Giant Killers: Madness upset method

• Recommend0
• Comments0
• Email

Ronald Martinez/Getty ImagesSeeking Cinderella: Who could be this season's VCU? We've got some ideas.

Admit it. You've been taking virtual tours of Murray State. You've asked friends to quiz you on the spelling of Valparaiso. You've wondered aloud whether Oral Roberts ever, in fact, owned a Golden Eagle. And you've examined stats. Lots and lots of stats.

In other words, you're ready for Giant Killers season.

Well, back at GK Central, we're busting our tails (not to mention our spreadsheets and calculators) to zoom in on this year's crop of Cinderellas well in advance of the NCAA tournament. From now until the last slingshot is fired in March (or, if someone can follow in VCU's footsteps, April), we'll refine our secret upset-spotting sauce, then slather it all over this year's potential Killers and their would-be victims. We can sense you slobbering all over your keyboard already.

But in case you're new to our world, or are suffering from traumatic memory loss, we'll offer our annual refresher course.

A Giant Killer is a team that beats an NCAA tournament opponent seeded at least five spots higher in any round. Squads from the six BCS power conferences are ineligible to be Killers. We also exclude Butler, BYU, Gonzaga, Memphis, Temple, UNLV and Xavier due to current or historical success. (Note that mathematically, any team seeded fifth or higher in its bracket, as Wichita State might be this year, cannot be a Giant Killer. And if you think another team should be exempt from GK status, explain why in the comments section.) A Slain Giant, on the other hand, is simply a team that loses to a Giant Killer.

Take VCU as an example from last year. When the Rams (an 11-seed) beat USC in the First Four and Florida State (a 10-seed) in the Sweet 16, those weren't GK wins. But VCU's upsets of No. 1 Kansas, No. 3 Purdue and No. 6 Georgetown all met GK criteria.

MORE GIANT KILLERS

GK Blog

Stay updated with upset info

Methodology

How our stats-based model works

History Lessons

Explaining VCU's Final Four run

Last year was a slow one for Giant Killers. Only three GKs emerged, and they took out five total Giants (because of VCU's three separate wins). That brought the overall total since 2004, the first year of our data, to 32 Killers and 39 slain Giants. The basis of our research, then, relies on figuring out what characteristics differentiate those 71 teams from the squads who failed to pull off major upsets or the Giants who avoided an early exit. A number of similarities have emerged, and in a nutshell, successful Giant Killers tend to play a higher risk/reward game than their losing mid-major counterparts.

That means they tend to excel in at least one of a few key areas. They take a lot of 3-pointers and connect at a high percentage. They force a lot of turnovers and protect the ball when they have possession. They dominate the offensive glass and hold their own on the defensive boards. The best Giant Killers embody many of these qualities, and, when you think about it, it makes perfect sense that those strategies lead to upsets. When everything clicks, a team that plays this way generates a ton of extra possessions (and maximizes them with the 3-ball). And all it takes in the tourney is that one good day to move on to the next round. So yes, these same teams are also more likely to get blown out than one with a more conservative approach. But who really cares whether you lose by 10 or 30 in March?

Similarly, high seeds that are more vulnerable to upsets tend to have common traits. They struggle in the same key areas where Giant Killers thrive: turnover margin, offensive rebounding and 3-point defense, in particular. And they also tend to rely too much on free throw shooting for their scoring margin, and not enough on 2-point buckets.

Traditional box scores don't reveal these hidden truths, though. And that's why we rely on the tempo-free stats at Ken Pomeroy's website to help enhance our model. Simply put, if you don't account for schedule strength and pace of play, you're not getting an accurate read of a team's true strength.

Fortunately, our model has turned out to be a finely tuned predictive tool. From 2004-10, it says there should have been 28.9 Giant Killings. There were 29, and it also nailed each of the likeliest Killers in those given years. As soon as the data finishes simmering and we can take a similar look back at last season, we'll update those numbers. (You can stay tuned in with the Giant Killers Blog.) But we can tell you what we wrote about Richmond against Vanderbilt in the first round last season: "If our model had pockets, it would empty them on this one. If it had an ATM card, it would max out the withdrawal. If it could play poker … you get the idea. This is your best Giant Killer bet of the first round." And we know it was a big fan of VCU against Georgetown, too.

So stay with us over the next few weeks, as we feed you all the info you'll need to breeze through your bracket. Let Killing season commence.

Want more Giant Killers? Follow Jordan Brenner (@jordanbrenner) and Peter Keating (@PKStatsBlog) on Twitter. Jordan Brenner is a regular contributor to ESPN Insider and ESPN The Magazine. You can read his online archives here.

• Recommend0
• Comments0
• Email

Jordan Brenner | email

ESPN The Magazine contributing writer

• Archive

Brenner writes for ESPN The Magazine and ESPN Insider. He covers the NBA and college basketball.