In ThePowerRank.com's projection, No. 2-seeded Kansas has a better title shot than some No. 1 seeds. (Jeff Moffett/Icon SMI)
There are a lot of folks who are dabbling in predictive projections of the NCAA tournament these days, but I'm not sure I've seen an output that's more visually compelling than the one put together by ThePowerRank.com. It's equal parts addictive and seizure-inducing. Don't say I didn't warn you before you head down the wormhole.
First off, if you played arcade games in the 1980s, this setup reminds me a lot of Tempest (h/t @cwyaco for remembering the name of the game). I just kept moving around the edge of the bracket, wanting to shoot opponents as they came from the national title game core.
After that quick flashback, I tucked into the numbers, which were intriguing in some spots: None of the 1-seeds advanced past the 16-seed more than 90.2 percent of the time. A win rate in the high 80s on a neutral floor is nothing to sneeze at, but it struck me as a couple points lower than what previous seasons would have projected. Is the data suggesting that 1-seeds indeed are more vulnerable this season than in the recent past, or am I having some sort of bias in thinking that number would have been higher on average in other years? Gonzaga has the best chance to win its first game, and more than an eight percent chance to win the national title coming out of a 1-seed. In this projection, Kansas is a 2-seed and has a better chance to win the national title than some of the 1-seeds. That said, none of the other three 2-seeds (Michigan, Georgetown or New Mexico) were above 3.5 percent to win it all. That's a significant dropoff from the 1-seeds in a year where very few people perceive a quality difference of that magnitude among the top eight teams or so. A lot of the difference comes in drawing a 15-seed instead of a 16-seed in the Round of 64. All three of those less-regarded 2-seeds have at least a 1 in 5 chance of losing their first game, according to this model. VCU, coming out of a 7-seed, has a 1.3 percent chance of winning the national title. Just warning you in advance.
What do you think about this predictive model or the layout itself? Addictive? Silly? Hit me @AndyGlockner on Twitter.