How to spot a dark horse in the NCAA tournament

We look to history to give you pointers on how to pick your dark horse in the upcoming NCAA tournament.

As we creep ever closer to Selection Sunday, it’s time for another bracketology primer. Today, we’ll be looking to past tournament results to help your bracket with the most unpredictable moments in March: upsets, and dark horse teams.

Though part of the beauty of March Madness is its unpredictability, that’s not something you want to be whispering to your $20 when you’re depositing it with your office bracket. And while I can’t offer you certainty, I can help point you to historical profiles of previous surprises and dark horses to help guide your hand as you move through the bracket. Filling out a bracket by favored mascots might be fun, but glory lasts forever. Let’s get to the analysis.

Much like some other articles, I’ll be using tournament results from the KenPom era — 2001-02 plus — along with available stats and characteristics of the team. This time, I’ll be using what was developed to identify common characteristics in teams that unexpectedly crashed the Sweet 16 – and beyond, to help guide your choices next week. For a fuller explanation of some of the methodology — especially expected wins — check out this article.

On to the analysis. Here’s how you find the right dark horse team to take over your bracket.

Mid-Major underrated because of a weak schedule

This is the Wichita State spot — a non-power conference team dismissed by the Committee’s seeding practices because of underwhelming scheduling.

Forget that metrics can adjust for strength of opponents and margins of victory to try and combat comparison problems in scheduling differences — PLAYED NO ONE. While the committee appears to be moving closer to using more KenPom-esque statistics to make its seeding decisions, we haven’t gotten there yet. And it certainly can’t be erased from history because it happens all that time.

Gonzaga pulled an 11th seed last year, despite an efficiency margin that marked it closer to a 6 or 7 seed. The Zags easily dispatched two major conference teams they were clear favorites over, by adjusted efficiency, before falling victim to Syracuse’s own magical dark horse run.

And they were nearly joined by 14th seed Stephen F. Austin, who missed out on the Sweet 16 by a single point. Though led by a legitimate National Player of the Year candidate (Thomas Walkup) and strong metrics (top 45), the Lumberjacks were seeded like a typical auto-bid team. They proceeded to made easy work of 3-seed West Virginia, and took Notre Dame to the wire before falling just short.

History smiles upon this category. Restricting teams to: non-Power 6 conference, underseeded by at least two lines, above a 5 seed, and with an overall schedule 80th or worst in the country, we get some interesting results.

Of the 25 squads that have met these criteria in the last fifteen tournaments, seven have made it to at least the Sweet 16. There are also two Elite Eight appearances (2008 Davidson and 2002 Kent State), and a Final Four run (2006 George Mason). That’s over a quarter of these teams who, mind you, have an average seed of 11.12 have made Sweet 16s. All other teams between a 6-seed and a 14th — the highest seed in this group — makes the Sweet 16 just 11.5 percent of the time.

The lesson? Don’t dismiss these teams in your bracket just because they haven’t played many quality opponents. They’re still good teams, and you could be missing out on the opportunity for a play that runs counter to popular opinion.

Possible candidates: Wichita State, Saint Mary’s, Middle Tennessee State

Underseeded Power Conference Team

These are by far the easiest to spot. Anyone with the KenPom website address and a bracket can identify a ‘Power 5’/Big East team that looks stronger than its seed.

And yet, it happens all the time. In 2015, a Michigan State squad with a 23-11 record going into the NCAA tournament garnered a 7th seed. Their adjusted efficiency pegged them as the 16th best team in the country,  a margin more in line with a 4- or 5-seed. Sure enough, Tom Izzo’s crew crashed the Final Four, taking out three higher seeds along the way.

And lest you think this is just the Izzo magic talking, Tennessee pulled the same trick in 2014. Despite an adjusted efficiency, per KenPom, that ranked 14th in the country before the tournament, Cuonzo Martin’s team earned an 11-seed. They promptly trampled their way to the Sweet 16, before falling to Michigan.

While these teams are easy to spot — and you should certainly be aware of them — analysis doesn’t show that they’re any more likely than other low seeds to crash the Sweet 16. Major conference teams — seeded between 6 and 12 — that landed at least two lines, or 8 spots, below their adjusted efficiency rank made it to the Sweet 16 or farther 16 percent of the time. This is slightly better than the population at large; about 15 percent of all other 6th through 12th seeded teams make it to the Sweet 16. But it’s not a significant difference.

While you should be aware of teams like this, they’re not much better bets than the field. Don’t fall too hard, especially as these type of teams are likely to be among the most popular sleepers. Let’s dig deeper.

Possible candidates: Oklahoma State, Michigan, Vanderbilt, Marquette.

The Thunderbolts

This section will be on thunderbolts — or a dark horse no one saw coming. A team rising from an unspectacular resume to crash the gates in March. It’s one of the best parts of the dance, but it’s also a pain in the ass from our perspective; how do you predict the unpredictable?

Who can forget Dunk City’s run as the 15-seed past Georgetown and San Diego State? Florida Gulf Coast’s record and efficiency profile hardly suggested they’d become the first 15-seed to ever reach the Sweet 16. But two double digit wins over higher seeds — and a ton of dunks — later, and they were forever entered into tournament lore.

And VCU’s surprise run from First Four to Final Four will be the standard for Cinderellas for years to come. Behind a constant press and an offense that (unsurprisingly) thrived in transition, VCU bedeviled opponents until finally falling to Butler in an all mid-major Final Four game. Despite being the underdog from an adjusted efficiency perspective in each game, the Rams logged four double digit victories en route.

From an aggregate perspective, this is a dangerous bet to make. The type of runs we saw from VCU and FGCU, or 2013 Ohio, 2010 Cornell, and even 2013 La Salle, are inherently rare events. Teams that deserve their seed — or are worse than it — don’t usually perform particularly well. But nailing these teams can set your bracket apart, and give you a place among the truly prescient.

So let’s look for some common features among these teams. For these purposes, I separated the 10 teams with an adjusted efficiency ranking below 50, prior to the tournament, that made the Sweet 16. I looked at some common characteristics in the statistical profiles of each team to try and identify the best markers to look for in these thunderbolts.

The first that pops out is 3 point attempt rate. It’s long accepted wisdom that the 3-pointer is an excellent tool in a ‘David’ strategy; underdog teams should try to use the 3-pointer as an equalizer, or high risk, high reward strategy, against teams with more talent than them. It’s no surprise, then, that four of the ten teams in this sample came into the tournament with 3PA rates in the top 60 of the country. A fifth landed in the top 90. And the same amount landed in the top 80 of the country, in their respective years, when it came to actually converting those 3s. An ideal candidate in this category will probably take and make 3s at a good clip.

A second is the steal rate. There’s incredible inherent value in generating steals — it ends a defensive possession while giving up no points, and leads to a great opportunity on the other end for your team. Seven of the 10 Sweet 16-crashing David squads had steal rates in the top 100, and only 2012 Xavier managed the trick without a rate in the top half of the country.  VCU may be the most famous example, but Florida Gulf Coast and Ohio, in 2012, were both excellent at creating steals. It’s another key tool for the ideal ‘Thunderbolt’ candidate.

And finally, an underrated little gem: luck! While it’s impossible to measure luck in an exact way, KenPom does an excellent job at approximating it, mainly by measuring how well a team does in close games, and how good its record is vs. its point differential. While this is something I’ve seen previously in analysis, ‘luck’ played an important factor for each of our underdogs. Every team in this group of ten, other than 2007 Vanderbilt, rated well above average in the ‘luck’ category. While winning in close games does often come down to luck, there’s something to be said for teams that are used to being in close games come tournament time, as it’s a situation they’re already used to. The little margins often count in March, and bringing that end-game experience can really matter.

Possible candidates: Dayton, Michigan, Princeton, Jacksonville State