NHL Analytic Teams’ State of the Union

Pure-mathematics-formulæ-blackboard

Fandom means a lot of different things to different people. But one thing unites us all: we hope our favorite team will win, and spend a great deal of time thinking how they can.

For those of us who dig a little deeper on the “how” side and use analytics, we hope that our work will eventually make its way to a front office. In some ways, it already has: numerous “hockey bloggers” hirings have been made recently.

But how many and for which teams?

With some research, I’ve culled a working document on all analytics hires for NHL teams and how they may be using analytics. The following descriptions comes from a variety of sources including Craig Custance’s Great Analytics Rankings [Paywall], fellow bloggers from across the internet, media reports, word of mouth and anonymous insiders.

It should be noted that just because a team has made an “analytics hiring”, it doesn’t necessarily mean that they value their input or use the analysis provided properly. In fact, hires can be made simply for PR reasons, and some teams may even give analytics tasks as secondary duties to staff members who do not posses any formal background in the subject. Teams may also have hired private firms providing proprietary data, which in reality may not provide any tangible, verifiable value than what is free and readily available online.

Continue reading

Bayes-Adjusted Fenwick Close Numbers – An Introduction

With the season upon us, and multiple stat sites now hosting team and player fancystats, it is pretty tempting for a hockey fan (well, one who’s into fancystats) to try and check how his team is doing in possession in close situations – in other words, in Fenwick Close (alternatively, score adjusted fenwick). The problem with this, of course, is that the sample sizes are currently so small as to make the #s pretty meaningless – some teams have played as few as 3 games, so you can’t make any judgments based upon these numbers on their own.

But, as I mentioned on twitter, we can still try and take these numbers and make something out of them, using our prior knowledge of the NHL to make judgments. For example, I can look at current fenwick close #s and pretty confidently state “Buffalo is going to be a terrible terrible team” at this point, despite the sample size, given our prior knowledge of what the Sabres are. In other words, we can incorporate current fenwick close #s into a Bayesian Analysis.

Continue reading

2014-2015 Season Preview: The Metro Division

Image from Michael Miller via Wikimedia Commons

Last year, in preseason, the Metro Division, was considered by far the strongest division in the East and the likely bet to take both Wild Cards.  The whole division, minus the Pens, promptly started the season by getting hammered, only recovering later in the season to grab one of the two wild cards.

This year again, the top 5 of the division looks strong enough to take two wild cards.  The bottom 3, particularly the bottom 2, are very weak, but the top 5 is strong and near evenly matched such that they could wind up in any order.  But, given the requirement to project the division, these are how I believe the division should finish up, from worst to first:

Continue reading

NHL Team History, Possession, and Winning the Stanley Cup

Photo by “JulieAndSteve”, via Wikimedia Commons

Gabe Desjardins dropped a comment over at my Tumblr awhile ago, asking me if I could put together a graph expanding on a metric I came up with, 2-Period Shot Percentage (or 2pS%). 2pS% is an historical possession metric that takes shots-for and shots-against in just the first two periods of a game and expresses it as a percentage for the team being analyzed. The idea was that I was trying to get a rough possession measure from the period that would avoid score effects, or the tendency for teams with a lead to sit on the lead and thus give up shots late in the game. Having recently completed a database of period-by-period shot data going back to 1952-53, I have been able to test this metric a bit and the results were good for 2pS% as a possession measure. Returning to Gabe’s request, he wanted to know if I could chart the 2pS% data from year-to-year, with one line following the league leader in the metric and the other line following the Stanley Cup winner. I’d been curious about this myself; certainly there are a number of different ways to express the value of the metric, but this particular one could be interesting because it toes the line between what the Old and New Guard feel is important in this kind of analysis.

Well, I was right that it would be interesting:
Continue reading

NHL Career Charting: The Pre-BTN Era and What We Can Still Do With Historical Data

File:BrendanShanahan.jpg

Photo by “IrisKawling”, via Wikimedia Commons

Hockey statistics have always been fairly historically limited; most of the so-called “fancy stats” have only been tracked (and easily track-able league-wide) back through the 2007-08 season. The prior years have a veil of fog over them, though there is fairly decent shot data going all the way back to the 1952-53 season (thanks to the Hockey Summary Project; I’ve been able to bring the data together), good game-by-game individual player data going back to 1987-88 (thanks to Hockey Reference via Dan Diamond & Associates), and gradually-improving TOI data going back to 1997-98 (thanks to NHL.com and Hockey Reference). Unfortunately, this has lead to a relative dearth of research into the years of the “Pre-BTN” Era, so-called because 2007-08 was the first year we received in-depth, league-wide data from Gabe Desjardins’ Behind the Net stats site and Vic Ferrari’s timeonice.com.

Having a background in history, and also having grown up as a fan of the league in this grey statistical era, I have spent the last couple years trying to compile and present statistics from the Pre-BTN Era in ways that can help provide a window into those years (and possibly inform our understanding of the present-day game). I’m somewhat indebted to Iain Fyffe, a guy who’s been doing similar yeoman’s work much longer than myself at Hockey Prospectus, though more recently he’s been sharing his work at his own site, Hockey Historysis.

The fact of the matter is that there is actually an enormous amount of information out there, and more importantly with graph work we can really do some interesting things. First case in-point is what I call “career charting;” essentially, charting a player’s shots in a game relative to their team’s shots in those same games. Using the metric %TSh, or percentage of team shots, this provides an interesting glimpse into player contributions, workload, and development in the Pre-BTN Era. Adding some artistic (and informational flourish), I present to you Pierre Turgeon:

Continue reading