Fancy stats: An introduction to hockey analytics and the importance of shots

Matt Pfeffer
welcomepostpic-181765732

You’ve seen Moneyball. You’re thinking “do they have that for hockey?” They sure do!

It’s a different sport and it’s much harder to track the kind of stats that can be tracked in baseball, but advanced – or fancy – stats is a growing phenomenon in hockey, as fans and teams alike look for more ways to evaluate players beyond the traditional numbers we’re used to seeing in the box score.

Advanced stats – be it from the Canadiens tracking plus-minus in the ‘50s or Roger Neilson tracking scoring chances – used to strictly be used for behind closed doors information. Now, with the wealth of data the NHL provides every game, along with some heroic efforts inside the hockey blogger community to track new stats not covered by the NHL, there is a ton of information available for fans. It’s an exciting time to be a nerdy hockey fan. So get your taped up glasses out and let’s talk about the new generation of hockey stats.

In this column I’ll be your host as we traverse the world of analytical thinking in hockey and why you should care about stats named after NHL goalie coaches (corsi) and anonymous Internet commentators (PDO). Yeah, really.

The first concept that’s crucial to understanding analytical thinking in hockey is shots. Shots are like goals, except they include the ones that don’t go in too. We care about shots because, well, this:

graph

This is every team’s shooting percentage by season since 2008-2009. It’s craziness! The point of this grotesque and horrifying graph is to make the very important point that shooting percentage is not a sustainable skill especially amongst teams, but also at a player level, too. (This wouldn’t be an article on stats without ifs, ands and buts, but more on those later.)
If we accept the notion that shots are just goals that go in at a somewhat standard amount of time, then shots become the thing that we’re really after. Shot differential is goal differential except the mostly-random effect of shooting percentage isn’t factored in.

We can assess a player or team’s luck in shooting with a simple stat called PDO. PDO is just Shooting percentage + Save percentage. If you’re number is over 1000, it means that your shooting and/or saving luck has been better than your opponents – and no rabbit foot in the world can help you sustain a PDO over 1000 consistently.

The concept of shots and shot differential is expressed in many different analytical stats. There’s corsi, named after Sabres goalie coach Jim Corsi, which includes simply every shot taken on the net, and then there’s fenwick, named after Matt Fenwick, which excludes blocked shots from the calculation, since a) blocked shots are somewhat of a skill, and it’s possible there might be something to a team blocking a lot of shots or having a lot of shots blocked that is indicative of something else and b) blocked shots are counted rather differently across NHL rinks. These stats are usually expressed as percentages. A player whose team shoots three-quarters of its shots when he’s on the ice would have a corsi for, or CF% of 75%.

Instead of just regular corsi though, analysts often like to look at shot differential only in tied or close situations, as shooting percentage drops when teams are chasing a significant lead, which hampers our ability to treat corsi as a proxy for goals. Corsi is also used frequently as an estimation of possession, as teams with the puck shoot the puck, and teams without the puck are often getting shot on.

All sorts of things out of a player’s control can affect a player’s corsi numbers, though. Things like Zone starts (the amount of faceoffs a player gets in the offensive zone relative to the defensive zone), the quality of your teammates and the quality of competition you face all have effects on not just a player’s advanced numbers, but his normal ones too.

Understanding that shots are a proxy for goals without all the wildness of shooting luck is a basic and somewhat incomplete way of looking at hockey statistics, but it’s a fundamental part of looking at hockey from an analytical perspective. The Los Angeles Kings snuck into the eighth seed of the Western Conference in 2011-12 with the worst shooting percentage in the league, but their true talent was shown to be much greater as they went on to win the Stanley Cup. The same year, the Minnesota Wild started hot with a 20-7-3 record, but their success was largely driven by shot luck, and as such they regressed heavily down to earth.

Plus-minus, or goal differential, is the goal of any player or team. Scoring more goals than the opponent is the only way to consistently win hockey games. But goals can be misleading as a stat. By using shots as a barometer for success, we get a larger sample size to work with and less of the variability that comes with shooting and save percentages.