Predicting the QMJHL playoff champion using SRS

The Saint John Sea Dogs and their league leading post-deadline SRS of 2.56 have a strong chance of being crowned President’s Cup champions this year. But if an upset were to happen, look for the Charlottetown Islanders to win for the first time in club history with their post-deadline SRS of 2.39. Here’s why:

Looking at regular season data from 2007 to 2016, the team that has won the Jean Rougeau trophy for having the most points in the regular season has won the President’s Cup 6 times. In that same time period, 3 “underdog” teams have won. However, those 3 teams were all very talented and could have easily been considered the favourites to win it all. This example shows the limitation of looking solely at points to evaluate the quality of teams following the regular season.

I am a strong believer in using the Simple Rating System (SRS) approach to evaluating teams. SRS was initially created by Doug Drinen of Pro Football Reference in 2006 to measure all college football teams along one metric despite their unbalanced schedules. While the calculations can get complicated, it essentially combines a team’s strength of schedule and their scoring differential. This is great for two reasons:

  1. Wins and losses are relatively limited compared to goals scored and conceded. The much larger sample size of goals data increases its predictive accuracy. Teams that win a lot of 1 goal games are far more prone to “bad puck luck” in situations like the playoffs compared to teams that win the same amount of games but consistently outscore their opponents by multiple goals.
  2. In the Q, the schedules are highly unbalanced. For example, Halifax played 26 of its games this season against 3 of the top teams in the league (Saint John, Charlottetown, and Bathurst), while Québec played those same teams only 6 times. Halifax had a much tougher schedule this year than Québec, but this won’t show in the regular season standings.

One additional modification that I made to the SRS approach is that I only looked at post trade deadline games. The Q is a unique league in the sense that teams can be significantly stronger or weaker following the trade deadline. Since the trade deadline is just past the midpoint of the season, there is still a 30 game sample size per team. Based on my analysis, the post trade deadline SRS is a stronger predictor of playoff success than full season SRS.

President's Cup Winners

The table above looks at the President’s Cup winners since 2007, and compares their regular season point rank and their SRS rank. For the 3 “underdog” teams that won, SRS had successfully evaluated 2 of them as being the best team in the league (2014 Val-d’Or Foreurs and the 2008 Gatineau Olympiques). The only winner that SRS missed was the 2010 Moncton Wildcats; however, the SRS margin between the Wildcats and the top SRS team that year, the Saint John Sea Dogs, was only 5%. This means that SRS can not only predict who is most likely to win, it can also predict by how comfortable of a margin we can expect them to win.

2017 SRS Standings Smaller

Analyzing the 2017 regular season (see above), the Saint John Sea Dogs had the highest post trade deadline SRS at 2.56. That being said, the Charlottetown Islanders are close behind with an SRS of 2.39, which is only 3% behind Saint John. If history is any indication, these 2 teams will be the ones battling it out to win the President’s Cup.


Tagged with: , , , , , , , , , ,
Posted in SRS Analysis
Follow on Twitter