I was a bit bored on Sunday and decided to take a deep dive into Kenpom, looking at some of the team’s in the Atlantic 10. In particular, I found myself on Duquesne’s page. It’s hard to believe that this 18-9 team ranks just 156th in Kenpom, but some of its wins have been incredibly hard-fought and by small margins. Kenpom’s rankings are largely reflected by margin of victory, and particularly, covering spreads. While Duquesne hasn’t been great at doing that this year, they’ve been great at what matters more: winning. Despite the odds being stacked against this team in countless games this year, Keith Dambrot has found a way to will the young Dukes to victory. Luckily, Kenpom gives win probability insight into each game at every scoring change throughout the contest. I found a shocking result.
Basically, there have been points in 6 games this season where Duquesne only had a 2.1%, 3.1%, etc. chance of pulling off the victory, and in these 6 contests, the Dukes ended up winning. Said another way, if you could simultaneously revisit every point in time where Duquesne was down big and seemed destined to lose, and figure out the odds of Duquesne winning all of those games, the chances are 8 * (10 ^ -9), or in other words, an incredibly small number.
I wish there was a way to see how compelling this actually is, because taken in a vacuum, I feel a bit skeptical. I could hypothetically revisit every game Duquesne has played and won this year, find the minimum probability of winning during the course of the game, and factor those odds in as well. I could do that for every team in all of Division 1 college basketball, and I’m guessing there would be some crazy small numbers as well.
However, I’d be very surprised if there was another team in college basketball that had less than a 10% chance of winning at a given point in 6 separate contests and came back to win all 6 of those games.
Kenpom goes as far as to rank the top 100 games of the year in which a team with the lowest win probability ended up coming back and winning. For example, Marquette’s 106-104 overtime win over Creighton ranks 1st, as the Golden Eagles had a 0.1% chance of winning in the final seconds but pulled it out nonetheless. Duquesne has one game in the top 100 this year, as the Dukes had a mere 2.1% chance of beating Illinois Chicago with a little less than 4 minutes to go in regulation. Alas, the Dukes pulled it out.
While this “comeback” feature is a fun tab to explore (I’d highly recommend paying $20 a year for a Kenpom membership, by the way), I’m not sure this captures the entire story for Duquesne. Most of the Dukes’ comebacks this season have come after being down big (15+ points at some point in the second half). When surfing through the lowest win probability games, a lot of teams made a 5 or 6 point comeback in the final 20 seconds or so, which is certainly unlikely and compelling, but not exactly the same story as Duquesne. The Dukes have dug themselves out of massive holes this season, and yes, late-game theatrics have played into some of their impressive wins, but it’s been more of a steady comeback than a quick spurt of 6 or 7 points in the final 20 seconds of games. I think it’d be interesting to filter the lowest win probabilities to only capture those with more than 5 minutes to play in the contest — this may be a better measure of massive comebacks rather than a late game meltdown from an opponent before the buzzer.
Regardless, this has been a pretty cool story to follow this year, and the numbers give good indication as to just how unlikely some of Duquesne’s wins have been this year. While I’m no expert in working with college basketball data, I enjoy diving into Kenpom’s site and seeing what I can unearth. The “1 in 100,000,000” stat is certainly insane to think about, but I’m wondering if there’s more to the story or at least a more convincing argument as to just how improbable Duquesne’s season has been — from a mathematical standpoint. I would love to hear anyone’s thoughts/comments on this — especially you numbers junkies who love to look at stuff like this as well. Duquesne is gonna keep on Duquesne-ing this year… I just wanted to make sure this team gets proper recognition for this.