Have you ever watched your favorite team dominate play, creating shot after shot only to have the other team’s goalie play like Martin Brodeur reincarnated? Then the other team limps down the ice, gets one or two shots on net, and of course one scores a goal. Some may refer to this as bad puck luck. The tricky thing about puck luck is that it is impossible to sustain, meaning that a change in luck is always just around the corner. This is an example of what stats nerds refer to as regression to the mean.
To put it simply, regression in the context of hockey is the idea that when a player has a season in which he produces at a high level, there may be some percentages working in his favor that are not at all related to his skill. There are of course many factors that can influence a player’s production; natural skill development, better teammates, more favorable deployment, and even a change in mindset or confidence plus many more. However, there are also factors that the player has little control over, and the idea is that these percentage-based events will mostly even themselves out over time. So, if a player experiences an unusual amount of good puck luck within a season they are unlikely to repeat their performance in the following year.
All data comes from evolving-hockey.com
A Case Study
One of the easiest percentages to examine is shooting percentage (sh%). A player’s sh% is the percent of shots that end up being goals out of all shots on net he takes. While there is definitely a skill component to sh%, (i.e.- players can sustain a higher/lower shooting percentage over their careers) a year-to-year fluctuation is a glaring sign of puck luck.
Let’s use Kyle Turris as an example. Turris began his career with the Coyotes, was traded to Ottawa in 2011, and remained there until the 2017-2018 season. His career sh% until then was about 10%. In the 2016-2017 season, he had a sh% of 14.6% (a 46% increase over his career average) and piled up a career high 27 goals. Skeptical observers may have exclaimed regression!, but the Nashville Predators were not skeptical observers. That season also happened to be the last year of Turris’s contract, and Nashville traded for him and promptly gave him a hefty raise to go along with a six year contract. Here’s what his sh% has looked like since:
He has not been ideal in Nashville, as his sh% and goal totals declined every year.
Players Due for Regression in 2019-20
Now that we are aware of the dangers of an unsustainable sh%, who might be due for some bad puck luck this upcoming season? Below are the four players from last season who had the biggest difference between their 2018-19 shooting percentage and their career average.
These four players all had highly unusual shooting percentages. The one that stands out the most to me is Casey Cizikas. After hovering around 9.5% for four straight years, he scored on 18% of his shots this year.
Another interesting name here: Matt Duchene. Duchene happened to be in the final year of his contract this past year and put up the highest shooting percentage of his career. Guess which team signed him to a huge contract? That’s right, the Nashville Predators. Fool me once…
A Pleasant Plot Twist
Remember everything from above where we predicted everyone to have worse seasons? Well it turns out that regression works in both directions, meaning players who had unlucky seasons can regress to the mean in a positive way. These are players who are due for more bounces to go their way, and that usually means a higher shooting percentage. Here are four players from this past season who are likely to have more luck in the upcoming season.
David Backes (BOS): A 35 year old veteran player with a rich cap hit, Backes has seen his sh% decline for the fifth straight year. After being a healthy scratch during Boston’s playoff run, many people wonder how much he has left.
Tobias Rieder (EDM): Rieder scored 0 goals but somehow that wasn’t the most outlandish shooting percentage (in terms of z-score). Will he repeat that performance this year? (Hint: no)
Jaden Schwartz (STL): Ah yes, the Blues are definitely the team in need of more bounces to go their way.
James Neal (The Real Deal): The player who is most due for puck luck. Neal’s 18-19 shooting percentage was 2.46 standard deviations below his career average. He will start his redemption tour on a new team after being acquired by Edmonton after spending one season in Calgary. Will Edmonton actually be on the winning end of a 1 for 1 trade?
Thanks for making it this far. I hope you have a clearer understanding of how regression plays a factor in hockey. Remember, whether a player is having a ‘career year’ or is one game away from being run out of town, check the percentages.
Top 10 Overperformers in 2018-19
- Travis Zajac
- Matt Duchene
- Jakob Silfverberg
- Casey Cizikas
- Valtteri Filppula
- James van Riemsdyk
- Joe Pavelski
- Ryan Strome
- Tom Wilson
- Patrice Bergeron
Top 10 Underperformers in 2018-19
- James Neal
- Ryan Kesler
- Jaden Schwartz
- Tobias Rieder
- David Backes
- Riley Nash
- Tom Pyatt
- Tyler Toffoli
- Jeff Carter
- Ondrej Palat