The emergence of advance metrics in sports has brought about all sorts of new statistical measures. For example football has air yards per pass completion and Rushing yards before contact per rushing attempt. Hockey decided to develop a stat called Corsi.
The term Corsi is often used as an umbrella term, but in reality it is the total number of Corsi For (CF) events minus the total number of Corsi Against (CA) events. A Corsi event is a shot, block, or miss.
The CF of a team or player is all shots + blocks + misses that team or player makes. Conversely, the CA of a team or player is all shots + blocks + misses the opposing team makes. The difference between those two (CF – CA) is Corsi.
At its core, Corsi is a measure of possession. A team with positive Corsi has more shots, blocks, or misses than their opponents. But is this a good measure of how successful a team will be? This post attempts to answer that.
(A great, very in-depth, article on Corsi can be found here.)
First, I wanted to look at the most successful teams in the NHL last season. Below you will see teams ordered by their Corsi. The most positive ones are at the top, while the most negative are on the bottom. The colors represent the team’s playoff spot last season.
Only 12 of 31 teams (38.7%) last season actually had a positive Corsi. Of those 12 teams, 10 made the playoffs – three 1 seeds, two 2 seeds, three 3 seeds, and two wildcard spots. The Carolina Hurricanes had the best Corsi last season at 748, yet they were a wildcard team. The Montreal Canadiens were the fourth best Corsi team at 664, but didn’t make the playoffs.
Of the 19 teams with negative Corsi one was a 1 seed, two were 2 seeds, one was a 3 seed, and two were wild cards. The playoff team with the worst Corsi was the New York Islanders with -319 for the season. It’s fairly clear from this that Corsi does not necessarily equate to success.
Looks at this another way: below are team’s Corsi and how it relates to a team’s number of wins. Again, the color represent the team’s playoff spot last season. The light blue shaded area is for positive Corsi and the light orange is for negative. The linear trend line is added as well.
There was a positive relationship between Corsi and wins last season. On average, a team would need around 86 more CF events than CA events before they would increase their total number of wins by one.
The Tampa Bay Lightning are clearly a large outlier last season. They had 62 wins and a Corsi of 236, making them the 9th best in terms of Corsi.
The most interesting item above is where teams sit relative to the trend line. Every team above the trend line was a playoff team (12 of the 16 playoff teams).
Three other playoff teams: the Carolina Hurricanes, San Jose Sharks, and Vegas Golden Knights, all are below the trend line but they are ranked first through third in the Corsi standings, respectively.
Finally, a historical look at how Corsi relates to a teams success should be a better measure. Below is every team’s Corsi and how it relates to their number of wins since the 2007-2008 season. I excluded the 2013 season which was shortened due to the lockout. The light blue shaded area is for positive Corsi and the light orange is for negative. The linear trend line is added as well.
The increased data points provide even more insight. This plot suggests that historically it takes an additional 111 Corsi for the season to increase a team’s win total by one.
At the extremes of the Corsi measure are some very dominant and very poor teams. The 2014-15 and 2013-14 Buffalo Sabers certainly take the cake for some of the worst teams recently. The 2009-10 Chicago Blackhawks and 2013-2014 LA Kings were both Stanley Cup Champion teams.
So is Corsi a good indicator of success? How about this hypothetical example:
Say a team has a CF of 100 in a game (they shot, blocked, and missed the puck 100 times) and the have a CA of 0 – meaning their Corsi is 100. At it’s surface this would seem phenomenal; most would be willing to bet this team won. However, if all 100 points of their CA were misses that would mean they didn’t block the opponent puck once or shoot the puck. In theory, the opponent could’ve scored 20 goals on them and still have a CA of 0.
While rather extreme, this shows that Corsi does not necessarily relate to the success of a team.
What all this implies is Corsi’s insight into a team’s success is relative to other teams. It is not enough to know if a team’s Corsi is positive or negative.
Pingback: Predicting NHL Hockey Games – The Commute Sports