Advanced statistics seem like a chore. You may have tried to understand them already, and waded online into a pool of one of the absolutely incomprehensible spreadsheets that can be found at Behind The Net or Hockey Analysis.
There are a growing number of bloggers and fans who pay close attention to these advanced hockey statistics, that go beyond basic “goals” “assists” and “+/-” along with more traditional ways of evaluating player performance. Some of the best in the business are Gabriel Desjardins, JLikens, Kent Wilson and Jonathan Willis, all of whom wade through these waters and pop out with readable analysis on certain players.
Why, for instance, will a player score 25 goals one year, but dip to 19 goals the next? Why do unlikely players top the end-of-year +/- list seemingly every season? How do we determine who the best defensive players in hockey are? How can I win my hockey pool?
We approach these questions the same way that legendary baseball statistician Bill James approached similar questions. You count things.
The basic advanced statistic has become the ‘Corsi’ number, named after a Buffalo goaltender coach Jim Corsi who developed the metric as a way of evaluating defensive performance around his goaltenders. The Corsi is an expanded form of +/-: it records not only every goal, but every shot, miss and block that took place at both nets when a player was on the ice.
From Arctic Ice Hockey
“This metric was presumably adopted by the Sabres because it’s a better indicator of a team’s play than goals for and against, which are highly-driven by factors outside of a team’s control. Shot volume is much more a function of a team’s ability, and a much better predictor of future performance than goal-scoring metrics – in other words, there is basically no such thing as a team that shoots efficiently, just teams that get a lot of shots on goal…or not.”
Why the Corsi number matters is that the team that has led the league in Corsi has never done worse than making the Stanley Cup Finals–the ’08 and ’09 Red Wings, the ’10 Blackhawks and the ’11 Canucks. It’s a number accepted as a “possession” statistic, as in, you can tell where the puck was when the player was on the ice. You can check it out here.
We know that Mason Raymond isn’t the best player in the NHL, and the Corsi number doesn’t attempt to prove that. The Corsi number is a possession-driven statistic that generally shows where the puck was when a player was on the ice, as in, you may want to compare it to other players on his team. You can check out a player’s “relative Corsi” number There are some differences. Again, in no universe is Mikhail Grabovski the best forward in the NHL or Dustin Byfuglien its best defenseman. While the Corsi number favours average players on good teams, the relative Corsi number does the reverse: favouring good players on average teams.
To account for this, the best way to use a possession number is to see how a player did compared to the rest of his team, or even his linemates. The other thing to do is check and see whether his shifts were “protected”. Protected shifts indicate both where a player started his shifts or the quality of players he faced. Again, these are numbers better stacked up against players of the same team, but for every player, his “zone starts” can be found here.
One thing that you might notice is that the Vancouver Canucks have quite a significant gap between their top and bottom players for starting in the zone. Their head coach Alain Vigneault experimented with, and to great success, last year, avoiding matchups against personnel and instead matching up the player vis-a-vis a certain situation. Third-line centre Manny Malhotra started, and won faceoffs of, shifts in the defensive zone, and Henrik Sedin, his first-line centre, began an inordinate number of shifts in the offensive zone.
The last thing that a hockey analyst should look at, and the most important for winning your hockey pools, especially if there are midseason trades, is the PDO. PDO does not stand for anything, but was named for an anonymous commenter on an Edmonton Oilers blog who came up with the idea. The PDO tries to mirror “Batting average on balls in play” which is accepted as baseball’s “luck” statistic. It does this by adding a player’s on-ice shooting percentage to his on-ice save percentage. Since every shot has to result in a save or a goal, the basic PDO number is 1. Anything higher, and the math suggests it will come back down. Anything lower, and the math suggests it will come back up. Here is every player’s PDO from last season of players who played more than 50 games.
Every PDO is pretty close to 1, because of the large number of minutes these players all played. The luck evens out, whether it evens out over the course of a week, a month, a season or even a career, the luck does even out. Shooting percentages are tricky to deal with, since some players are way better at shooting than others and thus more of their shots will go in. A basic rule is that if a player seems to be scoring a lot more goals than normal, check and see if his shooting percentage deviates too much from his overall career number. Corey Perry, the NHL’s leading goal scorer last season, shot 17.2% while his career number is just 12.6%. Expect a slight regression from his goal totals this upcoming season.
These are just a few things that modern hockey analysts are looking for to further understand the game. There is a more in-depth series of FAQs on Arctic Ice Hockey that allow you to understand the true application of these statistics, why they exist, and why we use them. In the spectrum of player movement over the last few offseasons, it becomes clear which NHL teams are using these statistics or a form of them to make player moves and, in essence, stack their odds by using a spreadsheet in an attempt to determine what happens on the ice.