The Future is Already Here: Advanced Stats and BoltProspects

The revolution in hockey analysis, late-developing and centered around the increased reliance on statistical information, has been underway for some time now. A small and marginalized movement in the past for various reasons, advanced analysis has gained such widespread traction that its relevance and influence is no longer deniable. That isn't to say that advanced stats are close to or ever will be the panacea for evaluating hockey players and teams. As Robert Vollman, one of the leading hockey statisticians out there, explained in the foreword to his recently-released 2011-2012 Player Usage Charts:

"...objective hockey analysis acts a useful supplement to everybody’s own experience-based understanding of the game..."

Part of a larger sports trend, this shift, predictably, has faced some well-reasoned skepticism (for example, Daniel Wagner's Kierkegaard, Choice and the Limitations of Advanced Statistics) and, from traditionalists, prolonged resistance and outright disavowals (here's looking at you, Brian Burke and Mike Milbrury). At present, there's a very charged discourse surrounding the merits of advanced statistics but, no matter one's take, clearly a new era in professional hockey has dawned when the powers that be are attending conferences on sports analysis, teams (including the Lightning) are adding analysts to their operational staff and mainstream sources are catching on. For these reasons alone, it's worth keeping up with the times.

What's particularly exciting about these developments is that the field is still in its infancy. It may seem odd to suggest that, but for all parties--teams, players, agents, the media and especially the fans--there are implications and possibilities. All stand to benefit from exploring and finding uses information that's already out there or in development. Again, we've only seen the tip of the iceberg; take, for example, the profound potential that the SportVu technology, initially introduced a half-decade ago, already established in soccer and football coverage and now being deployed for basketball), presents for hockey analysis. How teams conduct their business--scouting, signing free agents, conducting trades, preparing for opponents, evaluating their own personnel--is evolving; advanced stats, even if flawed, are here to stay and only to what extent they'll be useful remains to be seen.

I've long been interested in seeing something on par with baseball sabermetrics developed for hockey, but, admittedly, when first exploring the work being done several years ago (particularly Gabriel Desjardin's BehindtheNet.ca), I found myself discouraged and dismissive, perhaps partially due to my non-mathematical background but largely because I was bothered by how problematic it seemed to me to develop reliable, individualized advanced stats for a team sport. It wasn't until last summer that I began to revisit many of the concepts and, accepting their limitations and understanding their intended usage, really started to embrace the advanced stats available.

I was in the midst of initially building my own Lightning site, Electric Blue Hockey Test, and, though I found plenty to read on other teams, there seemed to be a lack of advanced analysis specifically devoted to the Lightning. I spent this past season beefing up my own comprehension with the intent of helping fill the perceived void once I fully committed to writing. Fortunately, Clare Austin of Raw Charge got the ball rolling on this ambition with an intriguing breakdown of Guy Boucher's player usage and evaluation of their resulting performance for last season.

It's in this vein that some of my contributions to this site will be made, as Chad explained in my introduction and it's my hope that those in the community here will find such work stimulative, regardless of whether one's prevailing views are challenged or reinforced. I expect, given the nature of focus of this site, exploring the concept of league equivalencies as a predictive tool for NHL production will, in particular, be an enjoyable addition to the existing dialogues concerning those in the Lightning's prospect pool.

If you aren't already familiar with advanced stats, here's a brief selection of informative readings to guide you:

Following are a few sources of publicly-available advanced metrics:

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

I don't want this idea to die

I don't want this idea to die, so I'll comment. Although the charts aren't simplistic to analyze, the are simplistic charts and standing alone give an incomplete picture of the player. It does at a quick glance give an analyst many ideas of things to look at.

First, if there was one improvement to these charts to make them easier to grasp at first glance, it would be this for me. There is no explanation of the size that a players name is typed at, yet there appears to be at least two different size fonts that the names are typed in. If there is no apparent reason for that, then make one - ice time. It could be as simplistic as total time on ice (or for the purposes of this chart, even strength TOI). Say the larger type size for 20+ min. a med size for 10 - 20 min, and a small type for less than 10. Then I can look at that great big blue circle for Labrie and quickly say to my self, little ice time, small sample size. But if Labrie were approaching UFA and had an individual chart showing multiple seasons of big blue circles albeit limited ice time - he might be a guy I'd look to sign and give an opportunity. The converse of this argument would be Hedman. He would be listed in the largest size type and that smaller blue circle would seem more impressive considering 60 - 65% of his non-neutral face-offs are in the defensive zone and it's an accomplishment to have a plus shot differential in that situation.

By itself, this chart seems to be reliable in that you can clearly spot the top two defensive pairings (Brewer/Hedman are in proximity, Kubina/Clark are right on top of each other) and forward lines ae reasonably close together.

Question: There's a giant blue bubble for Kessel who obviously doesn't play for us. He has a smaller yet still impressive blue bubble on Toronto's chart. Is this a total mistake or does this belong to another Bolt player (Brown is not represented)? Or, cou;d it be Gilroy's? He has no bubble indicating an even shot differential, sits right on the line for having a neutral QOC rating, and is almost 50% in offensive/defensive zone faceoffs. Could happen, but all three being statistically neutral seems like it would be rare.

Hedman seemingly not putting up expected offensive numbers at this point would seem to be justified by the percent of defensive vs offensive draws and the fact that he doesn't get a lot of 1st pp time with marty on the point.

According to this, Aulie was put in higher leveraged situations than say Lee or Mikkelson. Looking where he was position between the Brewer/Hedman and Kubina/Clark suggests he was being put on ice in #2/#3 dman situations versus 2nd/3rd pair where he likely would have been a bit less noticeably exposed.

Connolly's small shot deficit and middle of the chart positioning is encouraging to me.

Mike, I usually don't have a lot of time to devote to this, but keep it coming.

Response to Ken

Ken,

First off, thanks!

Yes, the player usage charts give us an additional means of evaluation. I share the same hesitation about them being a stand-alone tool.

The ice-time = size of font idea is really intriguing. Thanks for the idea; I can rework the chart to show that and will do so when I get a chance. One alternative would be to create a chart with the same X- and Y-axis data with the bubbles, which show a player's relative Corsi rating, changed to show their TOI. You'd get more precise scaling but, of course, would now be comparing multiple charts.

Regarding your observations on Labrie, Hedman, Aulie and Connolly. Exactly! Spot on.

The "Kessel" bubble is a typo--it's Brandon Segal.

Brown wasn't initially included because he played less than 10 games. I included him in the follow-up post I did showing the complete usage charts.

Thanks again for reading,

Mike

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.