For most of us, it's the most practical question sports analytics will ever answer: Which upsets should we pick in our NCAA tournament brackets?

Over at the Giant Killers blog, you'll find an answer I've worked on and believe in, a system that examines past NCAA upsets to find traits shared by winning underdogs and by fallen favorites. Every year, Jordan Brenner and I, assisted by Tom Haberstroh, wade through swamps of statistics for every program in the country to find the factors that correlate most strongly with surprising success. And over time, we've come to appreciate that most "Giant Killers" are either significantly better than traditional basketball statistics make them appear, or good at employing high-risk/high-reward strategies that increase the variability of their scoring. (To understand why that matters, see this ESPN The Magazine story.)