A guide to picking NCAA Tournament upsets

We all want to do well in our bracket pools and look like experts. It’s easy to go chalk and guarantee yourself at least a pretty good showing, but we all know that people who win the pool are the ones who hit on a few big upsets. With that in mind, what should you be looking for in picking an upset, and should KU be worried about any teams in their bracket?

In the KenPom era (since 2002), there have been 116 “big” upsets in the NCAA tournament. For my purposes, “big” tournament upsets are those in which the team that wins was seeded 5 or more lines lower than the “better” seed in any round. That works out to about 7 or 8 per year through all rounds, which means it’s a pretty frequent occurrence.

But is there a way to foresee these tournament upsets? I took a look at these 108 “big” surprises to see what predictive factors I might be able to find. Here is a list of observations you might want to keep in mind as you put your finishing touches on your bracket:

• KenPom rank matters. In 36% of the upsets I found, the winning team finished the year with a KenPom rank within 10 spots of the favored team. 11% of the time, they were actually ranked higher. If you see two teams with a big disparity in seeding but a small disparity in KP ranking, you might consider going with the underdog.
• Look out for turnovers! On average the teams getting upset had higher offensive turnover rates than the victors, while the underdog turned opponents over more often than the favored team. That might seem like a HUGE indicator at first glance, but we should tap the brakes a bit. Lower, non-at large seeds are typically teams that won (and in some cases, bulldozed through) a low or mid major conference, where they faced inferior competition. Even with this in mind, it seems noteworthy that a lot of the teams that pull of big upsets are teams that force a lot of turnovers (69% of the winning underdogs had opponent TO rates over 20% in their conference schedule, well above average).
• 3 point hoisting underdogs are, in fact, dangerous. Victorious underdogs in the games I looked at, on average, shot just over 3% more of their shots from behind the arc than their opponent. What’s interesting is that the underdogs were still at a shooting percentage disadvantage compared to the favored team. The underdog didn’t need to be a better 3 point shooting team, but shooting more of their shots from behind the arc helped (as long as they happened to make them that day). Beware picking teams based on 3PA%, though. Jump shooting outcomes contain a dangerous amount of variance. Some nights, it helps the teams that put up a bunch of threes if fortune is on their side, but there are plenty of times where it hurts, if they aren’t consistently making them.
• Tempo does not seem to be a factor. There was a lot of variance in the pace of both the upset-ers and the upset-ees. It came out roughly even in the end, with the upset-ees playing at a slightly faster average pace than their opponents. There is some thought that one should stay away from picking slow-paced tournament teams because it takes them longer to mount a comeback, should they fall behind, but a converse school of thought saying slower teams limit possessions, making an unlikely upset more likely by limiting the sample. This was not an area of significance in my observations.

These tips may not make or break your bracket, but they should give some guidance as to which teams are prime for an upset. Look out for underdogs who force a lot of turnovers (looking at you, East Tennessee State and Oklahoma State), put up a lot of threes (I see you, Winthrop and Vanderbilt), and matchups where KenPom, Sagarin and PSR (see what I did there?) don’t see much difference between the opponents (Oklahoma State vs Michigan comes to mind, as does Marquette vs South Carolina and Maryland vs Xavier).

Since we’re a KU site, I went ahead and perused the Midwest bracket to see if anyone met these criteria. Here’s what I found:

• Although a lot of people seem concerned about a possible North Carolina Central matchup in the first round, there’s nothing particularly concerning about them (or at least not concerning enough to see them taking out a one seed). They force a moderate amount of turnovers (just below the 20% threshold at 19.6% and only 18.6% in conference play). They did shoot 40% of their shots from behind the arc in MEAC play, but hit just under 34% of them, so they weren’t exactly scorching the nets.
• Neither Miami or Michigan State look especially troublesome from a generic standpoint. Miami forces few turnovers and shoots very few threes. Michigan State is terrible at turning teams over and shoots a very average number of threes. Neither is very close to KU in the computer rankings.
• Iowa State hits on all categories as a possible Sweet 16 matchup. They’ve beaten KU once, and they’re only seven spots behind Kansas in KenPom’s rankings. They also forced turnovers on over 21% of defensive possessions in Big 12 play, and as for three point shooting...well, we all remember what happened in Lawrence. That could be a landmine game if both teams make it there.

Best of luck to you and your brackets, and we’ll see where the first upset is Thursday!