Gonzaga’s Opponents Will Be Better Than They Look Right Now: Simulating the Rest of the NCAA Tournament

Since their brackets were unveiled, we’ve been simulating both the NCAA Tournament and the NIT 10,000 times per day. The ratings we use in these simulations start with KenPom, but we run the simulations “hot,” meaning if a team impresses in a certain simulated game, their rating improves for the next game in that specific simulation, and vice versa. Our aim with this is to adequately reflect the reality that ratings will change throughout each tournament.

We thought it’d be interesting today, as we take a breath and the NCAA Tournament resets, to look at where each team stands when it comes to reaching the Final Four and winning the national championship. Per our simulations…

To make the Final Four:

• Kansas: 54.3%
• Gonzaga: 50.5%
• UCLA: 41.4%
• Purdue: 38.7%
• Houston: 33.9%
• Arizona: 30.4%
• Villanova: 27.5%
• Texas Tech: 22.4%
• Duke: 19.9%
• North Carolina: 17.5%
• Iowa State: 17.4%
• Miami: 14.3%
• Providence: 14.0%
• Michigan: 8.2%
• Arkansas: 7.3%
• Saint Peter’s: 2.5%

And to win the tournament:

• Gonzaga: 21.8%
• Houston: 11.2%
• Arizona: 10.2%
• Kansas: 10.2%
• UCLA: 8.7%
• Texas Tech: 8.1%
• Villanova: 7.8%
• Duke: 6.5%
• Purdue: 6.2%
• North Carolina: 2.3%
• Arkansas: 1.6%
• Michigan: 1.5%
• Iowa State: 1.4%
• Providence: 1.3%
• Miami: 1.2%
• Saint Peter’s: 0.1%

A few takeaways from these numbers:

• Kansas’s advantage via path is jarring when looked at through these numbers. There’s an element, yes, where this is an in-tournament development. The 2, 3, 5, and 6-seeds have all been eliminated from the Midwest Region. But at the same time, Kansas drew far and away the worst 3-seed and 4-seed in their region, their 2-seed was suspect, and their 6-seed was a team that had just fired its coach. This can be fair—Wisconsin and Providence had great résumés, even if they weren’t rated highly by systems like KenPom—but it’s still noteworthy, and what’s striking is not Kansas’s high probability of reaching the Final Four so much as the gap between that and their probability of winning the title. Ditto Providence. Ditto Iowa State. Ditto Miami. That region is weak.
• Houston and Arizona are, per KenPom, the second and third-best teams in the country, and they’re playing on Thursday in the Sweet Sixteen. That’s a big game! Given that Houston’s only slightly better than 50% likely to beat the Wildcats, it stands to reason that a win, independent of other results, would double Houston’s championship probability. Were Gonzaga and Houston’s game times flipped, there’s a chance we could see Houston briefly favored instead of Gonzaga. Instead, it’s a situation where if Gonzaga loses, or if Gonzaga loses but Texas Tech or Duke beats the other so impressively that Houston or Arizona’s path is now demonstrably easier, the winner of Houston vs. Arizona could exit the weekend in front.
• There are a few clear gaps in championship probability, and a few gradient spaces. Houston and Arizona and Kansas have comparable chances. So do Arkansas and Michigan and Iowa State and Providence and Miami. Elsewhere, it’s less clumpy, but that said, Purdue is nearly three times as likely as UNC to win it all, with no one in between them. It’s more likely than not a nine-horse race right now, viewed generously (I’d put it closer to a three-horse race, given what we know about where the winner of Arizona and Houston will stand). Seven teams are very much outsiders, including Carolina.
• Compared to Ken Pomeroy’s own system of forecasting single-elimination tournaments, our simulations are low on Gonzaga, low on Kansas, low on Purdue, and high on everybody else. What’s happening here is that our model looks at Gonzaga’s hypothetical Elite Eight opponent not as present-day Texas Tech or present-day Duke, but as the Texas Tech or Duke team that will have beaten a very good opponent (present-day TTU or present-day Duke) to reach Gonzaga. There are plenty of simulations where the game is close enough that the ratings don’t move much, but given how close the line is, it’s hard for the winner of a game like that to have underperformed. Contrarily, Gonzaga or Purdue could play below expectations and still move on to the Elite Eight. Should running the simulations hot have this much of an impact? I’m not sure. But Houston’s risen from 5^th in the country to 2^nd in KenPom over the last eight days, and Saint Peter’s has risen from 121^st to 101^st over the last nine, and Miami was 55^th yesterday and now they’re 42^nd.
• The chalkiest route from here, by KenPom or our own modified KenPom, is for Gonzaga to beat Texas Tech in the West Regional Final, for Houston to beat Villanova in the South, for Kansas to beat Iowa State in the Midwest, and for UCLA to beat Purdue in the East. From there, it’s for Gonzaga to beat UCLA, Houston to beat Kansas, and Gonzaga to beat Houston in the title game. If you trust betting markets more than KenPom (there’s a case to be made for either in this situation), you can flip Purdue for UCLA, Arizona for Houston, and Miami for Iowa State, though there’s plenty of room for lines to still move, and we aren’t one hundred percent certain the market has Purdue better than UCLA.
• Finally, our last point: Gonzaga is far from inevitable. Even Pomeroy’s more Gonzaga-optimistic method has the Zags as less than three-in-ten likely to win the championship. You have a better chance of losing Russian roulette with two rounds in the cylinder. And just to make the Final Four is hardly a coin flip for even a team as good as Gonzaga or a team with Kansas’s advantages (also, Kansas is good—we aren’t trying to knock Kansas here, they’ve just caught a break). This thing is, by nature, wide open. That applies this year as much as any.