Mackenzie Irwin is a contributor to Hockey Prospectus and also contributes hockey and analytics coverage to the Washington Post.
Despite putting up one of the most dominant even-strength scoring seasons in years, Rick Nash’s Rocket Richard trophy chances are getting fainter with each passing game. Alex Ovechkin leads Nash 44 goals to 38, while Nash has three more games in the bank to make up the difference.
I’ve used a Monte Carlo analysis to project the chances Nash, Ovechkin, or any of the other top 7 goal scorers this season have of winning the Richard. The simulation estimates the number of shot attempts each player will take over the rest of the season by multiplying each player’s individual shot attempts per game at even strength and on the powerplay by their number of games remaining. For each of these shot attempts, a random number is generated in each simulation cycle. If the random number is less than their shooting percentage (we used three-year even strength and powerplay shooting percentages) the shot attempt is a goal, and if the random number is greater, a miss. By running the simulation through 10 million cycles we can see the average outcomes (goal totals) for each player, and compare the percentage of Rocket Richard trophy wins.
Below are the results for the 5 players with at least a 1% chance of coming second in the scoring race.
Ovechkin is likely to put up the fifth 50 goal season of his career: in 84.4% of the 10 million simulation cycles he scored 50 or more, and his average goal total came in at just over 52. Nash’s average goal total was just under 46, but he hit 50 goals in 8.7% of the simulations and won the Richard in 3.3 percent. A relative down year for Steven Stamkos still has him finishing second in league scoring in 30.8% of the simulation cycles and winning it in 1.7%.
Nash’s scoring this season has been driven by his 5 on 5 play. He has scored at a rate of 1.97 goals per 60 minutes – a better rate than Ovechkin has scored points at even strength. But while Ovechkin almost doubles his iSAT/60 on the powerplay, Nash is only slightly more dangerous than at 5 on 5.
Nash’s lackluster powerplay production has likely cost him the Rocket Richard trophy this year. Putting Nash to better use on the powerplay should be a priority for the Rangers’ coaching staff.
At this point in the season, Ovechkin is far and away the favourite to win the Richard. He won the trophy in 95% of the simulation cycles. Simulating the Richard race at the half-season mark in future seasons will likely be more interesting, and worth comparing to betting lines and how the scoring race turns out. A couple notes on the model and how it could be improved for next year.
Using three-year shooting percentages assumes the 2012-2015 even strength and powerplay shooting percentages to be each player’s “true talent” shooting percentages. Those percentages could be improved by regressing shooting percentages towards the mean depending on the size of the sample. For example, Tarasenko and Ovechkin have similar three-year powerplay shooting percentages, but we have a much larger sample for Ovechkin: His 166 career powerplay goals are more than Tarasenko’s 146 career shot attempts on the powerplay.
We used individual shot attempts per game to try and get at not only how the player has been performing (generating shots), but also how they are being used: the number of minutes they play, the quality of their linemates, and the roles they are used in will impact their shot attempt generation. A dominant performance like his 12 shot attempt effort against the Islanders could get Nash back in the race.
Simulating the Rocket Richard race with a Monte Carlo analysis has its limitations. Among them is injury; Tyler Seguin might well be atop the goal scoring standings had he not been injured on a low hit from Dmitri Kulikov. But shooting percentage can do crazy things for goal scoring – like when Alex Chiasson scored 9 goals in his first 11 NHL games – and a Monte Carlo analysis helps account for some of those possibilities.
Thanks to Pat Sandquist for his help with the Monte Carlo analysis in this article