The Ivy League offers another opportunity for Ken Pomeroy and Dan Hanner to disagree. Once again Hanner is the one that’s a little more bullish on one specific, but what this really comes down to is how good you think Harvard is going to be. The Crimson lose a ton from last season’s NCAA tournament team. Then again, Tommy Amaker has been stockpiling talent. Will those younger players step up when given a bigger role? If they do it could change the complexion of the league race.
Let’s start with the 10,000 sims for the Ivy League using Ken Pomeroy’s numbers. He has Harvard 71st and Princeton 75th. It’s almost no difference. Which is why you get this for the percentage of titles won (including ties):
- Harvard – 57.0%
- Princeton – 51.5%
- Columbia – 9.9%
- Cornell – 2.1%
- Pennsylvania – 1.9%
- Yale – 1.0%
- Dartmouth – 0.4%
- Brown – 0.03%
More than 20% of all the Ivy League sims ended in some sort of tie in the standings. Harvard won 40% of the titles outright, while Princeton won 34%. There was also a minuscule amount of undefeated teams, which makes sense considering how evenly matched the top two teams are according to Pomeroy. Harvard went undefeated about 1.3% of the time, about the same amount that Brown went winless. Columbia is also the decisive third team in Pomeroy’s ratings, averaging 1.5 wins more per season than Cornell.
What about Hanner’s rankings? What does he see? If you read his post last week you know he’s way down on Penn, but how do his other rankings impact the standings. Once again the percentage of league titles won including ties:
- Princeton – 74.4%
- Columbia – 28.6%
- Harvard – 14.6%
- Cornell – 4.1%
- Yale – 0.3%
- Dartmouth – 0.09%
- Brown – 0.09%
- Pennsylvania – 0.02%
After taking away the challenge from Harvard, Princeton won the title outright 57% of the time. Also, Columbia gets a big boost, almost tripling its chances. On the other side, Harvard plummets and Penn’s chances go from longshot to non-existant. In this scenario Penn went winless 90 times in the 10,000 sims, while Princeton went undefeated 378 times and averaged 11.8 wins per season. According to Hanner, Columbia is a solid favorite for second, averaging about a win more per season than Harvard.
Both of these ratings system seem to be undervaluing Cornell, but no matter who you ask Princeton seems to have the inside track at winning the Ivy League title in Mitch Henderson’s second season.
Hello Jon,
Princeton has talent with experience. Harvard has talent with not as much experience. Columbia has three talented players with experience, but not as talented overall (less bench depth) as either Princeton or Harvard.
The other five teams are back in the pack compared to the three top teams named above.
Ken Pomeroy is more experienced than Dan Hanner, yes?
Without knowing how the simulations were constructed, the rankings and percentages of titles won resulting from them are not comprehensible to me. Also, how accurately did the two repective models predict the actual Ivy League season reults last year? over the last 5 years?
Eiso
Bob – Thanks. All good questions and comments. Let me try and provide some more context. Yes, Pomeroy has been doing this for longer than Hanner. This is Dan’s second season doing predictions, but he completely revamped his model going into this season to use player projections. Pomeroy’s relies on past performance, returning possessions and minutes and a few other factors.
The simulations are conducted by me using a custom program that I built which simulates each game of the season individually and then looks at the standings to determine the results. Basically I use relative strengths of the team – adjusted for home and away – that Hanner and Pomeroy predict and then have a random number generator determine the outcome based on the calculated probabilities.
Because Hanner’s ratings system is relatively new – in fact this method is brand new this season – I don’t have any historical results to go on. In terms of Pomeroy – I looked at his Ivy League projections last season and they came in pretty strong. The predictions will be more robust right before the start of conference play.