What modern metrics should hockey fans pay attention to?

The Bruins had a Corsi rating of 54.7 percent last season, which means they took more of the shots in their games than their opponents did. Getty Images

Hockey lore is rich with tales of last-second goals, impossible saves, and scrappy comebacks. But hockey’s future may be measured, instead, with Corsi numbers, PDOs, and other advanced statistics.

Bringing sabermetrics to the ice has taken time, in part because hockey games aren’t made up of little, countable events, like pitches or at-bats. In hockey, the puck rarely stops moving and players can jump on and off the ice at virtually any moment.

Yet, with some creative approaches and some quirky names hockey analysts have found new ways to track what’s happening — and to expand any fan’s understanding of the game. Here are a few of the new measures.

Corsi

What does it measure?

Corsi measures shot attempts, including shots on goal, stray shots, and blocked shots.

Last year, the Bruins had a Corsi of 54.7 percent, which means that if you look at every shot taken in Bruins game, you find that 54.7 percent were taken by Bruins players. The other 45.3 percent came from opponents.

Why is it useful?

The team with the most total shots is generally the team that controls the puck, which makes Corsi a good way to gauge possession.

Teams with high Corsi numbers tend to spend most of the game on offense. Teams with lower Corsi numbers are skating backward.

What’s a “good” Corsi?

Anything over 50 percent is good. Last season, the Stanley Cup-winning LA Kings had the highest Corsi at 57.3 percent, while the playoff-missing Buffalo Sabres had the lowest, at 41.2.

How well does it track wins?

Pretty well. The chart below compares the overall Corsi of each team with the number of games they won during the 2013-2014 season. The dots move in a relatively clear pattern from the bottom left to the top right, which tells us that as Corsi goes up, wins do too.

What does Corsi stand for?

Nothing. It’s not an acronym. It’s named after Jim Corsi, a former goaltending coach for the Buffalo Sabres who’s generally considered the first to have tracked it.

Can it be used to evaluate individuals players?

Yes, you can also calculate Corsi for an individual player. You simply count all the shots taken while that player is on the ice (not just his shots, but any shots during his ice time).

These individual-level Corsi numbers have some limitations. They can be affected by the line you’re in, the quality of the team you play for, and how your coaches use you. There are ways to address these issues, but they can get pretty complicated.

Are there ways to improve Corsi?

If you tweak the Corsi numbers just a bit, you can make them evey more useful. One problem is that when teams are losing big they often take more desperate shots, which then affects their Corsi. To correct for that, you can use a measure called “Corsi close,” which only counts shots when the score is close.

There’s also “Corsi (even strength)” which excludes power play situations, since those also distort shot opportunities. The chart above makes both these adjustments, relying on what is called “Corsi (5v5 close).”

Fenwick

Why is it useful?

Fenwick is quite similar to Corsi. It measures shot attempts, only it leaves out blocked shots.

That makes Fenwick a better measure of scoring chances. If you think of each unblocked shot as a scoring opportunity, Fenwick tells you whether teams have more scoring opportunities than their opponents, or fewer.

As with Corsi, you can improve Fenwich by calculating “Fenwick (even strength)” and “Fenwick (5v5 close).”

How well does it track wins?

About as well as Corsi. In fact, with a large enough sample — like a full season — there’s very little difference between the two.

The Nashville Predators are one of the few teams that fared much better on Fenwick (14th) than on Corsi (22nd), but even that didn’t help them make the playoffs.

What does Fenwick stand for?

Fenwick is named for a hockey analyst named Matt Fenwick.

PDO

What does it measure?

PDO is a measure of luck. Have you ever thought, around mid-season, that some of the league leaders were more lucky than good? That for whatever reason they were getting a lot of fortunate deflections or friendly bounces?

One way to check that intuition is to see if their shooting percentages or their save percentages are surprisingly high. And that’s just what PDO does. It adds together a team’s shooting percentage and its save percentage to get a single number (ignoring power play situations).

An unusually high PDO is the mark of a lucky team.

Why is it useful?

PDO is based on the idea that good luck is actually bad news — because luck runs out.

If your team has a PDO above 100, that means they’ve been getting a lot of lucky wins. Moving forward, they’re likely to underperform.

By contrast, teams with PDOs below 100 are the ones being held back by bad bounces. They’re actually better than their records suggest and are poised to move up the standings.

Are there any problems with PDO?

One major problem. There’s more to PDO than just luck. Some teams really do have deadly shooters and unusually good goaltending, and those teams can sustain high PDOs.

The Bruins have kept their PDO above 100 in six of the last seven seasons. That’s likely a reflection of their skill, not just sheer luck.

What does PDO stand for?

Believe it or not, PDO isn’t an acronym either. It’s the unpronounceable username of the person who created it.

Must I give up on traditional statistics?

Not at all. Goal differential (also called “+/-”) is a traditional statistic whose enduring value lies in its ability to capture this simple truth: the surest way to win in hockey is to outscore your opponents.

There are problems with “+/-”, including the fact that it breaks down at small sample sizes, making it poorly suited to assessing individual players. But if you stick with larger samples (whole seasons and full teams), “+/-” tracks wins at least as well as any of the new-fangled statistics.