Ladies and gentlemen, your Corsi Rel leader this year. As you can see, he wants that puck, he needs that puck.
The League of Extraordinary Statisticians (LOES) is a weekly forum bringing together top analytical minds in the hockey world to answer a variety of questions that straddle the line between stats analysis and something you might hear floating around section 304. They have agreed to answer these questions in a few paragraphs or less, and with minimal formulae. Because this is a forum, we'd encourage you to use the comments section to answer the questions yourselves, or to discuss or debate the answers given.
The LOES is not meant to represent the entire of the hockey stats community. There are a number of people that either were too busy or too difficult to contact for the purposes of the forum.
Let's say you just heard of an NHL player for the first time and wanted to know how good he was. What statistics would you look at first? Up to the last 10 years or so, an overwhelming majority of us would say "Goals", "Points", maybe "Plus Minus." My grandpa would say "Fights." In the case of goaltenders, you might have said "Wins", "Goals Against Average", and "Shutouts". In more recent years, a wealth of newer metrics have become available to bring complexity to this issue, including Corsi (which actually had been around for quite awhile), GVT, Quality of Competition & Teammates, Zone Starts, Player Contributions, and so on. This week, the LOES lets you know which statistics they look to if they want to make a "quick-and-dirty" assessment of an NHL player...because believe it or not, every once in a while they don't mind just looking at a couple of numbers, too.
The question: When you initially want to get a sense of the talent of a player, what metric (or metrics) do you look at?
I've got a spreadsheet where I summarize 4-5 years of data for each player, and of them the primary statistics I use to understand a player's nature at a glance is GVT, PC and even-strength scoring rate.
Depending what I see, there are about a dozen other stats I look at next, but "initially" it's just those three. I know people that would add QualComp and/or Corsi/Fenwick to that list, and perhaps my routine will evolve in that direction some day as well.
The only key thing I want to reiterate is that I look at 4-5 years. People never look at enough data when trying to get a sense of a player's talent.
- Rob Vollman, Hockey Prospectus
The "metric" I first use to try to judge a player is watching him play. A good player is involved in a positive way in the play when he is on the ice. More traditional scouting gets an impression of a player faster than you can collect enough games to attempt any sabermetric analysis.
- Greg Ballentine, The Puck Stops Here at Kukla's Korner
If there's a player I know very little about, the first thing I look at is TOI, to see how many minutes he gets at even-strength and on special teams. Then I key in on even-strength statistics - contextual ones like QualComp (who he's playing against) QualTeam (who he's playing with) and Zone Start (where he's starting his shifts). Once I have an idea of how difficult his minutes are, I look at Corsi, Relative Corsi and PTS/60, to answer which direction the puck is going when he's on the ice, and whether he's contributing offence at even strength. Beyond that, I also check power play offence (PTS/60 and GFON/60) if he's playing while a man up and compare it to his teammates, although I take these results with a grain of salt and try to get several years of them. For penalty-killing, I don't go beyond time-on-ice, as I tend to believe penalty-killing is too team-dependent to measure effectively.
If you gave me one stat, I'd take average TOI per game. If you gave me two, I'd take a player's Qual Comp rank on his team and his offensive zone start percentage. If you gave me three, I'd take the last two plus Corsi or Fenwick. You're 90% of the way there with those stats - to fill in the rest, I'd also look at how good or bad a player's team was, and whether he himself had any special offensive skills.
I think we'd all be kidding ourselves if we didn't initially look at goals, assists, and points first. This is partly because of the availability and familiarity of those stats, but also because everyone knows what are good numbers there.
But if I want to see how good an NHL player is, I go to Corsi Rel and Zone Start first to see how he's controlling the play in the situations he's put in. Beyond that, Corsi Rel QoC is next.
I'm much more interested in how a player is used and whether they are pushing the play forward or not than how many goals and points he scores.
- Geoff Detweiler, Broad Street Hockey
These aren't your grandpa's statistics...hell, they aren't even your father's. But the use of these newer metrics aren't just an attitudinal or generational shift; they are the culmination of years of research, data compilation, and philosophizing, done by people genuinely interested in the workings of the game. Truth be told, this process is far from over, but we've also come a long ways in the matter finding out what counts and how to count it. Expect things to get even better as the LOES gets a few more years of data to work with.
Which metrics do you like to use for quick analysis? Are there other statistics you wish were tracked and recorded? Do you prefer easily countable stats or stats that involve extra calculation (the trade-off being a more precise indicator of talent)? Could one of you get the NHL to put sensors in the pucks and players' equipment, please?