<
>
Insider

Saturday preview (and a bit of gloating)

3/19/2010

We'll admit it, we let out quite a whoop here at GK Central when Murray State's last-second 15-footer gave one of our "Best Bets" a huge win. Mostly, though, we've spent most of the past 24 hours staring in awe as the events of an unprecedented day of Giant Killing drama unfolded. We told you, definitively, that this March would be a Killing season unlike any we'd seen. And Thursday's assassins didn't disappoint.

Consider this: Of the 16 games played on Thursday, 11 qualified for GK status. Three Davids actually pulled off the feat, which is a heck of a percentage when you look back at history. And that doesn't include a pair of victories by double-digit seeds that didn't count as GK wins -- Washington (because it's from the Pac-10) and Saint Mary's (because there wasn't a spread of at least five seeds). Add in the near-misses by Sam Houston State, Robert Morris and San Diego State, and Goliath was dodging stones everywhere he turned. Further, eight of the 11 potential Killers covered their point spreads, including six out of seven teams we labeled "Best Bets," "Worth a Long Look" or "Not Completely Crazy."

We did believe in UTEP -- not to the degree of Murray State or Cornell, but pretty strongly. And the Miners, after an impressive first half, absolutely imploded. That stung a bit. But it also got us thinking. One of the key attributes we've identified in GKs is that they tend to employ high-risk/high-reward strategies. And what's interesting about the Miners is that the composition of their roster could also be construed as high-risk/high-reward, given the collection of transfers found in El Paso. And when Butler unleashed its own barrage of high-reward basketball -- 3-point shooting -- UTEP absolutely collapsed mentally. It's one of the many things we wish we could account for in our model, but we're still working on a measurement for mental toughness and chemistry in our spreadsheets.

But for us, that's the best part about Giant Killers -- we're constantly learning. With every passing year, we have more data (due, in large part, to the greatness of Ken Pomeroy). That info allows us to fine-tune our approach and target new methods of analyzing David/Goliath games. So, while we're proud of knocking one out of the park with Murray State, pleased that we called Ohio over Georgetown "Not Completely Crazy" when the rest of the world gave the Bobcats no chance, thrilled that Sam Houston State scared the daylights out of Baylor and content with warning you to avoid a host of games with low upset probabilities, we always want to do better. We're doing something right -- just ask Vanderbilt fans. But we recognize we can do better, and we're constantly looking for ways for our model to more accurately discern GK talent. You'll see some of the issues involved in the following previews of Saturday's Giant Killer games.

SATURDAY'S GK GAMES

No. 5 Butler (24.2 Vulnerability Rating) vs. No. 13 Murray State (94.5 Giant Killer Rating)

UPSET CHANCE: 22.9 percent

On Thursday, the Racers put on a GK clinic, using speed and hustle to dominate Vanderbilt on the offensive boards (13-7), shooting 47 percent on 3s while limiting the Commodores to 33 percent and spreading their scoring around so effectively that they won even though their two biggest players saw just 32 minutes of action combined. (And Danero Thomas took one huge step toward making the All-Giant Killer team.) Our model can't say it's likely that the carpet ride will continue, because Butler is much better than Vanderbilt at generating turnovers and limiting opponents' offensive rebounds. But the warning signs are there. The Bulldogs are terrible on their own boards (31.3 percent of missed shots, 220th in the NCAA). And more than half their scoring margin comes from the free throw line. Everyone says free throws are ultra-important at this time of year, but you know what can happen to teams like that? Just ask Vanderbilt, which went 17-of-29 from the line while Murray State was shooting an even worse percentage at 9-of-17, but with 12 fewer attempts. If our model had a sense of humor, it would laugh.

No. 3 Baylor (26.1) vs. No. 11 Old Dominion (66.0)

UPSET CHANCE: 17.2 percent

In the first round, the Monarchs were outrebounded by Notre Dame on the offensive glass 15-8, and 35-30 overall. That's very unusual for Old Dominion, the No. 1 offensive rebounding team in the nation (42.4 percent of missed shots). On the other hand, in beating a Sam Houston State squad that was completely gassed by the end of the game, the Bears forced just 11 turnovers, which is typical (17.7 percent of opponent possessions on the season, 309th in the NCAA). If both teams play to type this weekend, Old Dominion will maximize its possessions, and our model says that gives the Monarchs a chance against Baylor's highly efficient offense.

No. 6 Tennessee (23.1) vs. No. 14 Ohio (31.3)

UPSET CHANCE: 7.2 percent

The Bobcats shot an incredible 58 percent against Georgetown in the first round; that will be tough to repeat against the Vols' efficient defense, which ranks eighth in the NCAA. And our model doesn't see Ohio as particularly able to maximize possessions: The Bobcats are adept at minimizing turnovers, but Tennessee is even better at generating them, and while Tennessee is weak on the offensive boards, Ohio is even weaker on the defensive boards. The Vols probably won't be able to shoot well from the outside (31.8 percent 3-point FG shooting, 267th in the country, versus Ohio's 29th-ranked defense). But that probably won't matter unless the Bobcats can grab some steals and turnovers and keep the Vols' big men cold for long stretches.

No. 2 Villanova (<2.0) vs. No. 10 Saint Mary's (80.8)

UPSET CHANCE: Less than 2.0 percent? *

This one we have to give an asterisk. Not because we want to move away from objective analysis toward subjective territory. But because our model, although it's as good as we can build it, can't account for a handful of factors, and this game involves three of them. First, we use pre-tourney stats, but sometimes a team's performance changes so drastically that there's reason to think its overall numbers (which include early-season games) no longer reflect its true talent. Injuries are an obvious reason this can happen (as we already addressed with Purdue). But so might be whatever the heck is happening to Villanova, which is now 3-5 over the past month, and which labored mightily to beat Robert Morris, a team with the 203rd-best offense in the country, in the first round. Second, our model cannot distinguish between high-scoring big men who can be neutralized easily and those who can dominate even tough competition, but the fantastic Saint Mary's center, Omar Samhan, who dropped 29 points and 12 boards on Richmond, certainly seems to be the latter. Finally, in the years our model is looking at (2004 through 2009), Giants with vulnerability scores as low as Villanova's have always won their matchups; hence the reason for the incredibly low upset percentage. But none has ever faced a Killer with a rating as high as the 2010 Gaels, who rank sixth in the NCAA in 3-point shooting, don't turn the ball over and rebound well. The math says the Cats should be safe, as we have noted. But Omar and the Aussies may just push our model into new territory in their first GK game (their first-round game didn't count since the seed differential was only three). We'll find out soon enough.