Do the best Corsi percentage teams really have the best records, and vice versa?

I’d like to look at a feature by TSN’s Travis Yost about how Corsi% can forecast the future success or failure of teams this week.

Cutting right to the chase I will go straight to his conclusion: “Corsi% is a repeatable team skill, one that correlates well with success over long periods of time. Teams that find brief success despite poor Corsi% generally cannot sustain their winning ways over long stretches. Teams burned by poor luck in the standings despite great Corsi% generally see things normalize, moving up the standings in quick fashion.”

Having measurements and statistics help tell the story of what occurred and we attempt to project what will happen in the future based on that data, but I have issues with his conclusion.

“Corsi% is a repeatable team skill.” That may be true, but that all depends on the coaches, players and injuries.

From one year to the next, no team stays intact. Changes occur and while you hope the changes are for the better, how can you derive from statistics alone if the new players will adapt and mesh well with their teammates and coaching staff?

The comment is loosely based on something called institutional memory, like that of a Board of Directors, but in this case call it puck possession memory. Assume we have a case where the team players do stay intact but the coach is changed, what happens to that memory? Do we use past year’s stats or do we throw them out the door? It depends on what kind of coach comes in to replace the previous one and what kind of relationships they can quickly develop with the players.

One of the more obvious errors in that single part of that conclusion are how injuries cannot be forecasted into the success or the more likely failure for the team that season. A bad injury happens to a key player and there goes the season. It happened to the Boston Bruins with David Krejci’s injury and almost to the New York Rangers with Henrik Lundqvist’s injury.

If someone wants to limit themselves in predicting the future only based on numerical data, that is their choice to do so. I can almost forgive that.

Let us jump into some statistics of our own though. I will be referring to information found on Hockey Abstract regarding luck and the NHL’s site for Corsi%. The statistics I will refer to will be from the 2012-13 season to 2014-15.

The Los Angeles Kings are a model for high and consistent Corsi% of 58.60, 55.75, and 54.56 over the last three years. But even though they had a very high percentage in 2014-15, they did not qualify for the playoffs. The projection that they would do well because of past Corsi% performance values failed them last season. Will it do so again this year?

Now take the Detroit Red Wings and their Corsi% of 54.17, 53.60, and 56.26. Sure they made the playoffs, but the only year they went further than the first round was in 2012-13 and they did not get past the second round. A similar story can be said of the San Jose Sharks (51.39%, 53.58%, and 50.77%).

One more case I want to point out is that of the Dallas Stars. In those three seasons their Corsi% was over 50, yet they only made the playoffs once and at in that they did not go past the first round.

Those examples were of teams that had consistent Corsi% values over 50, so why such vastly different outcomes?

I want to show two teams that had that improved over that same stretch of time, the Nashville Predators and Minnesota Wild.

The Predators had improved from a Corsi% of 44.44 to 50.22 to 51.89. They were only in the playoffs last year and they were ousted in the first round. Minnesota fared a little bit better starting with a Corsi% of 48.94, then 49.55 and ending with 52.10. They made the playoffs twice and reached the second round both times. Yost says that Corsi% is a repeatable skill, so why did they improve their previous values? Where was it in the stats they made us feel they would improve in the first place and will they continue improving this season?

The final two teams I want to focus on with values that dropped are the New York Rangers and the New Jersey Devils.

The Rangers went from 53.49, to 51.35, to 49.26 Corsi% and they stayed playoff contenders each year going to the second round twice and the third round once. Where are they projected to end up this year though? Their Corsi% is trending downwards so does that mean they will not fare so well this year? Did their numbers fall because their goalie was injured?

The Devils are another case. From 56.79, to 54.57, to 48.52, they never made the playoffs in any of those seasons but my poking and prodding will go one step further.

To quote the last part of Yost’s statement, “Teams burned by poor puck luck in the standings despite great Corsi% generally see things normalize.”

The puck luck of the Devils improved in each of those seasons, if you could call it that. It started with a -9.83, then -6.97 and then a -3.53 value. However, their team results stayed the same… no playoffs. Ask them if they feel burned as the stats did not normalize for them.

Speaking of puck luck, these are the values for the teams I mentioned earlier:

Can it be said that the Rangers and Predators had to be lucky in order to be in the playoffs?

One last place I want to take this is the emphasis that some, like Yost, are placing on Corsi%. Even the NHL site states: “For the past several seasons, full-season team shot attempts % close has been a significant indicator of which teams make the Stanley Cup Playoffs.”

Of the top ten Corsi% teams in the 2012-13 season, three did not make the playoffs, one played in the first round, three in the second round and two made it to the Finals. That sounds great if not perfect. A total of 15 teams had a value of 50% or better.

In 2013/14, three teams did not make the playoffs again, four lasted only one round, two made it to the third and one to the Finals. I do not know if you can call it a success here. A total of 16 teams had a value of 50% or better.

Finally from last year’s top ten in Corsi%, three more teams did not make the playoffs, four went no further than the first round, one went to the second round and two made the Finals. A total of 19 teams had a value of 50% or better.

While we see the finalists come from this top ten group, we also have about a 70% ratio of teams over the last two seasons that either do not make the playoffs or only play one round. That is hardly a success story in my books.

I understand the theory and logic behind puck possession. It has some sense to it, but you cannot accurately determine how successful a team will be in the future solely based on their past Corsi% values.

It at most it reduces the range of teams to select from as your Stanley Cup Finalists, so your odds are better but you and those teams still have to be lucky.   

Have a great Labor Day Weekend, punk!