A spin through the bracket to find the Giant Killers

It's a new season, with a new Giant Killers model.

In our never-ending quest for the perfect bracket -- or at least one better than your coworkers' -- we are diving into the numbers to find the edge and, ultimately, the teams that will surprise the nation (but hopefully not us).

The base of our model this season is ESPN's Basketball Power Index (BPI), the system that powers many of ESPN's college basketball predictions. But we're taking it a step further, because college hoops isn't just about how good one team is overall compared to another. It's about individual teams' strengths and weaknesses and how different teams match up with one another. And so we've incorporated that into our new model with a Giant Killers-specific wrinkle.

These advantages are derived from box score statistics that are adjusted for pace and opponent strength. Adjusting for opponent strength is critical in a sport with 351 teams and a wide range of talent.

Here's how it will work in practice: We start with our base-level expectations of a Giant Killer's chance to win based on BPI. But that number is then adjusted up or down depending on how the Giant Killer matches up against the Giant. That adjustment is what we'll call the "GK Factor." So it will work like this: BPI + GK factor = Giant Killer's chance to win.

Where can the biggest advantages be found over an opponent? For starters, it depends on how big the edge is in all the various categories. But it also depends on how each team's respective strengths and weaknesses interact. A team might not be very efficient shooting, but has a big advantage on the offensive glass. Given the right matchup, that could work, just as it did for former Giant North Carolina in 2017.

By and large, though, the statistics that we found to be most predictive of an advantage were ball control (fewer turnovers), offensive rebounding, and 2-point and 3-point shot volume and efficiency.

Let's get into some hypothetical matchups from different rounds of the NCAA tournament using Joe Lunardi's latest Bracketology.

Second round: (11) Notre Dame vs. (3) West Virginia

Notre Dame's BPI chance to win: 34 percent
GK Factor: Plus-6 percent

This matchup projects Notre Dame beating its first-round opponent to earn a matchup against Bob Huggins' defensive powerhouse. The Irish would be underdogs, but there are areas where they would have the edge.

It basically boils down to shooting, where Notre Dame is a shade better from 2, a tad better from beyond the arc, and a decent bit better from the free throw line. It doesn't hold the advantage in ball security (which would be a challenge against West Virginia), but it's not bad in that department, either. All in all, the model gives the Irish a slight boost against the Mountaineers, which might be all we need to make them a value pick.

One more thing to possibly like about this pick: Bonzie Colson. The All-America forward suffered a foot fracture in early January and it's unclear when he will return. If Colson is ready to play come March, it means that BPI -- and our Giant Killers model -- could be underrating Notre Dame because it doesn't know Colson has missed time (six games so far), and yet thinks the Irish would have a good shot in this matchup anyway.

First round: (11) SMU vs. (6) Seton Hall

SMU's BPI chance to win: 51 percent
GK Factor: Minus-1 percent

This matchup actually puts the Giant Killer at a disadvantage compared to BPI's projection, but that in no way disqualifies SMU from this exercise. Even with the negative GK Factor, our model thinks SMU is a favorite in this matchup by the slimmest of margins.

In this case, the advantage gained isn't by any particular matchup -- it's because SMU is just the better team. Shake Milton, who scored 33 points in the Mustangs' upset of Wichita State, has contributed the fifth-most win shares among all Division I players. It's worth noting that one of SMU's other key contributors, Jarrey Foster, is out for the season with a partially torn ACL.

To give you a sense of how much the matchups could swing things: SMU is far from Seton Hall's worst nightmare in this matchup. If Virginia Tech, a similarly strong team relative to SMU, according to BPI, were to play Seton Hall, the Hokies would have a GK Factor of plus-21 percent and an overall chance to win of 73 percent. Fortunately for the Pirates, Virginia Tech is currently on the outside looking in.

Elite Eight: (10) Missouri vs. (1) Virginia

Missouri's BPI chance to win: 14 percent
GK Factor: Plus-17 percent

This would be quite a run for Missouri (and probably without Michael Porter Jr., too). The Tigers would still be solid underdogs against Virginia, although their chances are not too shabby against a No. 1 seed.

As unlikely as this scenario is, it presents an interesting matchup with a big GK Factor. So what's making Virginia so vulnerable against Missouri? For starters, it's all about the boards. Missouri simply rebounds better than Virginia. The Tigers also hold the edge in free throw shooting. And although the Cavaliers are more efficient shooters, the Tigers shoot more 3-pointers (in lieu of some 2-pointers), which helps as an underdog.

Second round: (8) Saint Mary's Gaels vs. (1) Duke Blue Devils

Saint Mary's BPI chance to win: 28 percent
GK Factor: Minus-4 percent

Saint Mary's might not strike you as a serious Giant Killer with just over a 23 percent chance (after rounding), and in normal circumstances, it might not be. But compared to other teams with a No. 6 seed or worse, 23 percent against Duke is one of the best options out there.

The Gaels are no slouches. We'd give them a 53 percent chance to beat Oklahoma or a 41 percent shot against North Carolina on a neutral court.

Duke, however, is particularly tough. While Saint Mary's is a better 3-point shooting team, that isn't enough to boost the Gaels' GK Factor into a positive number. The Blue Devils, to no one's surprise, are more efficient on 2-pointers, rebound better, and hang onto the basketball better than Saint Mary's. It's almost as if they're one of the best teams in the country.