart coach who knows how college basketball works. He had his guys prepared and calm, and kept them focused on the prize.

Second, the game plan itself was smart. UVA has long been susceptible to teams who can overload with guards and make 3-pointers. UMBC exploited this deficiency to perfection (almost literally, as we’ll see).

Third, UVA was terrible. I’m not sure that I’ve seem them play so badly under Tony Bennett.

But mostly, it was luck. In fact, sheer luck is probably as important as all of the other factors put together.

Even before the tournament, UMBC was already the luckiest team in Division 1. And it’s not really that close. UVA was slightly above average in luck; UMBC was already a few standard deviations above the mean. They’ve had a magical season.

When we consider the game itself, the first half was pretty unremarkable. Sure, it’s cool for the #16 to be tied with the #1 at halftime, but UVA fans probably weren’t surprised. DeAndre Hunter was out, Devon Hall and Isaiah Wilkins were in some foul trouble, but UMBC had only 21 points. It looks pretty familiar to UVA fans. A few more points on UVA’s side would have been nice, but it wasn’t way out of the norm. And UMBC had already been hot, especially from 3-pt range. UVA was shooting about 10% from 3, and UMBC was over 40%. UMBC’s three-point shooting was keeping it close.

The second half was another story. UMBC’s performance was close to as efficient a performance as is possible. We can note first that UMBC scored 53 points in the second half, which is the average number of points UVA allowed per game this year. UMBC ended up with the highest score against UVA this season.

So did UMBC just take a ton of shots? Did UVA’s defense just completely fall apart? Did UMBC finally figure out how to speed UVA up? Not really.

Usually win percentage correlates pretty well with field goal percentage.

At extremes of pace (very few or very many attempts) the numbers aren’t quite as consistent. But in this chart, you can see that the basic relationship is pretty stable. For reference, the blue dot is UVA’s performance on Friday. The red dot is UMBC’s performance for the whole game. And the purple dot is UMBC’s second half performance, if that rate had continued for the whole game. The data producing this plot is all NCAA Div I games since the 2009 season. There are only a dozen or so team performances better than UMBC’s since then.

If we look just at UVA’s games, we can see an even more dramatic outlier.

Here are all of UVA’s (regular season) games since Tony Bennett became the head coach. Each dot shows the field goal attempts and percentage for each opponent. The orange dots represent opponent wins and the blue are opponent losses. The red dot is UMBC’s game on Friday, and the orange dot at the very top is the second half of that game, extrapolating the number of FG attempts to two halves.

Two features of this graph deserve note. First, there are a couple of UVA losses surprisingly close to the bottom left. Occasionally UVA just can’t score. Second, no one has shot over 60% for a game against UVA. 

In general, though, UVA is dominant against other teams. The boxplot shows the last four years of UVA versus other teams. The values here are scaled so that we can see how far from the norm UVA’s opponents are. Because the scores are scaled, UVA’s numbers center around 0. When UVA’s opponents don’t play UVA, they’re about 0.5 standard deviations above average in FG%, compared to all teams. (Most of them are in the ACC, which historically is a good to great conference.) When they play UVA, they are between 0.5 and 2 standard deviations below average. Ordinarily, UVA dominates pretty decent teams.

But, as is the case with any statistical distribution, there is a lot of overlap. Shooting 30% better than UVA is possible, even if it is very unlikely. But we already knew that UMBC’s win was unlikely. This chart just reminds us that it wasn’t impossible, even given the existing distributions of performance.

UVA and the NCAAs

So why does UVA struggle so much in the NCAA tournament? If we take the pundits’ view, it’s because of the style. But if we take the data analysis view, it’s just as likely that they haven’t had enough chances yet. Indeed, UVA fans can easily think of some quirky reason that made a solid season turn into a sketchy tournament. Injuries (most recently Hunter), bad draws (MSU twice?), etc. all seem like viable possibilities.

Sometimes people say that UVA struggles in March. But that’s not quite right. They struggle in the NCAA Tournament, but fans of other ACC teams should quickly realize that UVA doesn’t struggle in March. They’ve won the ACC or been close for several years in a row. So it seems as if there is something about the NCAA tournament that causes problems. If it’s not just random luck (still I think the best explanation), then what changes from the ACC tournament to the NCAA one?

The obvious thing that changes is the officials. And I wonder if this has something to do with it. UVA plays a distinctive kind of game, and I wonder if the ACC officials get used to it, as do ACC teams, and UVA gets used to the others being used to it. Then we get to the big tournament, and now there are officials who aren’t as familiar with the tactics, and call them differently. UVA does seem to get more fouls in the NCAA tournament, even though during the regular season they usually get very few. This seems like a small thing, but it might matter in tight games.

As I mentioned above, some people say that UVA just plays too slowly, so they can’t generate offense when necessary. I think folks should take a close look at the last 60 seconds of the game at Louisville this season. UVA can score fast when they need to, and focusing on pace of play misses the simpler point.

Fundamentally, UVA just struggles to score at all sometimes. For the last few years (basically since Anthony Gill graduated) UVA has been almost incapable of any kind of post game. Their big guys are excellent defenders, but they can’t score reliably.

This is, I think, the better way to think about UVA’s style. Their offense is one-dimensional, and though their guards are talented shooters, they understandably struggle when the opponent can load up opposing guards, feeling safe to ignore the big men. UVA would be very good indeed if the offense could alternate between an inside post game and the typical blocking/screening scheme they use.

Perhaps the best argument against UVA needing to change is UMBC itself. The Retrievers won basically by imitating UVA.

Other consistently excellent teams have sometimes struggled at the highest level. I’m a long-time fan of the Atlanta Braves, who won 15 consecutive division titles, but only one World Series. Consistent excellence isn’t showy, but it is excellence. UVA is building that kind of program, and I suspect that eventually the trend will swing the other way, and everyone will suddenly forget why they thought Tony Bennett’s teams were incapable of March success.