Blog Entry

Chris Snow talks statistical analysis in hockey

Posted on: July 15, 2011 9:16 pm
Edited on: July 15, 2011 9:46 pm
By: Adam Gretz

In any of the other three major North American sports we have a strong idea as to which teams utilize some form of advanced statistical analysis in their front office operations. In baseball we know it's the Boston Red Sox, Oakland A's and Cleveland Indians. In basketball we know it's the Dallas Mavericks. In the NFL we know it's the New England Patriots.

In hockey, it's still a bit of mystery. Perhaps because it's a subject that hasn't really been reported on enough (or at all), or simply because it's an area that teams haven't fully started to develop. It's not necessarily a matter of reinventing the wheel, as much as it is finding a way to add to the process of building a team and constructing a roster.

One person that has had a direct involvement in the still developing analytical community within the NHL is Chris Snow, a former sports writer that covered the Boston Red Sox for the Boston Globe. After spending a number of years in the Minnesota Wild front office as Director of Hockey Operations, Snow caught on with the Calgary Flames this summer as their Director of Video and Statistical Analysis. According to the team's official website, his job description includes the "implementation and oversight of the club’s video and statistical data mining programs including designing, developing and implementing a proprietary data base of hockey information for use by the club. In addition, he will integrate the data-based video system PUCKS into the team’s coaching, player preparation, scouting and planning processes."

I had a chance to speak with him on Friday afternoon to figure out what all of that means and how it compares to his previous role with the Wild.

"I would say it's much more defined," Snow told me. "With the Wild I went there with the expectation that I would be doing some work of this nature. Part of the appeal for [former Wild general manager] Doug Risebrough when he hired me was that I had been an observer to what the Boston Red Sox were doing, and what some pretty progressive baseball teams were doing. At the same time, he wanted someone who could assist the assistant general manager in the day-to-day operations."

Snow said this could have involved any number of responsibilities, including preparations for contract negotiations, arbitration hearings, planning for summer camps and simply being in a position to be around players and coaches on a day-to-day basis.

And anything that involved any sort of analysis.

"Without a doubt when we did things that were analytical in nature, and developed it for that matter, I was the person in charge of doing that," said Snow. "I would say it amounted to maybe 20-25 percent of my time, where now it will be the predominate part of the job."

As for the video analysis portion, Snow said the Flames already have a full-time employee -- Jamie Pringle -- committed to the video needs of their coaching staff and players, and his job will be to make sure they have all of the resources they need, and to complement the process.

"if I think the program they're using can be used to a greater capacity," said Snow. "I can maybe make recommendations to them. If there is, let's say a development component -- maybe they want the system to do more for them -- I might be somebody that kind of takes a week or two and works on that and then gives them a recommendation. I think I'll be a resource for them as opposed to someone on a day-to-day basis that is participating in their process. We haven't been yet been able to get into that day-to-day feel quite yet, but once the season gets close and we go through training camp it will become apparent as to how that will all work together."

I had an opportunity to have a similar discussion with Snow back in August of 2009 when he was still working with the Wild. During that interview I asked him if he knew of any other teams across the league that were involved in the type of analysis he was helping to develop. He couldn't give a definite answer because, as stated above, it hasn't been a subject that's been widely reported on (or accepted) across the league.

Has there been any sort of definitive progress in the two years since?

"I still get asked that question a lot," he said. "And I still don't know the answer."

"I think for teams that show curiosity or interest, and a lot of that honestly came in the past year as I was looking for a team that might be a fit to work for, certain teams demonstrated more curiosity about data and video information than others. But I would say even those teams that showed interest, I think they were probing more than I was able to probe, if that makes sense. They were looking for either what I might be able to offer, or what they should be doing without necessarily entertaining having me work for them. It's difficult unless you work for the team to know how a certain team is operating."

As to why it's so difficult to pinpoint which hockey teams are utilizing these new systems or resources, Snow pointed to two primary reasons.

For one, there hasn't been a team that's had a great deal of success building a winner with this method in the NHL (at least not that we know of). In baseball, it was easy for other small market teams to look at the Oakland A's winning the American League West on a shoestring budget every year and attempting to follow their path.

This hasn't yet happened in the NHL.

In the past, I've asked a number of general managers what, if any, statistical analysis they use and received a variety of answers, ranging from "we use everything," to others saying they're not quite sure what they're supposed to be looking at.

"Anyone needs to see some sort of evidence," Snow said. "For those that are experimenting and telling you they're not quite sure yet, they're being pretty honest with you. I suspect with each of these sports there was an event, a level of success with an organization that was identified, and that compelled an owner or other general managers to say hey, we need to follow that route. In baseball it probably started with with A's, and the Cleveland Indians, and San Diego Padres and Boston Red Sox."

The other reason: there seems to be a bit more fluidity to baseball front offices. People moving from one organization to another, producing a sort of general manager tree, which allows the ideas and practices to be spread throughout the league. Snow pointed to the John Hart Cleveland Indians as one example, having produced people like Chris Antonetti and Josh Byrnes. The Billy Beane A's, of course, would be another good one, having had Paul Depodesta and J.P. Ricciardi move on to take general manager jobs of their own.

"I don't see those types of trees that sprout in the NHL yet," said Snow. "And again, there also hasn't been one person or team that's had a great deal of success with analytics and required the league to pay attention to it."

At this point, we know of at least one team that's currently moving -- at least in some small part -- in this direction in the NHL and it should be interesting to see if the Flames can be the team others in the NHL can point to and eventually follow.

For more hockey news, rumors and analysis, follow @cbssportsnhl and @agretz on Twitter.

Since: Dec 2, 2011
Posted on: December 16, 2011 5:35 am
This comment has been removed.

Post Deleted by Administrator

Since: Dec 12, 2006
Posted on: July 18, 2011 5:14 pm

Chris Snow talks statistical analysis in hockey

The key thing here is not just advanced stats. It is stats tied to video. Snow is looking to use PUCKS to have instant call up of video for all recorded stats for all games. How the stats are used is up to the leadership of the team. Stats in any sport are history only and some form of indication of the future. The stability of the game through rules has influence on the predictability of the stats. Same holds true for the stock market. No one knows how the future will play out but having better information about the past may help win some games or score some talent.

Since: Jul 6, 2011
Posted on: July 18, 2011 11:25 am

Chris Snow talks statistical analysis in hockey

It seems to me that those who argue that advanced statistics would be of no value to hockey because it is such a fluid game are missing one of the big benefits of such statistics,  Looking at the Oakland A's, who are often cited as a great example of a team that made use of this kind of analysis, if you read "Moneyball" (great read by the way), they use the statistics not so much to inform their on-field strategies and decisions, but to inform their personnel decisions - who to draft, who to trade or trade for, who to sign as a free agent.  Hockey does seem way behind in this regard and, while the "gut feel" of scouts can never be completely replaced by looking at complex statistics, these statistics certainly have a place in complementing what they see.

Since: Jul 10, 2010
Posted on: July 17, 2011 12:12 am

Chris Snow talks statistical analysis in hockey

The league needs to strengthen it's statistical analysis capabilities so that it can provide more quantitative justification for disciplinary action.

A start would be to make publicly available a player's PIM per TOI. Two stats that are already available could help to quantify which players really are the most penalized. There is a ton more that could be done to expand on this very basic start.

Any thoughts? 

Since: Aug 5, 2010
Posted on: July 16, 2011 8:01 pm

Chris Snow talks statistical analysis in hockey

Knock the picture all you want bozos, but it's a legit shot from the 2011 Heritage Classic. No Photoshop done to it. The flames you see were a pyrotechnic special effect used when Calgary scored.

That being said, I think many of you are failing to recognize a key piece of information about advanced statistical analysis. Just to be clear, it is much more involved than sitting in a film room and analyzing video/game stats.

Much of what is being talked about in this article has to do with modern computer software programs that analyze every kind of hockey stat imaginable on a magnitude unseen before in the NHL.  Such programs crunch data and produce a unique set of results, which in turn helps front offices analyze individual and team performance in a brand new way.

This type of data analysis is the way of the future. Being able to input nearly every aspect of the game into a software program that will yield easy-to-understand results will only help teams -- and individuals -- become better.

Think of it this way. Major Corporation's all across the globe use some form of advance data analysis -- The results of which are vital to the modern day success of a given company.  There is no denying this. And in sports, I believe we can expect it to work the same way. The more you know, the more educated your decisions can be ... even in a sport as unpredictable as hockey.

Just my two cents. Except for the photo thing mentioned in the first graf ... that's a fact.

Since: Dec 9, 2006
Posted on: July 16, 2011 12:27 pm

Chris Snow talks statistical analysis in hockey

fats and ark-love the comments. i couldn't agree more. as a former player (not in the nhl) you are forced to make split second decisions because if not, you are getting rail-roaded into the boards. you don't have time to analyze data and then think of the best percentage play. sometimes its move the puck and don't get killed.
i was hoping to read this article and SEE some of the data that this gentleman has been looking at, but unfortunately (maybe for secretive reasons?) there was none listed.
i do think you can analyze pp/pk units and positioning when compared to umbrella style pp's/overloads etc. I am not sure how, but i am sure there are different metrics you can look at on each of those units and how successful they may be. For example, maybe you play a diamond pk or a high low pk. When paired up against a rhanded shooter from the point, maybe you can see how many blocked shots compared to attempts/chances a pker has....can be confusing and would take judgement calls but may be a possible stat in the future.
i agree that +/- is somewhat of an antiquated stat and really doesn't tell the whole story of a player or how effective they are. It needs to be paired with another stat (maybe if they have a high +, how many of those scoring chances were they involved in, and were they able to record a point when their team scored, or are they the benefactor of playing on a high scoring line). again not sure these would be easy to find or figure out, but i am sure before WHIP  was popular with pitchers in the majors, they relied on ERA until someone figured out that ERA does not tell the whole story of a pitchers effectiveness.
The nhl is a different animal as you guys have said because of its constant movement and fluidity, there are no times to set plays up on the fly. that comes with playing on a line with someone for years and developing that chemistry...which is hard to pin a statistic on.

Since: Feb 10, 2009
Posted on: July 16, 2011 9:25 am

Chris Snow talks statistical analysis in hockey

whats up with that picture, it makes no sense and looks like an 8 year olds work on photoshop

Since: Jul 16, 2011
Posted on: July 16, 2011 2:45 am
This comment has been removed.

Post Deleted by Administrator

Since: Mar 26, 2011
Posted on: July 16, 2011 1:51 am

Chris Snow talks statistical analysis in hockey

5 stars fatslags. I still think there's some room for situational/statistical analysis, but it's such a fluid game; any meaningful analysis would require a cadre of individuals. I'm not so sure there are going to be any major new revelations anyway. Gee, stay out of the penalty box, take advantage of the power play, clog the middle, but don't block the goalie's view; and don't give up odd-man rushes. Where I do see benefit is to whether or not teams should try forechecking more to get a turnover at the risk of giving up the middle of the ice, etc.

Since: Apr 23, 2010
Posted on: July 16, 2011 1:14 am

Chris Snow talks statistical analysis in hockey

Sorry, forgot to expand upon my fourth point. Hockey is far less reliant on pre-planned tactics than other sports. While players are often prepared for every situation, a pre-planned approach often does not work, it has to morph as a play unfolds (response positioning). Just think of the many scrambles along the boards and around the net. Most offensive maneuvers done behind the opponent's net is a spontaneous decision responding to opponents movements and scrambling to keep control.

The views expressed in this blog are solely those of the author and do not reflect the views of CBS Sports or