[octothorp]

Something I've always wanted to do is my own, mathematically derived, non-subjective power rankings. So after spending some time fiddling with a spreadsheet, I've got my first iteration.

It uses a Sagarin-style formula that takes the margins of all scores in the season thus far, then comes up with a ranking formula that produces the least error, including taking into account a home edge, which is also calculated in the forumla.

Note that there is no time bias. A quality win in week 1 counts for the same as a quality win in week 19. This is similar to the way college football rankings work, in that it's designed to rank teams over the entire season. I'd like to do one that biases recent games.

Also, note that this is purely points. There isn't much difference between winning by one point and losing by one point, but there's a big difference between losing by one and losing by 30.

Overtime games are counted as their regulation scores + 1 for the winning side.

So through twelve weeks, here are the rankings:

1) Calgary (37.3)
2) Montreal (31.3)
3) Winnipeg (27.0)
4) Hamilton (26.9)
5) Saskatchewan (25.2)
6) BC (24.1)
7) Toronto (23.6)
8) Edmonton (15.0)

Current home edge is predicted at 7.4. So if you were to use this formula to predict games, you'd add 7.4 to the home team's score (number in bracket), subtract the away team's score, and this would give you a margin of error. I'm not responsible for any property you might lose should you make any bets based on my rankings.

I'll try to update on a weekly basis until the end of the season, and I'll be testing some time-bias formulas to see what works best for CFL scores.

Edit: something was just not right with Hamilton's rankings, so I checked my tables and sure enough I input one of their scores wrong.