clock menu more-arrow no yes

Filed under:

NarbTom Bracketology Recap

New, 6 comments

Now that the official bracket is out, it's time to analyze how our predictive measures worked and why Narber and Rev are better at bracketology than the committee itself.

Kyle Terada-USA TODAY Sports

Selection Sunday brings a myriad of emotions to college basketball fans everywhere.

Yesterday was no exception for Narber and myself, however, there were different emotions than in years past. I found myself thrilled with Iowa State's draw, which would typically make for a great day, but I also found myself irrationally upset about things like St. Bonaventure being left out, Tulsa getting in and basketball teams in the state of Oregon being drastically overseeded.

This exercise in bracketology was a test run for us here at Wide Right & Natty Lite, but it's quickly developed into so much more. For those not familiar with the practice, there's a site called Bracket Matrix, that ranks the accuracy of bracketologists each year...

And what do you know!

Your favorite bracketologists have ended up on the list (you'll have to scroll down to Tier 5, but we're there, I promise)!

Now that we have this upgraded street cred, we had to figure out where we would rank among the experts. Using the ranking system at the Matrix, we scored a whopping 292. For those who are anti link clicking, here's where that would rank in comparison to recent annual averages.

Year 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015
Average 284.4 285.9 309.1 303.6 303.5 311.4 327.2 323.5 330.3 326.5

So, as you can see, we're stuck in 2007-08, which is about right for a blog of this caliber, just getting our feet wet. Now, the Matrix analyzes 152 total bracketologists, 134 of which have completed brackets for this year.

We ranked 129th.

But you know what? I thought we did a better job than what these results show, so my anger quickly turned to the committee. If they actually used the criteria that they always say they will, we should be able to better than 292, right?

NarbTom is (tied for) the best

If you're looking to project 16 seeds, look no further than NarbTom. Our metrics correctly predicted all six of the 16-seeds. Don't give me this crap that only seven prognosticators incorrectly picked a 16-seed, just accept that we're the best and move along.

Your boys managed to correctly pick 64 of the 68 teams that made the tournament, correctly seed 24 of the 64 teams that we picked and, of the remaining 40 teams, we were within one seed of actual seeding on 28 of them. In the end, we only incorrectly seeded two teams by more than two seed lines: Oregon State and Gonzaga.

So who did we miss?

NarbTom Selections: Saint Mary's (10), Florida (10), St. Bonaventure (11), San Diego State (11)
Committee Selections: Temple (10), Syracuse (10), Tulsa (11), Wichita State (11)

This is screwed up in so many ways. Let's look at four ranking systems that the committee has been noted as heavily observing during their at-large selection process. By averaging these eight teams' rankings in terms of RPI, Sagarin, KPI and KenPom ratings, we get the following results...

This is crazy. Three of NarbTom's four should have been selected. Temple of all teams is an average of almost 13 spots behind the next best team (Tulsa), which is an average of 7.5 spots below the next team (Syracuse), which finally brings you up to Florida and St. Bonaventure range. IT'S NOT EVEN CLOSE.

Look, I know that Temple won the American Athletic Conference regular season title, but you know what that gets you with a résumé like that? A 1-seed in the NIT. Not only that, but they get to play two hours from home. Well done, committee.

We're notable Wichita State haters, which is fine, and I'm good with their selection as they're top 50 by almost every metric. But they have one single top-50 win: Utah at home. By contrast, NarbTom's lowest field member (St. Bonaventure) has three. What about top-100 wins? Wichita State has four, St. Bonaventure has seven. What are we even talking about here?

Our two biggest seed whiffs were Oregon State and Gonzaga — four seed lines off on each, and if we would have swapped them we would have nailed them both. With a bubble as vague as it was this year, all of the bubble teams could really be anywhere from 7-11. So we're really splitting hairs here, though it may not look like it. Either way, let's use the same metrics as we used above:

Whatever, I'm done here.

Here's one decision that doesn't matter much but just goes to show the ineptitude of the committee. You'll notice that all along we've had Kansas (the top overall seed) in the Midwest Region. Now before you go to Google Maps and check this,, we already know that Louisville is THREE MILES CLOSER TO LAWRENCE THAN CHICAGO.

Great job, Selection Committee. By getting KU three miles closer you've moved Virginia (who we can also dispute getting a 1-seed anyway) 245 MILES FARTHER AWAY.

So way to not do your job.

Not only that, but if chalk holds, Virginia has to play 2-seed Michigan State in the Elite Eight, in Chicago, while the Spartans are 518 MILES CLOSER TO CHICAGO THAN VIRGINIA IS. As Iowa State fans, we know a little something about playing Michigan State "on the road" in the Elite Eight.

So what did we learn?

We were completely naîve to believe that the NCAA would do anything with any sort of logic. I have no doubt we'll be better next year as we learn to game the system, but in the meantime, you'll have to live with reading the bracketology predictions of the 129th best prognosticators in the free world.