By Julian Ryan
While the majority of HSAC posts have nothing to do with Harvard, at the end of the day we are Harvard students and more importantly, Harvard basketball fans. So given this year’s team is arguably the best ever, it is only reasonable that we analyze how the Crimson will fare in conference play this year.
Harvard is currently sitting pretty at 13-2, with its only losses on the road against Colorado and UConn. With big man Kenyatta Smith and point guard Brandyn Curry returning to the rotation, and its main rival, Princeton, graduating Ian Hummer, last year’s Ivy League Player of the Year, Harvard is the favorite for the Ivy crown. But how likely is that event, and could the squad put up the first 14-0 Ivy season since Cornell did so six years ago?
To answer these questions, I took the adjusted offensive and defensive ratings from kenpom.com for each of the eight squads, then adjusted those numbers by 1.4% to adjust for home court advantage as advised by Ken Pomeroy. From those numbers I could calculate a home and road Pythagorean winning percentage for each team using an exponent of 11.5 in order to work out the win probabilities for each of the 56 Ivy games this season by using the log5 formula. I then simulated the season 50,000 times, modeling each game as an independent event under a Bernoulli distribution with the probability determined by the log5 analysis.
Moving on from the math, let’s see what the simulation spit out:
The model is putting Harvard at just under 12 wins on average, over two wins more than closest rivals Princeton and Columbia. However, to some degree it doesn’t matter how many wins you get if you don’t win the conference and get the NCAA tournament bid. Instead we want to see the probability that Harvard wins the conference.
The blue bars represent each team’s chances of gaining at least a share of the conference title, and the orange bars show the probability of winning it outright. Harvard claims its fourth Ivy title in as many years 85% of the time and is outright champion in 70% of the simulations. The next best challengers in Princeton and Columbia are not even close and it seems as though at the very least Harvard should expect to be hanging up a banner in Lavietes Pavillion this season.
While getting a bid is obviously the priority for the Crimson, the team will also be playing for seeding in the tournament and will be trying to rack up conference wins. What is the likelihood that Harvard will get to 13 or even 14 wins?
On the face of it 14 wins seems like a stretch for the squad, only occurring in about 10% of simulations. 12 wins appears to be most likely but 13 is certainly not out of the question and would be a reasonable aim for the fans.
So if ever there was a season for Crimson fans to be optimistic about, this would be it. After decades of mediocrity, Harvard appears poised for its fourth consecutive Ancient Eight Championship. However, our gloating would not be complete without a quick shout out to Cornell. The former conference powerhouse, who not four years ago were in the sweet sixteen, are now one of the worst teams in all of Division I and indeed are proud of owners of the very worst defense by kenpom’s estimation. So Harvard fans should enjoy our time at the top now – because it is not guaranteed to last.
My comments about your use of Monte Carlo simulation: http://netprophetblog.blogspot.com/2014/01/about-monte-carlo-simulations.html
The games are not independent events.