For a better browsing experience, please upgrade your browser.

FiveThirtyEight

Sports

Most office pools — and a certain billion-dollar bracket contest — insist the NCAA Tournament still has the pleasing, symmetrical, 64-team format: Six rounds, each winnowing the field by half, for a total of 63 games, 63 losers and one winner. Fans, likewise, haven’t gotten over the habit of calling the round of 64, played out over the manic Thursday and Friday of this week, the first round.

But the real first round is happening now in Dayton, Ohio, and for the fourth straight year, it really matters. In the so-called play-in games, eight teams will compete on Tuesday and Wednesday for four spots in the round of 64. And two of the winners will have a real chance at a deep tournament run — a better chance than you’d think if you’d glanced at the contenders’ resumes on Selection Sunday. Their first-round — I mean, round-of-64 — opponents could regret having to face them. And you, too, could regret filling out your bracket before the opening-round games are done: North Carolina State and the winner between Iowa and Tennessee could shake up the strong Midwest region.

The play-in games aren’t just a gimmick. Like wild-card rounds in the playoffs for the NFL and MLB, they help sort the contenders from the also-rans. Teams that have to play, and win, a contest before facing their next opponent have momentum, an actual phenomenon in college basketball.1

Most important, play-in games provide vital data about teams’ current strength, which is hard to get from their schedules, littered as they are with non-tournament teams and results from months ago. Winning the play-in game, the most recent and important contest to date against a strong opponent, is a big indicator of a team’s ability today.

The baseball playoffs have validated the potency of this combination of momentum and trial by fire. Wild-card teams that have won their first MLB playoff series have won their next series about half the time, despite facing opponents with home-field advantage and, usually, a better record. Winners in the NFL playoff wild-card round, on the other hand, have slightly underperformed expectations, winning three fewer games against rested opponents (out of 124) than would be expected based on their regular-season performance.2

Sorting contenders from also-rans is particularly helpful in college basketball, a sport that’s particularly hard to predict from regular-season results. Each team has played fewer than 10 percent of other Division I teams. Top teams come from more conferences in college basketball than in football, making each team’s average conference game less meaningful as a postseason preview. Many regular-season starting lineups are a mix of new players and players who have never played with them, meaning November results may predict little about March results.

From 2001, when the play-in concept was introduced, through 2011, this sorting mechanism didn’t matter much, because the single game decided which team would offer itself up for ritual sacrifice in the next round. In those days, play-in games pitted two would-be No. 16 seeds against each other for a chance at a game against a No. 1 seed. Those games aren’t unwinnable, yet they were never won.

Those 10 underdogs did slightly better than expected in the round of 64. Their Simple Rating System3 score heading into the tournament, along with the SRS of their top-seeded round-of-64 opponents, suggested they should have lost those games by an average of 29 points. Instead, they lost by an average of 27 points — a layup better per blowout.

Since 2011, though, the play-in round has expanded to four games, with four of the teams competing to be seeded from 11th to 14th. These teams have a lot more to play for: They aren’t going to face a top-two seed in their next game, so they have a fighting chance of winning.

The NCAA’s move was both innovative and retrospective: The 1983 and 1984 tournaments — with field sizes of 52 and 53 teams, respectively — also had play-in games, then called an opening round. Winners advanced to the first round, which was then also a kind of preliminary round of its own, pitting outsider teams against each other for a chance to play the top 16 teams, which each got two byes.

The back-to-the-future tournament restructuring of 2011 immediately paid dividends. Virginia Commonwealth beat the University of Southern California for an 11 seed in the Southwest region, where VCU was a 10-point underdog to Georgetown, according to pre-tournament SRS. Instead, VCU crushed Georgetown by 18 points. And that was no fluke — the Rams then routed third seed Purdue by 18 and went on to the Final Four.

VCU’s run is an outlier; you’d want good odds to bet on any play-in winner reaching this year’s Final Four in Arlington, Texas. But it’s also consistent with the historical data. Since 1980,4 61 percent of 109 teams that had to win an opening or first-round game exceeded SRS expectations in their next game, against an opponent with a bye. The data set spans the play-in games of the past 13 tournaments, plus the opening rounds and first rounds in the early 1980s, when more teams got at least one bye. And the average team outperformed its rating relative to its opponent by two points. An extra layup doesn’t matter in a blowout, but it could swing a close 5-12 matchup.

The sample size here is too small to be definitive: The standard deviation of teams’ performance relative to expectations is almost 10 points. But other findings corroborate this one. For instance, the analysis so far hasn’t accounted for how play-in teams that won their next game did later on in the tournament. But many went on to make deep tournament runs. VCU was the seventh opening-round winner to get to the Final Four. The 1980 Final Four featured three teams that had to play their way into the main, 32-team bracket. And Jim Valvano’s North Carolina State championship team of 1983 was a No. 6 seed that didn’t get a first-round bye.

Seven semifinalist berths is a remarkable yield from this group of teams. Treat the 32 opening-round winners who won their next game as you would any other team in that round of the tournament, and you’d expect seven of them to reach the Final Four. And yet these were no ordinary teams. Each was, after all, flawed — it was in the opening-round game for a reason. None was seeded in the top four in its region.

Don’t take this as advice to write in any of this week’s play-in winners for a trip to Arlington. The most important factor in predicting winners will remain teams’ relative strength through the season. But if you’re looking for an edge in a bracket contest, you could do worse than backing a play-in winner. And if you’re a fan of a team slotted to play one, hope your team’s coaching staff has been keeping a close eye on Dayton.

CORRECTION (March 19, 11:00 a.m.): An earlier version of this article said four teams would match up in two play-in games. Eight teams will play four play-in games this week in Dayton.

Footnotes

  1. As my colleague Benjamin Morris will show in an upcoming article. ^
  2. This finding is based on NFL playoff data provided by ESPN Stats & Information, combined with Simple Rating System scores from Pro Football Reference, and this formula for converting SRS into win probability (assuming home-field advantage is worth 2.5 points). Why are the NFL playoffs so different from baseball and basketball? Two untested hypotheses: 1) The extra week of rest matters more because the sport is so physically demanding; and 2) SRS understates the gap in quality between bye teams and wild-card winners because many top teams rest starters after clinching byes, artificially deflating their ratings. Also notable: More recently, the NFL playoffs have looked a lot more like MLB’s. Over the last nine postseasons, wild-card winners have won their next game against bye teams 15 times, compared to an expected total of 12 wins. ^
  3. Simple Rating System is, as its name suggests, a basic way of evaluating teams based on their schedule strength and margin of victory. ^
  4. As far back as our data set goes ^

Filed under , , , ,

Comments Add Comment

Never miss the best of FiveThirtyEight.

Subscribe to the FiveThirtyEight Newsletter
×

Sign up for our newsletters to keep up with our favorite articles, charts and regressions. We have two on offer: The first is a curated digest of the best of what’s run on FiveThirtyEight that week. The other is Ctrl + , our weekly look at the best data journalism from around the web. Enter your email below, and we’ll be in touch.




By clicking subscribe, you agree to the FanBridge Privacy Policy

Powered by WordPress.com VIP