This will be a fun post, though I need to type fast because the 2015 Stanley Cup playoffs officially kicked off last night! In this post I will present my playoff simulations, where I use team-Elo values heading into the playoffs to make the predictions.
While I acknowledge that there are multitude of factors, other than Elo, to consider for making Cup predictions, and that in the end chance plays a huge role, an Elo based prediction should make for some interesting reading and it never hurts to make predictions.
As a predictive measure of playoff performance, we will use the average team-Elo during the months of March and April before the playoffs. In the previous 4 seasons the correlation between winning playoff games and March/April Elo looks like this:
The first thing to note here is that predicting who will win the Cup is a bit of a crap-shoot. There is some correlation, however, as we can see that teams who don't make it out of the first round average an Elo of 1.52, while teams who win all 16 games (i.e. take the cup) have an average Elo of around 1.53. Remember "anything can happen" - just ask Boston in 2014! A team that was projected as the best in the playoffs by pretty much every predictive statistic. Ask Phillidelphia in 2010 'how did you get to the cup finals?' then ask them in 2011 'how did you not get to the cup finals?'
Anyhow - yes - predicting Cup winners is hard, but let's have a crack anyways. Armed with the knowledge of home and away winning percentages versus Elo difference from a previous post, I've run one hundred thousand Monte-Carlo simulations of the playoffs. In each game, the difference in March/April Elo ratings determines a probability of the home team winning or losing. The following table displays the results;
The team with the best form in the NHL is the New York Rangers, with an average on-ice Elo of 1.564 during March and April 2015. The monte-Carlo simulation rates them as 62.6% to get past Pittsburgh in the first round, with an eventual probability to win it all of 15.5% - the highest probability in the league. Their most likely opponent from the West is, perhaps surprisingly, Minnesota! While Minnesota have only a 51% to exit their first round battle with the Blues, their great form (March/April Elo of 1.556) puts them in position to beat some slightly easier opponents after the first round.
The worst team in the playoffs (if you trust my Elo ratings) are the Calgary Flames with an average Elo of 1.496 in March/April. The flames only have a 0.8% chance of winning it all, according to my analysis. Some assumptions were made to deal with player injuries leading up to the playoffs. A smaller sample size was used for Pittsburgh, since they lost so many stars on their defense right before the playoffs. Similarly, Patrick Kane and Kimmo Timonen's stats were added to that for Chicago, and Max Pacciorety's stats were removed from Montreal's Elo calculation.
So there you have it folks. I've got New York beating Minnesota in the cup final this year!