<
>
Insider

Where No. 1 seeds face trouble

3/16/2011

So far this week in Giant Killers, we have been taking deep looks at matchups that could yield big upsets in what we now call the first two rounds of the NCAA tournament (the field of 68 and the field of 64). But there are potential Killers lurking in the middle seeds, too, whom you need to know about to fully fill out your brackets. Today we ask: If all the 1-seeds get by their 16-seed opponents, how safe will they be against 8- or 9-seed Giant Killers?

Short answer: Not very.

Three potential Giant Killers were given 8- or 9-seeds: George Mason in the East, Old Dominion in the Southeast and UNLV in the Southwest. (No Killer was seeded 8 or 9 in the West, and none was seeded 7 or 10 in any region, so this is a complete roster of teams that could be Giant Killers but that we haven't covered so far.) All three face tough matchups in their first games, appropriately enough for 8 vs. 9 contests: historically, 9-seeds have won 56 of 104, or 53.8 percent, of games against 8-seeds. The numbers suggest that Old Dominion has the best shot of advancing because the Monarchs play Butler, a team that is below average at offensive rebounding and poor at generating turnovers. But really, ODU probably has a slightly greater than 50 percent chance against Butler, with George Mason (which faces a possibly collapsing Villanova) and UNLV (which has to play a tough Illinois squad) just a hair behind.

To have a little fun, let's assume all three make it to the next round. How much trouble could they cause the top seeds in their brackets? Here are our estimates, ranked in order of the likelihood of underdog triumph:

No. 8 UNLV (50.1 Giant Killer score) vs. No. 1 Kansas (16.7 Vulnerability rating) in the Southwest

UPSET CHANCE: 21.7 percent

First of all, 1-seeds have lost to 8-seeds on nine of 48 occasions (18.8 percent), so the absolute invulnerability of the top teams doesn't carry beyond the opening round of the tournament. And UNLV is one of the top 10 Giant Killers our statistical model has seen since our data sample started in 2004. The Runnin' Rebels protect the ball, they grab offensive rebounds, they force turnovers and they block shots. It seems like UNLV always has the ball -- and when it doesn't, it doesn't let other teams do much with it: The Rebels limit opponents to shooting just 43.8 percent (29th in the NCAA) from inside the arc and 32.4 percent (64th) from downtown. UNLV is a highly efficient, possession-maximizing machine at both ends of the floor.

Of course, there are reasons many analysts think Kansas will be the national champion. For our purposes, what really matters is that the Jayhawks dominate inside, outscoring opponents by more than 15 points per 100 possessions on 2-pointers, and force more turnovers than they lose. Those are the two most important factors for depressing the chances of an upset, according to our model.

But if these teams meet, it will be a bruising game.

No. 8 George Mason (36.2) vs. No. 1 Ohio State (21.8) in the East

UPSET CHANCE: 16.5 percent

It's worth pausing for a moment to appreciate Ohio State's numbers. Typically, a good team will show statistical evidence of strong play inside, such as outscoring opponents on 2-point buckets, or outside, like high turnover or 3-point FG margins. But it's hard to dominate at both. The Buckeyes, however, are not just crushing opponents by more than 10 points per 100 possessions on inside scoring. They're also forcing turnovers on 23.5 percent of possessions while giving the ball up just 15.8 percent of the time, for a whopping 8.6-point gap that is the biggest among Giants. That's impressive.

The Patriots' strengths run right into Ohio State's: George Mason is outstanding at protecting the ball and generates a lot of turnovers off steals. Most crucially, Cam Long and Co. are terrific at shooting 3s (39.7 percent, 12th in the NCAA), while Ohio State is below average at defending them. If you're going to repeat as Cinderella, you might as well beat the No. 1 team in the country on your way back to the ball. And for the Patriots to triumph, they will have to follow the classic Giant Killer formula: take bundles of high-risk shots, and execute.

No. 9 Old Dominion (39.6) vs. No. 1 Pittsburgh (5.7) in the Southeast

UPSET CHANCE: 13.3 percent

If Old Dominion clears the hurdle of facing Butler, a highly vulnerable Giant, in its first game, it will have to contend with Pitt, one of the safest Giants, as a follow-up. Watch the Monarchs, and you'll see their strength in about 30 seconds: they crash the boards fearlessly, with offensive rebounds on a staggering 45.2 percent of misses, earning themselves boatloads of extra possessions. Keep a particular eye on F Frank Hassell, a favorite here at GK Central.

But while ODU is tops in the country at offensive rebounding, the Panthers are right behind them, with ORs on 42.7 percent of misses. Pitt also protects the ball ... and shoots well from inside the arc ... and doesn't let other teams penetrate (allowing opponents to shoot just 43.3 percent on 2-pointers, 22nd in the NCAA) ... or hit from outside (32.6 percent on 3s, ranking 70th). And even though the Panthers can roar from downtown, shooting 39 percent on 3s (18th in the NCAA), they rely on bombs for just 23.8 percent of their scoring, so unlike Duke or Wisconsin, they're not subject to criticism for abandoning their inside game for long-range shots. That's a tough package for any Killer to overcome. Truth is, Old Dominion would have had an easier time as a 10-seed in the East, facing turnover-prone UCLA and overrated Florida, than going through Butler and Pitt as a 9.

So there are the pickings, and they're pretty slim. Before Selection Sunday, we worried that potential Killers might land in these slots, and now that they have, they're going to be long shots indeed.

Finally, a note about our missing region: the West. Readers have asked whether we really believe that not having any Killers in the 6 through 11 seeds there somehow makes things easier for Duke, and the answer is yes, we do. At selection time, there still seems to be a bias in favor of power-conference teams, or at least in favor of the kind of numbers where power-conference teams do well. As a result, there are a batch of big names seeded too high and quite a few small-conference teams seeded too low, given their performance in areas that actually determine wins and losses. You can see this using almost any meaningful statistic. For example, whether you look at Ken Pomeroy's advanced metrics or Team Rankings.com, whose predictive ratings use a completely different system based on scoring differentials, Utah State is one of the top 20 teams in the country. But the Aggies got just a 12-seed.

When it comes to the West, the 7 through 10 seeds, Temple, Michigan, Tennesse and Penn State, are ranked between 32 and 54 by Pomeroy and between 30 and 55 by Team Rankings. UNLV and George Mason are qualitatively better than that group, and Old Dominion is right in the mix. And that's before you even consider the way that successful Killers play, intentionally pursuing high-risk/high-reward strategies to increase their scoring variability against better teams. Call us crazy, but it seems to us that George Mason is more likely to adjust its style of play as an underdog, and therefore to play over its head, than Penn State.

Would Duke rather face Tennessee in its second game, or UNLV? The answer is obvious, isn't it?