Thoughts on Corsi and advanced stats in hockey

Corsi

Nathan MacKinnon, Colorado Avalanche. Image courtesy of Wikimedia Commons.

What is Corsi? A formula that calculates shot attempt differential. Its purpose is to measure puck possession. It was invented by long-time goalie coach of the Buffalo Sabres, Jim Corsi, and has become a mainstream advanced hockey metric over the past five years or so.

How Corsi is calculated: The total shot attempts of one team (or player) is compared to the opposition’s same totals. There will be a differential. One team will have attempted more shots at goal than the other. This team will have produced the higher Corsi ranking, or Corsi For%. And people will say that said team is a better puck possession team.

Does Corsi perfectly measure puck possession? No, because it doesn’t measure time with the puck. Corsi isn’t a perfect science.

Corsi results are considered by people who possess logic. Logically, the team that attempts more shots on goal than its opponent has possession of the puck more often and is generating more scoring chances.

Corsi is the most popular known method used to measure puck possession, and that’s because we’re not privy to the advanced hockey metrics used by the Chicago Blackhawks.

Corsi is an insightful tool. However, it’s not the numbers themselves that provide the insight, it is the conclusions drawn from those numbers.

Blackhawks general manager Stan Bowman told the Sun Times, “Stats are what they are. There’s no disputing who scored the goal, or who was on the ice for the goal. That’s fact,” said Bowman. “What you do with that is sort of the real value. And I think there’s an art to it. The analytics themselves are very objective. But then you have to do something with them and draw conclusions.”

Advanced stats are informative. If the right conclusions are drawn from the numbers, they can be insightful. Stats, after all, can also be misleading — manipulated to prove a point one way, or another. As Mark Twain once popularized, “there are three kinds of lies: lies, damned lies, and statistics.”

But Corsi numbers don’t lie — the conclusions drawn from the numbers may tell a false story. The way in which the numbers are weighed in the grand scheme of things may be skewed. That’s on the interpreter, not the numbers themselves.

Corsi numbers provide a more accurate story over the long-term than they do in small sample sizes. It’s a big picture tool. 

The team with the best Corsi number doesn’t always win the game.

For example: The Bruins are playing the Canadiens. The game is entertaining. Lucic speared someone in the groin. P. K. Subban scored a blast from the point. Rask threw his stick at his own bench. Plekanec wore a turtle neck. Chara was tall. Desharnais was not. And Brad Marchand made you throw something at the television. Unless you’re a Bruins fan, in which case: Brendan Gallagher made you throw something at the television.

Montreal won 4-3. But the Bruins had more shot attempts (shots on goal, shots missed and shots blocked by the opposition). Canadiens won the game; Bruins had a better Corsi For%.

Yes, sometimes the team with the better Corsi For% loses the game. Corsi doesn’t determine winners, it provides insight. Teams with strong Corsi numbers tend to win more often than teams without. It’s about probability over a long period of time — generating offensive opportunities versus allowing them.

Not every team with strong Corsi numbers earns winning results and that’s because Corsi doesn’t paint the entire picture. Hockey is too complex to be explained using one statistical measure. But Corsi isn’t irrelevant, either.

Think about it logically. One team attempts 80 shots on goal and the other attempts 50. This is the NHL. Both teams’ goalies are good. Both teams have talented shooters. The only huge difference between the two clubs is that one attempted way more shots at goal than the other. Thus, it’s reasonable to conclude that the team with 80 shot attempts had the puck and generated more scoring chances than the team with 50 shot attempts.

Did they for sure? No.

But did they actually? Yes. And here’s why: Because in order to win a hockey game, a team must score more goals than its opponent. In order to score goals, a team must take shots on net. It is wise, as a team, to shoot the puck at the net, whilst possessing it. Because in doing so, goals may occur. And, as we’ve already agreed upon, scoring more goals than the other team is the only way to win a hockey game.

Having a good Corsi For% is where it’s at. Teams with good Corsi numbers generate more offensive scoring chances than they allow against; they shoot the puck at the net more often than their opponents shoot back. Which means, yes, they almost certainly have possession of the puck more often to accomplish this.

The team with the best Corsi For % may not score the most goals in the NHL because there is a human element at play in hockey. Skill level varies from team to team; player to player.

The New Jersey Devils finished the 2013-14 NHL season with the fourth best team Corsi For% in the league. They missed the postseason.

The Colorado Avalanche had the twenty-fifth best Corsi For % by the end of the 2013-14 season. They won their division.

The Devils lacked in other areas conducive to winning, despite having a strong puck possession team. It costed them a playoff spot.

The Avalanche excelled in some areas (goaltending and shooting percentage), but they weren’t a great puck possession team.

Colorado enjoyed regular season success and the Devils didn’t. But these teams were exceptions to the rule — proof that Corsi isn’t a perfect science. Proof that goaltending, special teams and other criteria are also relevant pieces of the puzzle.

Yet, the top three Corsi For%’s in the NHL this past season were the three most recent Stanley Cup champions: the Kings, Blackhawks and Bruins. The Oilers, Maple Leafs and Sabres were the worst, and all three missed the playoffs.

Corsi has its detractors.

There is a divide amongst hockey people. Some have wholeheartedly embraced advanced stats, meanwhile others have not. I believe this occurs because some folks overly praise advanced stats; weigh them too heavily. And there are those who believe advanced stats like Corsi are completely irrelevant. The two sides butt heads. Somewhere in the middle is probably the truth of it all.

I believe that a smart hockey mind can form excellent, on-point player personnel opinions without the use of advanced stats. But to ignore an available tool that winning franchises have embraced, seems closed-minded, also.

Is it impossible to evaluate talent without using advanced stats? Of course not. The trained eye sees much more than the untrained — the blogger, journalist, fan who yes, watches hockey religiously, but has never lived life in the NHL. I personally believe it’s easy to play armchair GM, and laugh at mistakes with the benefit of hindsight, while spewing out “I told you sos,” at the bar with friends or on Twitter. And we all do it because it is our right to do and it creates entertaining conversation.

But even the keenest eye for hockey — although very sharp and capable of recognizing talent — can’t be everywhere at once, watching every single game, focusing undivided attention on every single team. Advanced stats help fill in the blanks. They serve a purpose.

Folks who are NHL general managers and coaches: they assess talent well with their own eyes. These hockey geniuses already know which players drive possession because they absorb the minor details of the game so intently when they watch it. They have a keen eye for such things. Although, some GMs and coaches do make us wonder what they’re thinking, don’t they?

Is Corsi the ultimate answer? No.

Some people love Corsi a lot. They weigh its value heavily and think people who don’t worship Corsi are dinosaurs who know nothing about hockey.

For example, a blogger who worships Corsi, but has never worked in the NHL, probably doesn’t know more about NHL hockey than, say, Brian Burke. Brian Burke is president of hockey ops with the Calgary Flames, and he doesn’t believe in advanced hockey stats. 

Do I know more about NHL players and hockey than Brian Burke? I doubt it. Sure, I watch hockey a lot, always have. But to think I’m a better NHL talent assessor than a person who has been general manager of several organizations, won the Stanley Cup, and has lived the NHL life for several decades, would be egotistical and even a little bit delusional of me. I watch movies, too. Does that mean I’d make a good movie producer? No.

NHL people like Brian Burke have their reasons for dismissing advanced stats. To me, it seems like a purist mentality, but hey, Burke’s got credentials so he can believe what he wants and practice those beliefs. Agree or not, he’s an accomplished hockey guy. But he is also part of a shrinking population of NHL front office people still ignoring advanced stats, so that’s saying something.

What is Corsi? One of the best advanced stats formulas available to fans and media. Corsi is a tool that provides insight. It is a logical indicator of puck possession. Embrace Corsi, and other advanced hockey stats for what they are: insightful, informative tools.

NHL teams use their own advanced stats methods, and those are probably even more insightful than the stuff we fans are aware of. But that is proprietary information, as are all secrets that give one company a competitive advantage over its competition.

There’s Fenwick, which is similar to Corsi. And there’s PDO, too.

Extraskater.com seems to be the leader in advanced stats. Check them out. Google “Advanced Stats Hockey” and you’ll find many more resources. If you’re interested, I recommend learning about advanced stats. They’ve been part of the game for a while now, and they will only become more accepted as they continue to earn endorsement from Stanley Cup winning GMs like Stan Bowman.

Heck, Hockey Night In Canada even tosses advanced stats in our faces during broadcasts. It doesn’t get more mainstream than that.

Corsi is an insightful tool used to measure puck possession. Good teams are great at possessing the puck — Corsi quantifies that in more detail than traditional stats do.

However, there are other qualities a good team must possess to be successful. For example, along with puck possession: good goaltenders, special teams and clutch performers determine a team’s ability to win games. Strength in all of these four areas is required of a team if it wishes to be a winner. Some teams are good in two areas, some in three. The recent Stanley Cup championship teams are strong at all four.

And then there’s the bounces. That puck sure does bounce out there some nights. Players hit the post, whiff on shots, miss open nets, have their sticks explode while shooting from prime scoring territory. That is why hockey is so suspenseful, satisfying and entertaining. Sometimes the team with the Vezina calibre goalie, strong Corsi numbers, excellent penalty kill and proven core of winners doesn’t win the game. Sometimes they don’t win the series.

If you think Corsi numbers are going to give you all the answers, you’ll be disappointed in the end. But to dismiss them entirely, you may be ignoring an insightful tool that is readily available and easy to understand. Corsi is not the ultimate answer, but it is part of the big picture.