Look out! Joe Pritchard’s breaking out the CFL stats again!
Specifically speaking, The Grueling Truth’s resident CFL history enthusiast goes back to the now-classic advanced metric known as the “Pythagorean Expectation.” Now that the 2016 season is over, we may ask: Which teams outperformed a purely mathematical assessment of points scored and points allowed? Have many underperformed?
One could put it in other terms: Which teams were the luckiest and unluckiest in the 2016 CFL season? Joe has answers.
Before the beginning of the season, I experimented with Pythagorean Expectation as it relates to the Canadian Football League. This is a football website, so I won’t spend a lot of time discussing what’s under the hood here; long story short, it is a mathematical formula derived from a team’s points scored and points allowed, and how many wins a team can expect to get with such a spread.
For example, this year’s Edmonton Eskimos scored 549 points and gave up 496 points. Based on that point spread (+53), a team would expect to win 10.244 games in any given season. This year’s Edmonton team won 10 games, meeting expectations almost exactly.
On the flip side, this year’s Hamilton Tiger-Cats scored 507 points, and gave up 502. A team with that point spread (+5) would normally be expected to win 9.122 games, but Hamilton only won 7. This was the biggest “miss” by the Pythagorean Expectation system this season.
Aside from Hamilton, every team this season finished (or very very close to being) within a win of their “expected” wins this season besides Hamilton and, to a lesser extent, Calgary. Here are the standings, with real wins followed by “expected” wins. Note that ties are expressed as half a win, for simplicity.
Calgary 15.5 / 14.04
BC 12 / 11.21
Winnipeg 11 / 10.11
Edmonton 10 / 10.24
Saskatchewan 5 / 4.37
Ottawa 8.5 / 8.70
Hamilton 7 / 9.12
Montreal 7 / 8.02
Toronto 5 / 4.56
I came to a couple of conclusions by running data from the ten seasons before 2016 (2006-2015). Teams that were either very statistically “lucky” (exceeding their expectation more than half a win) or “unlucky” (falling short of their expectation by more than half a win) had very different seasons the next year, while teams that came close (like most this year) didn’t have much of a change the next season.
Teams that underachieved their Pythagorean Expectation by at least a half a win the season before, on average, added over 2 wins (actual number 2.13) the next season. The biggest positive swing of a team underachieving their Expectation was the 2014 REDBLACKS, increasing by 10 wins the next season, and the biggest negative swing of a team underachieving their Expectation was the 2007 Argos, who dropped 7 wins the next season.
Teams that overachieved their Pythagorean Expectation by at least a half a win the season before, on average, lost around 1.5 wins (again, actual number -1.47) the next season. On a whole, these teams had less seasons of huge swings than teams that underachieved the year before. The biggest positive swing of a team overachieving their Expectation was the 2010 Eskimos, who went up 4 wins the next season, and the biggest negative swing of a team overachieving their expectation was the 2014 Riders, who fell by 7 wins in 2015 after overachieving by 2.2 games in 2014.
Teams that came very close to matching their Pythagorean Expectation (by being within a half a win, over or under) averaged out to gaining .026 wins the next season, or basically, staying put. Matter of fact, all teams that fit this category besides the 2007 Stamps (who gained 5.5 wins the next season, ties counted as .5 wins in this study) and the 2016 Bombers (who gained 6 wins) stayed within 3 games of their record the season before. The teams that fell the most after hitting their Expectation the year before were the 2008 Lions and the 2009 Alouettes, both of who fell by 3 wins the next season.
So what can we take from this? For starters, we can begin to make an educated guess about what the standings in 2017 will look like! Here is what the system expects to happen in 2017. The 2016 regular-season record for each team is listed first, with 2017 “prediction” following.
Calgary 15-2-1 / 14-4
BC 12-6 / 10-7-1
Winnipeg 11-7 / 9-8-1
Edmonton 10-8 / 10-8
Saskatchewan 5-13 / 3-14-1
Ottawa 8-9-1 / 8-9-1
Hamilton 7-11 / 9-9
Montreal 7-11 / 9-9
Toronto 5-13 / 5-13
This gives us 79 total wins, of course, there are 81 games in a CFL season, so the expectation’s base predictions are a little short. I added 2 wins to teams that fell short over .5 wins, and subtracted 1.5 wins to teams that overachieved by over .5 wins, and left teams in between -.5 and .5 wins alone. Of course, teams don’t add or lose wins at these exact rates (they’re averages, after all), but this should give any prognosticator a decent base to work from!
Now, you may be wondering how the expectation would have predicted the 2016 season. First, let’s start with the 2015 actual wins and projected wins, like we did early on in this piece:
Edmonton 14 / 12.6
Calgary 14 / 12.7
BC 7 / 7.7
Winnipeg 5 / 5
Saskatchewan 3 / 5.8
Ottawa 12 / 9.3
Hamilton 10 / 12.5
Toronto 10 / 7.4
Montreal 6 / 8.6
And here is what just the averages would have predicted for 2016:
The two clear misses are Winnipeg and Hamilton, Winnipeg improved by leaps and bounds, gaining 6 wins, as noted earlier, and Hamilton cratered, losing 3 wins instead of gaining, like the averages would have expected. The system also grossly underestimated Calgary’s dominance this year, and came a bit short on BC as well. It did hit Saskatchewan dead on as well as coming close on Edmonton and Montreal.
2015 was a bit of an outlier, however. I do want to point out that of the 6 teams to over or underachieve their expectation by over 2.5 games, 5 came in 2015!
Here’s what the system gets right, though: It anticipated which way 6 of the 9 CFL teams would trend, missing only on Calgary, Hamilton, and Winnipeg.
Hopefully this system gives CFL fans a better idea of which teams to expect to improve or fall off their 2016 records.