Nate Silver’s N.C.A.A. Tournament Forecast
Nate Silver is estimating each team’s chance of advancing to any given stage of the N.C.A.A. bracket.
Even before the N.C.A.A. men’s basketball tournament bracket was announced on Sunday, there was plenty of discussion about how much parity there was in this year’s field. The chatter only increased after Louisville, the No. 1 overall seed, was placed in a brutally tough Midwest region that also includes Duke and Michigan State.
This condition is nothing new, however. Parity has been the rule for some time in the N.C.A.A. tournament.
Louisville is in fact the nominal favorite to win the tournament despite its tough draw, according to the FiveThirtyEight forecast. Still, Louisville has only a 23 percent chance of doing so, just ahead of Indiana at 20 percent.
In 2012, the FiveThirtyEight formula listed Kentucky as the tournament favorite. That call looks prescient since the Wildcats went on to win. Still, the result involved as much luck as skill, since the forecast gave Kentucky just a 27 percent chance of winning, only modestly better than Louisville and Indiana this year.
In 2011, the forecast looked similar to this year’s, with Duke, Ohio State and Kansas each listed as having between a 17 percent and a 22 percent chance of winning. (Connecticut, which the formula had as a long shot, won it instead.) With elite players quickly fleeing to the N.B.A. and midmajor teams posing a threat to big-name programs throughout the tournament, the days when a dominant team like U.C.L.A. or U.N.L.V. might enter the tournament with a 50-50 chance of winning it are probably long gone.
While the absolute difference in the strength of the teams may be on a long-term decline, the N.C.A.A. tournament selection committee seems to be doing a better job of sorting them out. There were relatively few controversies this year in the bracket picks; in fact, a large number of professional and amateur analysts predicted the 68-team field exactly.
This is not to suggest, however, that you might as well throw darts at the wall to pick your tournament bracket. A few teams, like Florida and Michigan, appear to be underseeded. Others are helped or harmed by their draw, or by how far they have to travel.
What follows, then, is an overview of how the FiveThirtyEight forecasts see each of the four regions. I won’t review the methodology in much detail because it is essentially unchanged from the past two years and is explained at considerable length here. Mostly these projections are based on a series of computer rankings that have had strong predictive power in the past. (Unlike the N.C.A.A.’s dubious R.P.I. formula, each of these computer systems accounts for margin of victory along with wins and losses.) To a lesser extent, the forecasts also rely on the N.C.A.A.’s 68-team “S-curve” — how it rated every team in the field — and preseason rankings, which serve as a measure of a team’s underlying talent level. (Teams that have surpassed preseason expectations in the regular season have tended to underachieve in the tournament, and vice versa.)
The FiveThirtyEight measure of team strength is also adjusted for recent injuries, like the one sustained by Jordan Adams of U.C.L.A. — and for players who were injured for part of the season but who have now returned, like Ryan Kelly of Duke. And it accounts for travel distance: teams hauling across the country during the tournament can be playing the equivalent of a road game, whereas a team playing in an arena an hour or so from its campus may effectively be playing a home game.
The Midwest is almost certainly the toughest region: there is about a one-in-three chance that the eventual tournament champion will emerge from it, the highest of the four sub-brackets. Louisville, the top seed, began the season ranked No. 2 in the country, and closed it by winning the Big East tournament. Duke could be the best No. 2 seed in the tournament and is healthier than it has been for much of the season. Michigan State, the No. 3 seed, is one of a number of Big Ten teams that the computers are enamored of, and it will get to play its first two games just down the road from East Lansing, Mich., in Auburn Hills. No. 7 Creighton, No. 9 Missouri and No. 12 Oregon all look a bit underseeded and might have been good upset picks were the top of the region not so tough.
The saving grace for Louisville is its travel itinerary: its road to the championship would run through Lexington, Ky. (just 70 miles from campus), then Indianapolis (just over 100 miles), then Atlanta (about 300 miles away). Its potential Round of 16 matchup, against No. 4 St. Louis or No. 5 Oklahoma State, also does not look especially tough. Mostly, though, the forecast favors Louisville simply because it is a good enough team to endure some difficult games.
The South is a bit upside down. Kansas is not necessarily undeserving of a No. 1 seed, but it is generally seen as behind Louisville and Indiana by the computer rankings. Georgetown, the No. 2 seed, is dinged by the computers for having a mediocre offense, which Ken Pomeroy’s rankings regard as being the 62nd best in the country.
But Florida, the No. 3 seed, is rated as the best team in the country by some of the computer systems, including Pomeroy’s, despite losing seven losses and playing in the mediocre Southeastern Conference. How come? The answer boils down to margin of victory. Florida’s losses came by margins of 1, 3, 3, 4, 6, 6 and 11 po
ints. By contrast, its wins came in blowouts; the Gators didn’t win a single game by fewer than 10 points.
As much as the conventional wisdom might chide Florida for having performed poorly in the clutch, there is an abundance of statistical evidence that a team’s record in close games is mostly a matter of luck, and that this luck turns around often enough. Had Florida split its single-digit games, for instance, it would have gone 29-4 this year, which may be a better indication of its strength.
One additional factor helping Florida is that Kansas could face a very tough Round of 16 game against No. 4 seed Michigan, which had until recently appeared to be in the running for a No. 1 or No. 2 seed and isn’t all that easy to differentiate from the Big Ten teams that are seeded ahead of it. North Carolina, the No. 8 seed, could also give Kansas problems if it begins to play up to its preseason billing.
Some Indiana backers were disappointed that the team was placed in the East, where it would play its regional games in Washington rather than Indianapolis. However, Indiana would probably not want to swap places with Indianapolis-bound Louisville, since the draw in the East looks much more favorable.
Miami, the No. 2 seed, is not regarded well by the computer formulas. The Hurricanes have as many high-quality wins as any team in the country, but also baffling losses to the likes of Indiana State, Wake Forest and Florida Gulf Coast University. The FiveThirtyEight model is also suspicious of Miami since it was not ranked to start the season, a sign of a team that might revert to the mean come tournament time.
Marquette is almost certainly the weakest No. 3 seed, and has about a 35 percent chance of being upset by No. 14 Davidson in its opening game. A potential Round of 16 game against No. 4 Syracuse in Washington could be Indiana’s toughest test.
With that said, there is no real reason to go searching for a fashionable upset pick, like No. 6 seed Butler, to emerge from the East. Indiana began the season ranked No. 1, has far and away the best offense in the country, and won the regular season title in the best conference.
The top two seeds in the West region, No. 1 Gonzaga and No. 2 Ohio State, come out almost equally in the power ratings despite very different profiles, with the Zags losing only two games to the Buckeyes’ seven, but against a much weaker schedule.
What differentiates the team is geography, something that is always a high-risk, high-reward proposition for teams in the western part of the country. If they’re placed in the West region, as Gonzaga is this year, they may benefit from playing a number of games against jet-lagged teams. Gonzaga could still face tough games against No. 8 Pittsburgh and No. 5 Wisconsin, for instance, teams that the computers regard as underseeded, but it will be easier when those games are played in Salt Lake City and Los Angeles, rather than in Newark or Charlotte, N.C. Ohio State, conversely, faces a potential Round of 16 game against No. 3 New Mexico, a team that the computers are not all that fond of but whichwould be playing closer to home.
Travel could work against Gonzaga if it reaches its first Final Four, however, when it would face a 2,000-mile journey to Atlanta.