UPSET ODDS BY REGION
West | Midwest | South | East
On Selection Sunday, we fired up the Giant Killers model to pump out the 10 most likely upsets in the opening round. But now it's time to go deeper. Much deeper.
For each of the four regions, we'll break down the six games with Giant Killers implications. (Reminder: A Giant Killer is a team that beats an opponent seeded at least five spots higher; check out our full methodology for more details.)
We also have sorted the games into four categories: Best Bets, Worth a Long Look, Not Completely Crazy and Stay Away. It's important to remember that we're dealing with probabilities here. No upset is a given -- if it were, it wouldn't be an upset in the first place. And even a game with a 10 percent chance of going in favor of the underdog means that one out of every 10 times, the underdog should win.
Instead, we can give you the likelihood of an upset, based on rigorous statistical analysis that merges a team's base power rating with our GK "Secret Sauce" -- the statistical similarity between potential Giant Killers and historical Davids, as well as Giants and their predecessors who were slain. By coming up with a rating for every team, we are then able to compute an upset percentage for each matchup.
What you choose to do with that information is up to you, and should probably depend on how many points you receive for picking an upset in your pool. Take a look at the upset chances for every game in all four regions, and check back with us on Wednesday when we look ahead to future rounds.
East Region
Cough, cough -- excuse us, but we looked at the East region and flashed back to 2011, when the air around our brackets was clogged with chalk dust. This sextet doesn't feature even one deep-'dog sharpshooter as good as Boise State, or an underseeded squad as talented as Oregon. Happy hunting elsewhere, and in the meantime, heed these warnings.
WORTH A LONG LOOK
No. 6 Butler Bulldogs (Giant rating: 57.4, on a scale of 0 to 100 percent chance of beating a generic Giant Killer) versus No. 11 Bucknell Bison (Giant Killer rating: 3.8, on a scale of 0 to 100 percent estimated chance of beating a generic Giant)
Upset chance: 29.4 percent
We will try not to belabor this analysis, because there's a pretty fair chance here that we're somehow going to insult fans of both of these teams. Let's first examine Butler, which rates as the second-most vulnerable Giant in our spreadsheets. The Bulldogs play smart and they play hard, and they've got obvious strengths (rebounding on both ends, interior defense) and weaknesses (turnovers, forcing turnovers). Overall, they have outscored opponents by 14.5 points per 100 possessions while playing in the Atlantic 10 this season. Thing is, that's not all that different from 10-seeds Colorado (scoring margin: 15 points per 100 possessions)
and Oklahoma (15.2), or 13-seed Boise State (16.0), all of which rank near Butler in BPI. Our model sees the Bulldogs as prone to an upset because they're seeded about four spots higher than teams of comparable strength.
But Butler looks pretty safe against its initial opponent. Fans and media types who are hopping on the Bucknell bandwagon are on to something: The Bison are highly efficient at both ends of the floor. They play at a perfect pace to keep a favorite from running away from them (just 63.2 possessions per game, ranking 285th in the NCAA). And there's just a few points separating them from Butler in fundamental strength, according to BPI as well as our spreadsheets.
Unfortunately, Bucknell's got a dried-out, crusty jar of special sauce in its cupboard. As we've explained, the Bison are a team of almost comical statistical extremes that happen to be the opposite of what Killers need to knock off Giants. Bucknell is one of the very best programs in the country at preventing steals and turnovers, but Killers have to force steals and turnovers, and in those categories, the Bison rank 346th out of 347 NCAA teams. Bucknell is second-best in the nation at stopping opponents from grabbing offensive rebounds, but Killers must generate offensive boards of their own, and there, the Bison rank 240th. And Bucknell can shoot 3s (36.0 percent, ranking 72nd in the NCAA), but while Killers should maximize the value of their possessions, the Bison hardly rely on bombs at all (20.6 percent of scoring from 3-point shots, ranking 320th).
There's an excellent reason for all of this: With the tallest team in the Patriot League this season, Bucknell used its inside presence to squash opponents and extended its dominance through utterly risk-averse play. In one January game against Lafayette, the Bison had all of 1 steal, 6 takeaways and 6 3-point attempts. But they turned the ball over only six times, amassed 28 defensive rebounds and held the Leopards to 19-for-55 shooting (34.5 percent), including 1-13 (7.7 percent) from long range, and won by 15.
Bucknell has 28 wins, and four of its five losses have come by a total of nine points. It's just not likely that this style will find success in the kinds of games it will face in the NCAA tournament. It's not how underdogs win. Killers press, steal, shoot bombs, get in position for offensive rebounds -- they amp up risk to boost reward, and if they do it right, they increase the variance in their scoring by enough to catch a Giant. You can't do that by playing it safe, even if you play hard and smart. Because remember, Butler plays hard and smart, too, and they've run up bigger margins on much better opponents.
The first level of advanced metrics correctly identifies Bucknell as a very good team. The next level says it's very hard to out-Giant any Giant.
NOT COMPLETELY CRAZY
No. 3 Marquette Golden Eagles (81.3) vs. No. 14 Davidson Wildcats (14.6)
Upset chance: 25.3 percent
At first glance, Marquette, a weak 3-seed -- are we really supposed to believe the Golden Eagles are on par with a Florida or Michigan State? -- looks vulnerable. And the Wildcats, who are riding the nation's longest winning streak (17 games) and can stroke it from anywhere (effective FG percentage: 53.3 percent, ranking 28th in the NCAA) seem like a potential Killer. And, hey, maybe some analysts would stop there. But we take you into the depths of the secret sauce, where the chemistry of Giant-Killing bubbles forth second thoughts.
Marquette is excellent on the offensive glass (offensive rebounds on 37.8 percent of missed shots, ranking 22nd) and at defending the perimeter. Davidson doesn't generate steals (8.8 percent of opponent possessions, ranking 258th) or offensive rebounds (ranking 252nd). Comparing their fundamental strengths straight up, the Wildcats would have a 1-in-3 chance of pulling off an upset. Include the factors that have been key to historical Giant-killing and safety, and those odds drop to 1 in 4.
No. 5 UNLV Rebels (76.0) vs. No. 12 California Golden Bears (5.8)
Upset chance: 22.3 percent
This is yet another case where an underdog's lack of Killer traits drags the odds of an upset way down. Cal lacks a true center, and runs a three-guard offense where PG Justin Cobbs, who plays more than 35 minutes a game, either penetrates on his own or gets the ball to SG Allen Crabbe, who plays more than 36 minutes. On defense, they cement two big forwards near the basket. And that's about it. The Golden Bears are just average at defending the perimeter, and are one of the worst teams in the country at generating turnovers (16.6 percent of opponent possessions, ranking 321st in the NCAA) and shooting 3s (30.4 percent, ranking 309th). So if their big men wander outside, or Cobbs gets into foul trouble, or Crabbe has an off night (which, you never know, could include coach Mike Montgomery pushing him), they don't have ways of climbing back into games.
Meanwhile, UNLV plays up-tempo and pressures opponents until they crack -- but in an interesting twist, the Rebels attack passing lanes rather than the ball, and that generates tons of blocks (14.7 percent of opponent shots, ranking 10th in the NCAA) and steals, rather than turnovers per se (ranking: just 252nd). Their transition defense leads to loads of easy baskets, rendering it less problematic that UNLV is somewhat sloppy with the ball (turnovers on 20.3 percent of possessions, ranking 200th) and has been plagued by poor long-distance shooting (32.7 percent on 3-point attempts, ranking 227th).
Watch this matchup, and the teams' roles will seem completely reversed. But UNLV, the pressing and pushing Giant, gives up just 88.3 points per 100 possessions, the 13th-lowest rate in the country. And predictable California is a Killer whose slingshot has no rocks.
No. 2 Miami (FL) Hurricanes (60.7) vs. No. 15 Pacific Tigers (8.8)
Upset chance: 15.9 percent
Here's why we spend so much time poring, and rhapsodizing, over teams like the Cal Poly Mustangs during conference tournaments: because when they lose, and a team like the Tigers nabs an auto-bid in their place, the world loses a chance to learn something interesting about a Giant a week later. In this case, while we yield to no one in our admiration for coach Jim Larranaga, his Hurricanes are considerably more vulnerable than most fans and analysts realize.
The Canes are overseeded (Miami is a top-16 team, according to most analytical models, not top-8). They don't crash the offensive boards as a matter of philosophy (ORs on 30.8 percent of missed shots, ranking 201st in the NCAA). And they don't generate turnovers (18.4 percent of opponent possessions, ranking 264th). Miami does handle the ball exceptionally well, and forces opponents to take lots of bad shots. But one effect of the overall Larranaga program is to broaden the Bell curve of a team's scoring, and of the 30 best teams in the country, Miami has the second-highest game-to-game variance in BPI. That's usually great news for a Killer, but bad for a Giant.
Pacific, however, is hardly likely to put the Canes to the test. The Tigers can hit 3-pointers (shooting 39.1 percent on bombs), but don't take all that many. They also don't generate offensive rebounds or steals, and they're ineffective on defense outside and inside. Few turnovers will keep down the number of possessions in this matchup, disguising the blowout, but a blowout it shall be. And maybe we'll find out more about Miami in the Round of 32.
STAY AWAY
No. 4 Syracuse Orange (92.1) No. 13 Montana Grizzlies (4.6)
Upset chance: 4.6 percent
What's this? Montana has already become something of a trendy upset pick? How many ways do you want us to take apart that silly idea? First, while Syracuse isn't quite as strong as the squad that earned a No. 1 seed last season, it's built according to the same basic blueprint. Offensively, the Orange are highly efficient, scoring 114.2 points per 100 possessions, because they're monsters on the offensive glass (ORs on 39.8 percent of missed shots, fifth-most in the NCAA) and their 2-point shooting and turnover rate rank in the top third of the country.
On defense, Syracuse doesn't collect many rebounds, but who cares? The 2-3 zone has enabled the Orange to rank 10th in the country in steals (13.5 percent of opponent possessions) and 24th in turnovers (23.4 percent). They don't rely on free throws, and they've played Killer-like nonconference opponents for target practice. They're not only a very good team, they're a Giant wearing a bulletproof vest.
As for Montana, if you don't even attempt to grab offensive rebounds (25.1 OR percentage, 335th in the NCAA), you had better be trying to cook up some other recipe for adding possessions to your column, but the Grizzlies also force very few turnovers (18.5 TO percentage, ranking 253rd). Maybe you don't need extra possessions when you're playing the 283rd-strongest schedule in the country -- quick, where is Minot State? What about Carroll College? -- but trust us, a few more balls in hand would be handy in the NCAA tournament.
We will stipulate that Kareem Jamar is an exciting player who can post up or shoot from behind the arc, and that the Grizzlies have no fewer than four players shooting 40 percent or better on 3-pointers (in 55 to 93 attempts apiece). But come on. Syracuse outscored opponents by nearly 25 points per 100 possessions while playing in the second-strongest conference in the country. Montana outscored opponents by 1.9 points per 100 possessions while playing in the 27th-strongest conference in the country. (Read those last two sentences again.) They're not just a Killer lacking a second shot, they're a mediocre team.
P.S.: Including "secret sauce" points gained or lost for playing like past Killers and Giants, the average gap in overall strength between 4-seeds and 13-seeds is 19.8 points per 100 possessions this year, according to our spreadsheets. That's actually higher than the average for 2-15, 3-14, 12-5 and 11-6 matchups. Which means that even though a few 13-seeds have popped through in recent years, 2013 is precisely the wrong spring to be hunting for upsets on the 4-13 line.
No. 1 Indiana Hoosiers (95.4) vs. No. 16 James Madison Dukes (5.8)
Upset chance: 2 percent
The Blackbirds and Dukes will tip off at 6:40 p.m. ET on Wednesday, and what more could you want? Both of these teams stole automatic bids from higher seeds (and better Killers) in their conferences, and will now fight for their tournament lives. One of them (LIU) could shoot an apple off your head from 100 paces, but couldn't defend eastern Australia in a game of Risk. The other (JMU) generates defensive efficiency through steals and blocks, but can't score. And they are separated by just three spots in BPI and less than one point per 100 possessions in our spreadsheets. We're telling you, their collision is going to be pure hoops fun. Unfortunately, both are also below-average in overall quality. And 45½ hours later, the winner will face the single safest Giant in the tournament.