Commentary

Giant Killers: Defining our task, tactics

Originally Published: March 11, 2009
By Jordan Brenner and Peter Keating | ESPN Insider

Cut to the chase: Just show me the lists!

Sure, the Final Four is the preferred destination for every college hoops team, but the opening weekend of the NCAA tournament is where you'll find the real action. Why? Upsets. Pulse-pounding, system-shocking, eternally memorable upsets. Princeton over UCLA. Valpo over Ole Miss. Hampton over Iowa State. For the fourth straight year, we've endeavored to make sense of these seemingly random acts of Madness, attempting to systematically and scientifically identify upsets before they happen.

The goal is to uncover the latest crop of Giant Killers. A Giant Killer is a team that beats a tourney opponent seeded at least five spots higher. Squads from the six BCS conferences are ineligible, as are Butler, Gonzaga, Memphis and Xavier -- they're not sneaking up on anybody. And, obviously, a Slain Giant is a team that loses to a GK. Over the past five seasons, 21 Killers have knocked off 25 Giants (with a handful of GKs claiming multiple victims). So, by our rules, No. 10 Davidson's first-round win over No. 7 Gonzaga wasn't a GK victory this past year, but Davidson's subsequent wins against No. 2 Georgetown and No. 3 Wisconsin qualified.

[+] EnlargeStephen Curry
Daniel Plassmann/US PresswireStephen Curry and the Wildcats crossed up two Giants in 2008, and almost felled the eventual national champions.

While we've had success in the past, identifying George Mason in 2006 and Winthrop in 2007, we've entered a new era this season by expanding and refining our approach. Now, we'll not only show you the top Killers, but also the Giants that are ready to crumble, too. And we've enhanced our methodology, creating a model that estimates each team's percentage chance of becoming a Giant Killer or a Slain Giant. That's the result of examining dozens of stats for every team that could have fallen into either category over the past five tournaments, from how many three-pointers they allowed to their record in close games to the number of seniors they started.

We looked for, and in some cases mathematically devised, stats that correlated strongly with spectacular success or failure. And then we performed a multiple regression analysis, which basically is a way to tell how strongly each part of a group of inputs (in this case, team stats) affects an output.

Here you'll find a list of potential Giant Killers (which will be pruned as teams are eliminated from tourney contention) and their percentage chance of pulling off a major upset, followed by projected Giants and their odds of an early exit. We'll update this list as teams' profiles change. Keep in mind, though, that these rankings have been calculated with one specific purpose -- to predict upsets. This won't help you figure out how Wake Forest might handle Kansas or how Cornell would fare against Northern Iowa. But over time, we've learned that teams on either side of a bracket-busting matchup are characterized by particular traits.

Traditional basketball stats often obscure the hidden gems that are Giant Killers, though. So we relied heavily on Ken Pomeroy's advanced metrics, which offer two key improvements. First, he adjusts for tempo, converting point totals to points-per-100-possession averages. It turns out that a lot of what we typically think about teams' scoring effectiveness comes simply from the pace they keep. In 2006, for example, Kansas averaged 75.2 points per game, while George Mason scored 69.4 ppg. But the Patriots were playing a slower game, and, at 107.8 points per 100 possessions, they were actually more efficient than the Jayhawks (106.8). Pomeroy also adjusts for teams' strength of schedule. To which all we can say is, thank God someone is doing that math.

So what did these next-level stats reveal? Among the most significant discoveries, Giant Killers have:

• High adjusted offensive and defensive efficiencies.

• Low turnover percentages.

• High offensive rebound percentages.

In short, they not only score many and allow few points per possession, they maximize their number of chances, too. That's why Niagara currently tops our chart -- the Purple Eagles excel at avoiding turnovers, are monsters on the offensive boards and are efficient at both ends of the floor.

Conversely, Slain Giants are inefficient and sloppy, at least compared to other top seeds. Florida State, for instance, features an adjusted offensive efficiency well below average for big-conference teams. The Seminoles' 22.9 percent turnover ratio is the highest of any big-time program. And more than 30 percent of the field goals they allow are three-pointers. These are all severe red flags in our model.

One other stat stands out for both Killers and Giants, and it's somewhat counterintuitive because it correlates with long-term success. Specifically, GKs are less reliant on free throws than potential killers that don't pull upsets -- and also less than Slain Giants. This past year, for example, Western Kentucky, which upset Drake, smashed its opponents by nearly 11 ppg, but actually got outscored by 156 points at the charity stripe. Meanwhile, 63 percent of Wisconsin's scoring margin over its opponents consisted of free throws, 61 percent for Drake, 54 percent for Vanderbilt. All three schools went down to GKs.

Apparently, it's easier to neutralize an attack that depends on drawing -- or preventing -- fouls. And that bodes well for a team like Davidson, if the Wildcats grab an at-large bid (12.9 ppg differential, -2.9 free throw margin per 100 possessions) and poorly for Xavier, Oklahoma and Syracuse. And if you really want to take a flyer on we call the "Magic Bullet" theory, check out Connecticut. There's no denying how effective the Huskies are at both ends of the floor: For every 100 possessions, they score 115.6 points and give up just 85.4. But 72 percent of that gap is due to their ability to get to the line while preventing other teams from doing the same. The Huskies have shot an incredible 426 more free throws than their opponents this season. If UConn runs into one cranky ref or smart defense or mid-major star who needs to be stopped by hacks, the team could go down hard.

Of course, our system isn't perfect at predicting NCAA upsets. Even the best teams can lose on any given day, and our spreadsheets are still scratching their heads over how Duke spit the bit against VCU in 2007. Besides, we wouldn't want to ruin every surprise ending, anyway. But if Florida State has to play Niagara in the first round, look out.

Now, let's go see how it all applies. Check out the list.

Jordan Brenner | email

ESPN The Magazine contributing writer
Brenner writes for ESPN The Magazine and ESPN Insider. He covers the NBA and college basketball.
Peter Keating is a senior writer at ESPN The Magazine, where he covers investigative and statistical subjects. He started writing "The Biz," a column looking at sports business from the fan's point of view, in 1999. He also coordinates the Magazine's annual "Ultimate Standings" project, which ranks all pro franchises according to how much they give back to fans. His work on concussions in football has earned awards from the Deadline Club, the New York Press Club and the Center for the Study of Sport in Society.