University Cup: Tournament odds

Just in time for today's opener between McGill and Manitoba:

Chances of finishing first in respective pools:
Pool A
Alberta 53.5%
Lakehead 28.9%
UQTR 17.6%

Pool B
SMU 41.4%
McGill 41.3%
Manitoba 17.4%

And what most people care about, the chances of winning the University Cup:
Alberta 32.5%
McGill 20.5%
SMU 17.5%
Lakehead 14.6%
UQTR 8.6%
Manitoba 6.6%
Next PostNewer Post Previous PostOlder Post Home


  1. How were these calculated?

  2. Sorry for the lack of explanation; had to get them up quickly this afternoon.

    I basically figured out how likely each outcome was, based on the teams' RPI and SRS rankings. An outcome is exactly what it sounds like: an example would be "Alberta beats Lakehead 4-2, UQTR loses to Lakehead 2-1, and UQTR loses to Alberta 3-0." That particular outcome's result (Alberta advances) would be associated with some probability, and adding up all the probabilities for all the results would tell us how often Alberta advances (which is about 53% of the time).

    From there, I was able to see how often each team won their pool, and at that point it's just a "who's the favourite in Sunday's final and by how much?" calculation, which is pretty easy.

    This calculation is quite a bit more complicated than the Final 8 odds for basketball, since it's not just "win and advance." (The possibility of ties in pool standings makes it tricky, but, hey, that's why there's a math guy on the blog.)

  3. That of course assumes a degree of parity between conferences, which may be a critically incorrect assumption.

    For example, the same calculation could be done with major junior, junior "A", and junior "B" teams from which it could be concluded that a given junior "B" team would win the Memorial Cup.

  4. Actually, you're a bit off since RPI includes strengh of schedule, i.e., the calibre of your opponents. Acadia, for instance, was third or fourth in RPI much of the season despite not having, on paper, as good a record as the OUA's best teams. That attested to AUS' strength.

    That is akin to how a 20-15-5 major junior team is better than a 30-8-2 Junior A team. I'm not smart enough to know if you just proved Rob's point.

  5. Neate's explanation is a little off: it's not just because we have strength of schedule, but also because we have enough interconference play (see the full schedule as a .pdf here) to get a sense of how good these teams are.

    The same calculation would not be done with Major Junior and Junior A/B teams because a) we do not have interconference play between the leagues (at least, none I'm aware of) and b) even if we did, the leagues have clearly different levels of play due to the different eligibility rules they have and due to numerous instances of a star Junior A player who can't hack it in the CHL. In other words, right from the beginning we wouldn't think of treating a .500 team from one league as equivalent to a .500 team from another. You could do that, but it would be irrelevant as an exercise because (as you correctly point out) the assumptions aren't valid.

  6. Rob - have you created your own RPI as I haven't seen anthing published or posted about an official RPI/SRS for CIS Hockey.

  7. Yes, it's an unofficial ranking, and it's produced by the development team here at CIS Blog Labs (i.e., me). Rankings are available here (see the links on the right under "Rankings" for other sports). As far as I know CIS does not produce any ranking other than the Top 10 voting.