CFL 2014 Predictions

As of November 23rd, the current predictions for the CFL playoffs are as follows:



Final prediction for the Grey Cup: Calgary Stampeders to win with a 65.3% chance.

Based on an analysis of past Grey Cup games, teams don't get any discernible home field advantage for playing in their division on the Grey Cup (if they're not already hosting), so no such factor has been applied to the Calgary-Hamilton odds.

Good luck to both teams! It should be a good one.

Previous write-up (Nov 17): 
Both semi-finals went as expected, with Edmonton and Montreal cleaning up their respective games. Next week's match-ups are:

• Edmonton at Calgary: Unfortunately, this one isn't expected to go Edmonton's way. Calgary has been by far the stronger team throughout the season, and with the home town advantage there's not much reason to expect this going any other way.
• Montreal at Hamilton: Because of how tight the east was this year, Montreal is actually expected to do a little bit better average for a visiting team. That being said, home town advantage is so high in the CFL that Montreal is still looking unlikely to accomplish much.

This leaves Calgary with an overall 45% chance of winning the cup, well ahead of any other team. We'll see what happens!

Previous write-up (Nov 3):

With one week left to go, here are some fun facts:

• Toronto's can still beat Hamilton into the playoffs if they win their next game against Ottawa, and Hamilton loses their next game against Montreal. The odds for this currently sit at 26%, so I wouldn't get your hopes up.
• If Hamilton wins against Montreal, they will end up first in the East and get a bye to the Eastern finals. Just doing this would increase their chances of making it to the Grey Cup to 40%, a huge swing against Montreal. This is obviously a pretty important game for both teams.
• Edmonton getting the home team advantage in the Western semi-finals will be a big boost, and is what contributes to their 70% chance of getting to the Division finals.

How the season's held up model-wise so far:

• Based on previous season, an away-team factor of 0.8 was applied (assuming the away teams only win 40% of games). Away teams actually won 41.6% of the time, so this factor doesn't seem inaccurate.
• The Pythagorean expectation for each team (with CFL factors instead of NFL) seems to have been reasonably accurate too! The expected vs actual wins so far have been:
• Calgary: Expected: 12.5, Actual: 14 (+1.5)
• Edmonton: Expected:12.8, Actual: 12 (-0.8)
• BC: Expected: 9.6, Actual: 9 (-0.6)
• Saskatchewan: Expected: 7.1, Actual: 9 (+1.9)
• Winnipeg: Expected: 6.7, Actual: 7 (+0.3)
• Montreal: Expected: 7.8, Actual: 9 (+1.2)
• Hamilton: Expected: 8.7, Actual: 8 (-0.7)
• Toronto: Expected 7.7, Actual: 7 (-0.7)
• Ottawa: Expected 3.6, Actual: 2 (-1.6)
• This has overall led to a 70.1% success rate in game prediction. The p-value for correctly predicting 54/77 sports games blindly is 0.0003, or 0.0430 if you factor in the home-town advantage. Either way, it looks as though the model is statistically valid.