Notre Dame’s basketball team has been a great story this season. Unranked to start the year — in fact, Notre Dame didn’t receive a single vote in the Associated Press preseason poll — they’ve compiled a 14-4 record in the deep Big East conference and a 25-5 record overall. Now ranked fourth in the country in the A.P. poll, they could secure a No. 1 seed in the N.C.A.A. tournament with a strong performance during the Big East tournament at Madison Square Garden.
The Notre Dame team is experienced — their lineup often consists of five seniors — but not deep or particularly athletic (they rank just 162nd in the country, for instance, in shot-blocking, which is a decent indicator of raw physical talent). Their best player, Ben Hansbrough, will probably not be drafted until the 2nd round of the N.B.A. draft if he is selected at all. Instead, they’re a smartly-coached, overachieving team that has had a knack for winning close games.
History would suggest, however, that when teams with this profile enter the tournament, the tables are sometimes turned. Instead of exceeding expectations, they often become the victims of upsets.
Since the tournament field expanded to 64 teams in 1985, there have been 41 cases in which a school that was unranked to start the season entered the tournament ranked in the A.P. top 10 (excluding a couple of cases where the team was ineligible for tournament play). These schools, as you might expect, were seeded highly: eight were No. 1 seeds and another 17 were seeded No. 2.
Absolutely none of these teams have made the Final Four, however: they are 0-for-41. Instead, they have been the victims of some notorious upsets, like the No. 2-seeded South Carolina Gamecocks losing — by 13 points! — to the No. 15 seed Coppin State in the opening round of the 1997 tournament, and the No. 3 seed Wisconsin Badgers being one of Davidson’s victims in 2008. Three of the No. 1 seeds that fit this description — Michigan in 1985, St. John’s in 1986, and Cincinnati in 2002 — lost in the Round of 32.
We can take a more comprehensive view of this by comparing the number of games that these teams were expected to win based on their seed — for instance, since 1985, No. 2 seeds have won an average of 2.4 tournament games — to the number that they actually won. Collectively, these 41 teams ought to have won about 95 games based on their seeding. Instead, they won just 70.
In contrast to polls conducted during the regular season, which aspire to rank teams on the merits, the preseason poll is essentially a prediction of how the teams are likely to perform. The writers who vote in the poll presumably consider things like coaching, the quality of talent on the roster, and how the team has performed in recent seasons.
Although we all like to make fun of sportswriters, these predictions are actually pretty decent. Since 2003, the team ranked higher in the A.P. preseason poll (excluding cases where neither team received at least 5 votes) has won 72 percent of tournament games. That’s exactly the same number, 72 percent, as the fraction of games won by the better seed. And it’s a little better than the 71 percent won by teams with the superior Ratings Percentage Index, the statistical formula that the seeding committee prefers. (More sophisticated statistical ratings, like Ken Pomeroy’s, do only a little better, with a 73 percent success rate.)
What’s going on here? It’s almost certainly a case of reversion to the mean.
Consider what would happen if a team ranked No. 2 to start the year — Duke, say — lost its first game to an average team. Quite rightly, your opinion of Duke would go down. But you probably wouldn’t conclude that, for instance, they were only the 200th best team in the country. Instead, you’d figure that they probably had been a bit overrated to begin with but also had been a bit unlucky to lose the game. Maybe you’d think they deserved to be ranked something like No. 20 instead — still a good team, but perhaps not a great one.
As more and more games are played, you’d place more and more weight on their actual performance and less on your initial expectations: nobody would claim that Duke would deserve to be ranked if they started out the year 1-4.
Even though the college basketball season is fairly long, however, it turns out to be a mistake to entirely dismiss preseason expectations, even late in the year. Instead — I’ve studied this issue in preparation for the N.C.A.A. tournament projections that we’re going to release next week — the optimal blend for predictive purposes turns out to be something like five parts in-season performance to one part preseason expectations.
Obviously, this implies that in-season performance — such as measured by computer power ratings — ought to be weighted much more heavily. But preseason expectations do deserve some consideration, and accounting for them might allow you to win an extra game or two in your tournament pool. Teams that have vastly overachieved expectations, like Notre Dame or San Diego State this year, are decent picks to be upset.
You may be wondering if the opposite also holds: do teams that underachieve expectations during the regular season tend to overperform their seeds in the tournament?
Here, the evidence is more mixed. We identified 33 teams that began the year ranked in the A.P. top 10, but that had dropped out of the top 25 entirely by the time the tournament was played (but who nevertheless qualified for the field).
Of these 33 teams, 19 lost in the opening round (bad, but not quite as bad as it sounds since most were fairly low seeds). But there are also some notable successes. Kansas, which underachieved during the regular season, won the tournament as a No. 6 seed in 1988. North Carolina reached the Final Four as a No. 8 seed in 2000. Four of these teams reached the Elite Eight, including Missouri in 2002 as a No. 12 seed.
Obviously, these teams are hit-and-miss: some really have been unlucky, or have had to contend with injury issues that are resolved by the time the tournament began. Others, despite the lofty expectations, were really pretty lousy. Overall, however, they won 33 games when 28 were expected based on their seeds, so they overperformed slightly.
This year’s set of underachieving schools — like Gonzaga, Butler, Kansas State and particularly Michigan State — need to be taken on a case-by-case basis, but they are plausible picks to pull off a surprise or two.
None of this, by the way, should imply that a team like Notre Dame is not deserving of a No. 1 seed (especially if they win the Big East tournament). The seeding committee’s goal, in my view, should be to seed the teams based on their performance, and not to predict how will they will do. I also don’t like the committee’s practice of sometimes docking a team because of a late-season injury, like it did to Cincinnati in 2000 after Kenyon Martin was hurt, which surely affected how they performed in the tournament – Cincinnati, as a No. 2 seed, lost its second game to Tulsa — but did not take away from what they had accomplished beforehand.
Nevertheless, as they say in the mutual fund brochures, past performance is no guarantee of future results. And if Notre Dame is a No. 1 seed, whichever No. 2 seed they are paired with will have to feel pretty good about its draw.